Home

MAROS - phaas.net

1. Plume Analysis Menu ien cmm Pm mem nece mieria Tir oer tuo Wet Dy Fb dg IQ V ei IP pb Mea i mtt TT The M ann Kendall Statistics menu will be displayed Se Data Censebdalian arse ER Der d MOTA Stattelicn Plume Analyode Cee Emi imm prom Dt Dnm nm Spatial Moment Anatyos saai es ti Pee Aca Syp M External fume Irformation feries PU ot Iun emi ose Pm aem e eod an eens fee 7 MAROS Analysin eop ere qim cef oa smart 2 The M ann Kendall Statistics screen is used to view the Mann Kendall Trend Analysis results by well and constituent Statistical analysis results displayed include ae e The Coefficient of Variation COV a statistical measure of how the individual data points vary about the mean value e TheMann Kendall Statistic MK S measures the trend in the data e The Confidence in Trend is the statistical confidence that the constituent concentration is increasing S20 or decreasing S lt 0 e The Concentration Trend for each well Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not Applicable Insufficient Data Further details on this methodology are provided on Page 28 of the User M anual wt pter oen Sys ie ARNS Cie Mann Kendall Statisucs Statistical analysis for the benzene data is displayed beet aes acd ay CAU eres d ueri itis es mier
2. Region 9 Regions C TCEQ Constituent Caso Region9 Region3 TCEQ Custom Goal 4 1 1 1 2 TETRACHLOROETHANE 630206 43E 04 416 04 14E 01 2 DICHLOROBENZENE 95501 37E 01 5 0E 01 BARIUM 7440393 25E 00 2 6E 00 2 0E 00 23E 00 BENZENE 71432 3 9E 04 36E 04 5 0E 03 COPPER 7440508 14E 00 1 5E 00 1 3E 00 ETHYLBENZENE 100414 1 3E 00 1 3E 00 7 0E 01 LEAD 7439921 40E 03 1 5E 02 PERCHLORATE 1 8E 02 9 2E 02 TOLUENE 108883 7 2E 01 7 5E 01 1 0E 00 XYLENES TOTAL 12E 01 1 0E 01 ZINC 7440666 14E 01 34E 01 4 screen Help Choose from the list of generic Preliminary Remediation Goal PRG recommendations Click on the appropriate standard to be used in database comparisons for COC recommendations Enter your own modifications to cleanup goals under custom goals in mg L Note User entered cleanup standards will supersede chosen standards Back Returns the user to the COC Decision Next Takes the user to the COC Recommendation Screen Help Provides information on the screen specific input requirements COC Recommendation accessed from the Risk Level A ssessment screen allows the user to choose COCs based on Toxicity Prevalence and Mobility of samples from the dataset the site in the boxes to the right Toxicity based COCs amp Monitoring and Remediation Optimization System MARIOS Prevalence based COCs Constituents of Concern Decision Below is a summarized list of CO
3. 2p Zane M arr Gata Wesel Graph Fas itawa Caton cd Mas atta MAROS Site Results mat Pe TA Anabi Iosune certa Fri met PP mm rm dd mp em IESESEREERESSES Page 1 of the report is displayed To select Page 2 dick on the arrow next to 1 at the bottom of the screen ee easi mana amain y Irt orn aper Seetrm AAS Plume characteristics The plume characteristics are displayed on this report The source is classified as PD probably decreasing and the plume tail as PD decreasing Version 2 1 A 11 66 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The plume mass is decreasing zeroth moment result D and the plume is moving from away from the source first moment result I The plume is spreading second moment result PI and I To view the trend results for individual wells select the report Plume Analysis Summary Report Tm Select Plume Analysis Summary MAROS Output Reports Graphs Report from the first list of options duisi ae tahini ida under the heading Report Select View Print Report to display the report Click here to dose The Mann Kendall and Linear Regression trend results for each well are consistent except for well MW 2 For this well review the data spatially Select Mann Kendall Graphs from the main menu
4. awe Lu Hel w fite MW mimmi Version 2 1 A 11 30 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 3 M oment Analysis Statistics allows the user to view the Spatial Moment Analysis results by well and constituent The zeroth moment is a mass estimate for each sample event and COC The mass estimated indicate the change in total mass of the plume over time The first moment estimates the center of mass of the plume coordinates Xc and Yc for each sample event and COC The center of mass locations indicate the movement of the center of mass over time The second moment indicates the spread of the contaminant about the center of mass Sxx and Syy or the distance of contamination from the center of mass The Second Moment represents the spread of the plume over time Displayed are the spatial moment Spatial Momeet Raat rats Rasuna analysis results for each sample event utres Aan cud tr rU gece iment n e mache for the constituent benzene Hoe edjiev mimii td Click Next to proceed to the Zeroth Vue M oment Plot screen The next screens will go through each moment analysis result in detail as well as looking at trends in the data over time Click hereto proceed Notes If more than one COC was being used the user would navigate the results for individual constituents by clicking on the tabs at the top of the scree
5. Data Sufficiency Analysis Options statistical test The Target Power is the coca GE Led RUD Lowel Power Limit mpi probability to detect a true change in the EE appe 005 eso F oon concentration level before and after the remediation It equals to 1 minus type II error The default values for the Alpha Use default Level and Target Power are 0 05 and values 0 80 respectively The Detection Limit is used in the risk based power analysis to indicate that a projected concentration is below the detection limit Set to Default Help In this example all default parameters will be used Click the Back button to return to the D ata Sufficiency Analysis M enu screen Selet Analysis 1 from the Data memeres x Sufficiency Analysis Menu screen to perform the individual well power analysis The Individual Wel Cleanup Status screen will appear Individual Well Cleanup Status This module will determing achieved at individual wells The user can choose to mended or original data for analysis The user should U se yearl y the cleanup status will be evaluated The statistical also be calculated averages 1 Select the type of data for cleanup status evaluation r ptigpz First select the type of data for analysis by dicking on one of the options buttons One choice is to use yearly averages and the other is to use original data Using yearly averages can avoid po
6. Note To edit sample events choose the sample event name and change the range Auto Event Allows the user to update sample events automatically The software will assign the actual sample date as the effective date Also each sample event will be assigned to a unique original date This option should only be used if the data only has one date per sampling event Next Takes the user to the Source Tail Zone Selection screen Back Returns the user back to the Site Information screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 21 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Source Tail Zone Selection accessed from the Sample Events Screen allows the user to define the well type for the wells in the database The MAROS software divides the wells for the site into two different zones e g Source zone and Tail zone The Source area include zones with NAPLs contaminated vadose zone soils and areas where aqueous phase releases have been introduced into groundwater The source area is generally the location with the highest groundwater concentrations of constituents of concern The downgradient groundwater plume Tail zone is the area downgradient of the contaminant source zone The Tail only contains contaminants in the dissolved phase and the sorbed phase but contains no sources of contam
7. Observations in the time series must be mutually independent Since a single constituent in a single monitoring well is tested the homogeneity of variance is generally true Procedures testing wells individually Step 1 Arrange measurements x i 1 2 n in time sequential order and determine the sign of the difference between consecutive measurements for all x as sgn xj Xk 1 if Xj Xk gt 0 sgn x Xk 0 if Xj Xk 0 sgn x Xk 1 if Xj Xx lt 0 where sgn x Xk is an indicator function that results in the values 1 0 or 1 according to the sign of x xx and j gt k Step 2 Calculate the Mann Kendall statistic S which is defined as the sum of the number of positive differences minus the number of negative differences or hl n S Y y sens x k l jk Step 3 Consult a Kendall probability table with the unsigned M ann Kendall statistic S and the number of samples n to find the confidence in the trend CT The Kendall probability table can be found in many statistics textbooks e g Hollander M and Wolfe D A 1973 Step 4 Calculate coefficient of variation COV which measures how individual data points vary about the mean value The COV is defined as the standard deviation of the data divided by the mean or Version 2 1 A 7 22 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE COV X A COV near 1 00 indicates
8. To determine the relative importance of potential locations in the monitoring network we define Slope Factor SF for each potential location to measure the information conveyed by each of them The SF of a location is defined as the standardized difference between the logarithmic scales of its measured concentration and its estimated concentration Since the spatial distribution of groundwater quality data tends to follow lognormal distribution using logarithmic scale of the concentrations will make the plume surface more smoothly Using logarithmic transformations of the concentrations for estimating the average plume concentration were seen in some studies Rice et al 1995 M ace et al 1997 To be consistent the SF calculation is thus based on the logarithmic scale of the concentrations The following steps are used to calculate SF 1 For a given node No find its natural neighbors Ni i e the set of nodes that are directly connected to this node by an edge of a Delaunay triangle Figure A 3 2 Delaunay Voronoi diaaram Figure A 3 2 Illustration of Natural Neighbors 2 The estimated logarithmic concentration ECo of node No is computed as the inverse distance weighted average of logarithmic concentrations of its natural neighbors n 1 mn Bue NO 3 where n number of natural neighbors NC measured concentration in logarithmic scale at node N i 2 1 2 n Version 2 1 A 4 2 Air Force Center for October 2004 En
9. 6 52 0 22 0 048 2 9 11 2 9 85 0 74 0 543 0 162 0 916 3 7 25 3 8 08 0 82 0 678 0 607 0 007 4 5 99 4 4 87 1 13 1 268 0 927 3 801 5 7 64 1 12 1 247 1 257 0 000 6 10 79 0 94 0 885 1 050 0 031 7 9 94 1 86 3 455 1 749 0 843 8 4 53 0 34 0 114 0 627 4 823 9 5 30 1 22 1 484 0 411 0 776 10 9 13 0 72 0 519 0 878 0 248 11 8 21 0 14 0 019 0 098 0 734 12 4 60 0 26 0 069 0 036 0 158 13 6 40 0 12 0 014 0 031 0 021 14 10 36 0 52 0 267 0 061 0 403 15 6 91 1 17 1 373 0 606 2 851 16 4 34 0 53 0 278 0 618 0 415 Overall mean 7 33 p 12 261 4 084 16 027 d 0 33 D 1 31 12 00 10 00 4 o 8 00 E c 9 600 S t 400 c 9 o 2 00 0 00 Oo 1 2 83 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Time No Quarter Figure A 7 4 Time series plot of quarterly contaminant concentrations Version 2 1 A 7 40 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The Strategy for D ata Evaluation The use of an appropriate data evaluation strategy in a long term monitoring program will result in reduction or better control of false positive and false negative rates In this study the strategies for assessing data before testing are presented and appropriate statistical methods for testing the data are recommended A summary of these methods and strategies are presented in tables A 7 17 through A 7 20 The general data evaluation procedures are described below This is a general ou
10. AZIZ ET AL 2000 Version 2 1 October 2004 A 4 14 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Lawrence Livermore Study Numerous correlations were conducted as part of this chlorinated solvent plume study The authors concluded that Another important conclusion dehalogenation exert less impact on plume length than source strength and groundwater velocity Thus plumes with weaker source strength and slower groundwater velocities may be better candidates for the application of natural attenuation remedies is that CVOC transformation rates through CHLORINATED SOLVENT TREND DATA USED FOR STEP 4 Lawrence Livermore Study As part of the Lawrence Livermore National Laboratory chlorinated solvent plume study McNab et al 1999 a time series analysis was performed This analysis divided the chlorinated solvent plumes into two groups a group with Strong Reductive D echlorination processes see Table A 4 6 and No or Weak Reductive D echlorination processes see Table A 4 7 TABLE A 4 6 TEMPORAL TRENDS IN PLUME LENGTH FOR CVOC PLUMES FROM THE STRONG REDUCTIVE DECHLORINATION GROUP CHARACTERIZED BY MONITORING DATA FROM THREE OR MORE YEARS SOURCE MCNAB ET AL 1999 p value Plumes D ecreasing In Plumes Increasing In Plumes With No Length Length Significant Trend 96 Sites Number 96 Sites Number 96 Sites Number s
11. Click the Next gt gt button to proceed The Centerline Regression Projected Concentrations screen will appear The Projected Concentration is the concentration projected to the H SCB If this value is less than the previously defined detection limit a check mark will appear in the box besides it In this example since all projected concentrations are below the detection limit intuitively the risk based cleanup status should be Attained If some wel Is need to be excluded Monitoring and Remediation Optimization System MAROS x Risk Based Power Analysis Results from the analysis dick the Select e uu E ONES n isk based power analysis results are given below for each sampling event classified by ample Size is Wells button and finish the selection a EAE x assumed to be normally or lognormally distributed and results under both assumptions are given for comparison in the Wed Selection Form screen In this screen deselect a well by men unchecki ng the checkbox in the Used Eum Sample Cleanup Power Expected Distibuion e eA ampling Event size Achieved of Test Sample Size Assumption in Analysis column The deselected SaeEenS 12 Hei OD Normal Sample Evert 9 12 Attained 1 000 Sample Event 10 12 Attained 1 000 Sample Event 11 12 Attained 1 000 Sample Event 12 12 Attained 1 000 Sample Event 13 12 Attained 1 000 Sample Event 14 12 Attained 1 000 Sample Event 15 12 Attained 1 000
12. Level of Monitoring Effort Indicated by Analysi Limited 2 Spatial Moment Analysis Results MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 2 Coefficient Mann Kendall Confidence Moment Moment Type Consituent of Variation S Statistic in Trend Trend Zeroth Moment Mass BENZENE 0 88 91 100 096 D 1st Moment Distance to Source BENZENE 0 23 83 100 096 2nd Moment Sigma XX BENZENE 0 35 35 95 496 2nd Moment Sigma YY BENZENE 0 42 53 99 6 Note The following assumptions were applied for the calculation of the Zeroth Moment Porosity 0 30 Saturated Thickness Uniform 12 ft Mann Kendall Trend test performed on all sample events for each constituent Increasing I Probably Increasing Pl Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 2 of 2 MAROS Plume Analysis Summary Project User Name Location Service Station State Texas Time Period 10 4 1988 to 12 19 1998 Consolidation Period No Time Consolidation Consolidation Type Median Duplicate Consolidation Average ND Values 1 2 Detection Limit J Flag Values Actual Value Number Number All Source of of Average Median Samples Mann Linear Constituent Well Tail Samples Detects mg L mg L ND Kendall Regression Modeling Empirical BENZENE MW 8 S 15 1 6 7E 04 5 0E 04 No S
13. Local geologic and topographic maps Geologic data Hydraulic data Biological data Geochemical data and Contaminant concentration and distribution data e Determination of additional data requirements including Vertical profiling locations boring locations and monitoring well spacing in three dimensions A sampling and analysis plan SAP and Other data requirements Conceptual mode development prior to use of the MAROS software will allow more accurate site evaluation through quality data input i e identification of source and tail wells etc as well as viewing the MAROS results in light of site specific conditions The conceptual model will also allow the user to gain insight into the type and extent of site data that is needed to fulfill minimum data requirements in order to fully utilize the MAROS software Minimum Data Requirements Compliance Monitoring data evaluation must be based on data from a consistent set of wells over a series of periodic sampling events Statistical validity of the constituent trend analysis requires constraints on the minimum data input To ensure a meaningful comparison of COC Version 2 1 A 2 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE concentrations over time and space the following minimum requirements were imposed on the time series groundwater monitoring data e Number of Wels Evaluation sh
14. Note If more than one COC was being used the user would navigate the results for individual constituents by clicking on the tabs at the top of the screen The results can be printed by selecting the View Report button EI Nt 3 Statistical and Plume Information Summary Weighting allows the user to weight the individual lines of evidence i e Mann Kendall Trend Analysis Linear Regression Analysis Modeling and Empirical results Choices for weighting trend methods are High Medium Low and Not Used If you choose not to weight trend methods leave the default of All Chemicals and Medium weight Version 2 1 A 11 38 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Since no modeling analysis or empirical Miles eee ce e n evidence is bei ng used the weighti ng for Statistical and Piume urformation Summary o these trends should be changed to Not deinen Used Click on the drop down arrow under the Source Weight text box to the right of Modeling Analysis A list of choices will appear Select N ot Used Repeat for the text box to the right Tail Weight and for the two text boxes adjacent to Empirical Evidence v 4 E o u x Tortue am sharp Puer MARS Vien bares Leer us t Statistical and Plume Wforinauonp Summary oreas Peveeitalinn Opites The top two rows for Mann Kendall Trend Analysis an
15. tt zl the tail of the monitoring network z eim ENTER Then select Sample Event 1 from the From dropdown list and Sample Event 15 from the To dropdown list The software will calculate the risk based cleanup status for all sample events To sel ect p u me centerl i ne wel Is at amp Monitoring and Remediation Optimization MAROS B xj least th T ired f th Plume Centerline Regression Results ea ree we S are requir or e The regression coefficients from the plume centerline regression analysis are given below for sampling analysis consult the plume contour fanelomed E Aa agam le dance deem he pe catene S E concentrations using the results here will be calculated in the next step map and use the best judgment to pick wells that are located on or dose to the plume centerline In this example MW 1 MW 4 and MW 12 BENZENE No of Regression Confidence in Sampling Event EffectiveDate Wels Coefficient 19 Coefficient Sample Event 1 10 4 1988 3 0 028784 87 0 0 042515 90 7 can be used for an approximate Negative c came EE analysis Use the gt gt button to add coeffident se ara the three walls into Plume centerline Sone Bet WE 3 sons ui Sample Event 8 10 3991 3 0 038254 88 8 M wells group Use the button to dadete a wel from the Plume centerline wells group Note when the number of
16. 4 where NC logarithmic concentration at vertex Ni N C logarithmic concentration at vertex N 2 N C3 logarithmic concentration at vertex N 3 A1 Area of sub part A A Area of sub part A2 A3 Area of sub part A3 After elimination of unimportant locations those with smallest SF values the estimation of average plume concentration and triangulation area might be affected By judging the values of CR and AR information loss can be evaluated CR and AR values close to 1 indicate that the information about the plume after elimination of locations is well kept CR and AR values closing to 0 represent a large estimation discrepancy and thus indicate greater information loss By setting the acceptable level of information loss we can judge when to stop eliminating locations Those eliminated locations are called redundant locations and the rest of potential locations are non redundant ones and should be kept An interpretation of the above decision process is given in Figure A 3 4 The optimization process is iterative It starts by eliminating the location s with smallest SF value s then followed by a check of information loss If information loss is not significant within the acceptable range repeat the process until significant information loss happens Two kinds of thresholds are defined to judge whether or not to 1 eliminate a location or 2 to terminate the optimization The SF threshold is defined for the first pur
17. Click on View Print Graph to display a plot of data A plot of data for MW 1 will be displayed Select MW 2 from the drop down box in the top right hand corner of the screen then dick on Graph to plot Version 2 1 A 11 67 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The plot for MW 2 shows that one data point Marin Kendall Plot that may be an error The source of this data ae should be reviewed ma OT 3 oses ET If this point was not included in the data set a trend may be identified by the Mann Kendall analysis DIM Lupe men Ve ater s Select Back to return to the MAROS IT Mh ees sc teens Output Reports G raphs screen ANALYSIS Mn Coo estan Ties WT Vom vitem If Satay meea oae ip Pata D wee PC jisme B iP rt PT ah aapa MONI hy metta TREND tare View Hepat 4 Results can be exported to a database format The user can then use the result to plot data in GIS or other software Select Export MAROS Analysis MAROS Output Reports Graphs Results Casa bw fa Be hiis fw Lower v veil M ahPtatuoo uhi B io ctm dap d mt L i The Export MAROS Analysis Results to June dd Sere Face Access File screen is displayed irma nyman a et Martane etd arena ctine Oprivtivation Syslein MAIS Enter the folder name and the name of Export MAROS Analysis Results the file to c
18. HQ Air Force Center for Environmental Excellence Monitoring and Remediation Optimization System MAROS SOFTWARE version 2 1 User s Guide Julia Aziz Mindy Vanderford Ph D and Charles Newell Ph D P E Groundwater Services Inc Houston Texas Meng Ling and Hanadi S Rifai Ph D P E University of Houston Houston Texas James R Gonzales Technology Transfer Division Air Force Center for Environmental Excellence Brooks AFB San Antonio Texas Air Force Center for Environmental Excellence Version 2 1 November 2004 GSI Job No 2236 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE AFCEE Monitoring and Remediation O ptimization System MAROS Software Table of Contents Section Page INTRODUCTION irit i pt n Eh ane dip ead 1 INTENDED USES FOR MAROS essent nanena nnt 1 FUNDAMENTALS OF COMPLIANCE MONITORING rrnnnnnne 3 QUICK START c 3 MAROS SOFTWARE STEP BY STEP csssssssssssssssssssssssssssssssssssseseseesseeenssassssssst 5 MAROS DETAILED SCREEN DESCRIPTIONS sssri 11 DATABASE COMPACTION ocsssssssssssssssssssssssssseessessssssssssssssseesessenseeneseeeeeeeeees 82 APPENDICES A 1 DATA IMPORT FILE FORMATS erronee A l1 A 2 STATISTICAL TREND ANALYSIS METHODS A 2 1 A 3 WELL REDUNDANCY SUFFICIENCY ANALYSIS DELAUNAY METHOD icti tcd dc cael A 31 A 4 QUALITATIVE EVIDENCE EMPIRICAL DATA METHOD AA1 A 5 SPATIAL MOMENT ANALY
19. Linear Regression is a parametric statistical procedure that is typically used for analyzing trends in data over time However with the usual approach of interpreting the log slope of the regression line concentration trends may often be obscured by data scatter arising from non ideal hydrogeologic conditions sampling and analysis conditions etc Even though the scatter may be of such magnitude as to yield a poor goodness of fit typically characterized by a low correlation coefficient e g R lt 1 for the first order relationship confidence intervals can nonetheless be constructed on the estimated first order coefficient i e the slope of the log transformed data Using this type of analysis a higher degree of scatter simply corresponds to a wider confidence interval about the average log slope Assuming the sign i e positive or negative of the estimated log slopeis correct a level of confidence that the slope is not zero can be easily determined Thus despite a poor goodness of fit the overall trend in the data may still be ascertained where low levels of confidence correspond to Stable or No Trend conditions depending on the degree of scatter and higher levels of confidence indicate the stronger likelihood of a trend The coefficient of variation defined as the standard deviation divided by the average is used as a secondary measure of scatter to distinguish between Stable or No Trend conditions for negative slopes
20. Location Service Station State Texas Contaminants of Concern COC s BENZENE MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 2 of MAROS Linear Regression Statistics Well MW 13 Time Period 10 4 1988 to 12 19 1998 Well Type T Consolidation Period No Time Consolidation COC BENZENE Consolidation Type Median Duplicate Consolidation Average ND Values Specified Detection Limit J Flag Values Actual Value Date COV SPECS LE SLOP SEEPS 101 Confidence in 1 Trend _ 100 0 2 B 0 1 Ln Slope c E 5 1 3E 03 E 0 01 S LR Concentration g Trend 0 001 0 0001 Consolidation Data Table Consolidation Number of Number of Well Well Type Date Constituent Result mg L Flag Samples Detects MW 13 T 10 4 1988 BENZENE 3 5E 02 1 1 MW 13 T 11 17 1989 BENZENE 2 6E 02 1 1 MW 13 T 3 1 1990 BENZENE 4 9E 02 1 1 MW 13 T 5 31 1990 BENZENE 5 2E 02 1 1 MW 13 T 9 13 1990 BENZENE 1 5E 02 1 1 MW 13 T 4 3 1991 BENZENE 1 9E 02 1 1 MW 13 T 7 0 1991 BENZENE 2 9E 02 1 1 MW 13 T 10 3 1991 BENZENE 3 5E 02 1 1 MW 13 T 5 2 1992 BENZENE 8 0E 03 1 1 MW 13 T 1 11 1994 BENZENE 1 0E 03 ND 1 0 MW 13 T 5 28 1996 BENZENE 1 0E 03 ND 1 0 MW 13 T 6 27 1997 BENZENE 1 0E 03 ND 1 0 MW 13 T 12 10 1997 BENZENE 5 2E 04 1 1 MW 13 T 6 19 1998 BENZENE 1 0E 03 ND 1 0 MW 13 T 12 19 1998 BENZENE 1 0E 03 ND 1 0 Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No
21. MA 12 s Sample Event 15 2 19 1 99 BENZENE ND 0 001 IMW 14 S Sample Event 2 1 17 S8 ETHYLBENZENE 0 012 0 001 MVR s Sample Event 15 2 19 1 98 ETHYLBENZENE 0 0058 0 001 MA 15 s Sample Event 14 5 19 1 99E BENZENE ND 0 001 M 15 S Sample Event 14 3 9 99 ETHYLBENZENE ND 0 001 IMA 14 S Sample Event 14 5 19 1 98 BENZENE ND 0 001 IMA 14 s Sample Event 14 5 19 1 S9E ETHYLBENZENE 0 005 0 001 lt lt Back Help At this point your data has been imported the wells have been divided into source and tail zones and the constituents of concern have been selected You may now proceed to Trend Analysis to analyze the plume behavior eMac pad RE TE ug ince Napier MANGAL m Continue to Step 3 Returns the user to the Site Detalls Complete Main Menu to proceed to Trend Analysis to analyze the plume behavior The M ain M enu irte serta talia bon dct screen will be displayed 4 Timur rooe amp Create MAROS Archive File There is also an Gane te Beg 35 option to create an archive file of the site details which have been entered Create MAROS Archive File links to a dialog box where a mdb file containing the imported data site details and source and tail well designations can be stored for later importation Import MAROS Archive File under Data Management The mdb file created should be named to distinguish it from MAROS output files and other site related databases Version 2 1 27 Air Force Center for Oct
22. Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events Moments are not calculated for sample events with less than 6 wells Im iu MM m EE ee X T PP n MAROS Version 2 2002 AFCEE 12 1 2003 Page 1 of 1 MAROS First Moment Analysis Project Tutorial User Name Charles Newell Location Service Station State Texas COC BENZENE Change in Location of Center of Mass Over Time Groundwater Flow Direction 20 40 z 11 89 6990 V 60 04 91 o amp 98 99 03 90 gt Source Coordinate 80 100 06 9 06 98 Y 0 e 12 97 12 98 120 Xc ft Effective Date Constituent Xc ft Yc ft Distance from Source ft Number of Wells 10 4 1988 BENZENE 48 53 71 12 11 17 1989 BENZENE 39 51 64 12 3 1 1990 BENZENE 49 64 81 12 5 31 1990 BENZENE 50 52 72 12 9 13 1990 BENZENE 44 64 77 12 4 3 1991 BENZENE 42 58 71 12 7 10 1991 BENZENE 42 64 76 12 10 3 1991 BENZENE 45 65 79 12 5 2 1992 BENZENE 43 77 89 12 1 11 1994 BENZENE 46 86 98 12 5 28 1996 BENZENE 44 81 92 12 6 27 1997 BENZENE 52 100 113 12 12 10 1997 BENZENE 49 108 119 12 6 19 1998 BENZENE 58 101 116 12 12 19 1998 BENZENE 59 106 122 12 Note Increasing I Probably Increasing PI
23. Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events Moments are not calculated for sample events with less than 6 wells re i J Z A M ee PP H Jx P MAROS Version 2 2002 AFCEE 12 1 2003 Page 1 of 1 MAROS Second Moment Analysis Project Tutorial User Name Charles Newell Location Service Station State Texas COC BENZENE Change in Plume Spread Over Time Mann Kendall S Statistic FI UR WU EEES ers 100000 Trend 99 9 10000 Coefficient of Variation 1000 0 38 Second Moment 100 Trend 10 l Date F e vef FORE S Pa S es a Mann Kendall S Statistic Confidence in Trend 94 3 Coefficient of Variation 0 31 Second Moment Trend 1 PI Syy 2 sq ft e e eo Sxx 2 sq ft gt eo E e Data Table Effective Date Constituent Sigma XX sq ft Sigma YY sq ft Number of Wells 10 4 1988 BENZENE 1 104 2 893 12 11 17 1989 BENZENE 1 329 6 063 12 3 1 1990 BENZENE 1 366 3 215 12 5 31 1990 BENZENE 1 533 4 347 12 9 13 1990 BENZENE 1 142 3 694 12 4 3 1991 BENZENE 1 007 3 502 12 7 10 1991 BENZENE 1 008 3 718 12 10 3 1991 BENZENE 1 025 3 824 12 5 2 1992 BENZENE 1 436 6 019 12 1 11 1994 BENZENE 1 308 4 493 12 5 28 1996 BENZENE 1 076 3 962 12 6 27 1997 BENZENE 1 294 4 735 12 MAROS Vers
24. Trend Summary Result allows the user to view print graphical Trend Summary Results in Excel Version 2 1 A 11 64 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Trend results should be reviewed for all wells to check they make sense Version 2 1 A 11 65 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Click here to dose Select Print Chart to print the current summary graph Click on the X at thetop right hand corner of the screen to dose or select Back to Access to return to the Trend Summary Results Select Back to Access to return to the MAROS Output Reports Graphs Screen Note Do not change the name or content of the worksheet xIsLO ET rendResults or move it to other folders The xIsLO ET rendResults worksheet will remain open until the user closes it All the results and graph output are kept if the user chooses to save the file before closing it The user should save the file under a different name by choosing Save as under the Excel menu option File 2 To view the Site Report select Site EE Report from the first list of options MAROS Output Reports Graphs under the heading Report Doisn initi ibo du itr tod BM ia Select View Print Report to display asd Wess nanena 5000s the MAROS Site Results report RATES Wt waaa ear ago
25. 11 2 for example For example if a sentry well provides early warning for a downgradient receptor its sampling frequency may need to be kept quarterly even if all the measurements are nondetects and the recommendation is biennial ini x amp Monitoring and Remediation Optimization System MAROS Sampling Frequency Recommendation Sampling frequency is determined considering both recent and overall trends so Sampling Frequency is the final recommendation Recent result is the frequency determined based on recent short period of sampling Overallt result is the frequency determined based on overall long period of sampling BENZENE The results of each monitoring well for a certain COC are listed below Well Name Sampling Frequency Recentresult Overall result wet Annual Annual Annual ww 12 Annual Annual Annual Mw 13 Biennial Annual Annual Marta Biennial Annual Annual ws Biennial Annual Annual w 2 Biennial Annual Annual Mw 3 Annual Annual Annual wo Annual Annual Annual wes Annual Annual Annual zl SAMPLING OPTIMIZATION lt lt Back View Report Next gt gt Click the Next gt gt button to proceed The Sampling Frequency Analysis Complete screen will appear indicating that the sampling frequency analysis has been completed E Air Force LTMPMonitoring and Remediation Optimization System MARO Sampling Frequency Analy
26. 16x1 61 11 28 k SS crours i l SS error SS tora SS ggoups 15 95 11 28 4 67 Step 4 The F statistic f is SS croues k 1 11 28 4 1 _ 3 76 f SS orror N k 467 06 4 0 39 9 65 Step 5 The critical value F for F distribution with a 0 01 3 numerator degrees of freedom and 12 denominator degrees of freedom is 5 95 Therefore f 2 9 65 F 2 5 95 the assumption of equal variances is rejected Version 2 1 A 7 28 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 14 Example calculation for Levene s test Well Original data Group mean Absolute Residual group Overall residual xij xi residuals z mean zi mean z 13 96 2 75 12 77 1 56 MW 1 9 66 11 21 1 55 2 15 8 46 2 75 8 77 1 79 3 87 3 11 MW 2 5 15 6 97 1 82 2 46 10 11 3 14 8 56 0 07 1 61 8 16 0 47 MW 3 8 97 8 63 0 34 0 27 8 84 0 21 11 05 1 66 10 82 1 43 MW 4 8 68 9 39 0 70 1 55 7 00 2 39 If we are confident that the variability estimated from the above set of data is true the parametric ANOVA test for detecting differences among the group means is not advisable Approaches that account for unequal variances such as data transformation to stabilize variances intra well Shewart CUSUM control charts or intrawell prediction limits should be used instead Conversely the conclusion of unequal variances might not be true if we consider th
27. B and McNichols R J 1994a Ground Water Monitoring Statistics Update Part I Progress Since 1988 Ground Water Monitoring and Remediation Vol 14 No 4 pp 148 158 Davis C B and McNichols R J 1994b Ground Water Monitoring Statistics Update Part II Nonparametric Prediction Limits Ground Water Monitoring and Remediation Vol 14 No 4 pp 159 175 EPA 1989 Statistical Analysis of Ground Water Monitoring Data at RCRA Facilities Interim Final Guidance Washington D C Office of Solid Waste U S Environmental Protection Agency EPA 1991a Monitoring Guidance for the National Estuary Program Interim Final Washington D C Office of Wetlands Oceans and Watersheds U S Environmental Protection Agency EPA 1991b Solid Waste Disposal Facility Criteria Final Rule Federal Register 56 50 978 51 119 EPA 1992a Statistical Analysis of Ground Water Monitoring Data at RCRA Facilities Addendum to Interim Final Guidance Washington D C Office of Solid Waste U S Environmental Protection Agency EPA 1992b Methods for Evaluating the Attainment of Cleanup Standards Volume 2 Ground Water Washington D C Office of Policy Planning and Evaluation U S Environmental Protection Agency EPA 2000 Guidance for Data Quality Assessment Practical Methods for Data Analysis EPA QA G 9 QA00 Update Washington DC Office of Environmental Information U S Environmental Protection Agency July 2000 EPA 600 R 96
28. COC Decision Mobility Below is a list of COC recommendations from the available dataset based on the Mobility of the compounds Repr Conc coc Above PRG PERCHLORATE Above PRG BENZENE Above PRG TOLUENE Above PRG 1 1 1 2 TETRACHLOROETHANE Above PRG 1 2 DICHLOROBENZENE Above PRG LEAD Above PRG BARIUM Above PRG COPPER Above PRG ETHYLBENZENE Below PRG IZINC Below PRG XYLENES TOTAL D Kd s for metals ID Insufficient Data lt lt Back Kd 986 02 35E 01 86E 01 1 88900 1 0E 01 14E 01 4 0E 01 14E 00 1BE 01 128400 Fy Note Top COCs by mobiliy were determined by examining each detected compound in the dataset and comparing their mobillies Koc s for organics assume foc 0 001 an imd COC Decision Mobility shows a list of COC recommendations from the available dataset based on the Mobility of the compounds Top COCs by mobility were determined by examining each detected compound in the dataset and comparing their mobilities Koc s for organics assume foc 0 001 and Kd s for metals Compounds listed first are those above the PRG and are shown on the COC Decision screen Toxicity of the compounds Representative Concentration mg PRG 1 0E 01 4 0E 03 1 1 2 TETRACHLOROETHANE 38E 01 43E 04 BENZENE 14E 01 356 04 PERCHLORATE 126 01 18E 02 2 DICHLOROBENZENE 88E 01 37E 01 TOLUENE 17E 00 72E 01 BARIUM 32E 00 2 36 00 Note Top COCs by toxicity were determined by exa
29. DL Step 5 Consult Cohen s table e g EPA 2000 Table A 10 of Appendix A with h and y to determine the value of the parameter If the exact value of h and y do not appear in the table use double linear interpolation to estimate Ai h Step 6 Estimate the corrected sample mean x and sample variance s2 which account for the data below the DL as Xx X X X DL and s s X x DL An example of this adjustment is available in EPA guidance EPA 2000 pages 4 44 Procedures Nonparametric Prediction Limit Step 1 Let n be the total number of background measurements and denote the number of monitoring wells for future comparison asr Step 2 Determine the resample plan e g One of Two Samples in Bounds plan or One of Three Samples in Bounds plan and usethe largest or next to largest background measurement asthe nonparametric prediction limit Step 3 Usen r and choices from step 2 in the tables from Davis and M cNichols 1994 or Gibbons 1994 to determine the Per Constituent significance level o or Per Constituent confidence level 1 a respectively Step 4 Inverse problems can be solved by fixing the desired Per Constituent significance level a and using the tables to inversely determine the number of background measurements n or the number of wells for future comparison r If N constituents are involved the Per Constituent significance level o should be calculated as Version 2 1 A 7 34 Air F
30. E vw xi E z 4 E e MACIA 102 0 20 0 4 4 E MAIS MOD 258 E E to be redundant For example a sentinel S we a 2 well might be unchecked since it cannot MW 3 350 100 E 1 De E MALA 550 37 0 4 E be eliminated O MVS 40 70 0 4 l xl 6 E H H Z Removable e e E Select All Sets all the sampling locations a lt lt Back Options Preliminary Analysis gt gt Back Returns the user to the Well Redundancy Analysis D elaunay M ethod screen Options Shows screen Wall Redundancy Analysis Options where the optimization parameters can be set Otherwise the default settings or the settings from the previous analysis will be used Preliminary Analysis Calculates the sampling events averaged Slope Factor SF values for all locations for each COC and then proceeds to the A ccess M odule Slope Factor V alues screen Help Provides additional information on software operation and screen specific input requirements Steps for use 1 Browse sampling locations for each COC by clicking the tab on the page frame For example dick Benzene to view sampling locations where Benzene concentrations were measured 2 Remove a location from the potential locations by unselecting the Selected check box Select Removable check box to decide if a location can be eliminated by the optimizing process 3 Set up the properties of potential locations for all COCs and then proceed to Prdiminary Analysis During the proces
31. MAROS individual Well Cleanup Status Results yearly averages or origi nal data that is ae eae ae te dene edm se emm eEum used in the evaluation normally or lognormally distributed and results under both assumptions can be compared Also available is an optional analysis where power analysis based on Student s ttest on mean difference is performed i Besuks shown are based on vest averages YES Cleanup Achieved Indicates whether BENZENE ETHYLBENZENE TOLUENE xyLENES TOTAL the cleanup goal is achieved in the well Sample Cleanup Distribution Assumption Resu Its cou l d be A ttai n ed C on t 3 am pl i n g Mon Name Size Achieved 1 H i aneen ET continue sampling N ot Attained or N C 2 ZE Ses RES not conducted dueto insufficient data Lee MW 3 g Attained u i ws g Aftaned ViewLog To facilitate the power analysis z MAS 6 Attained r E WS 8 Cori Sanna plcna Buwer concentration data are assumed to be 9 vu EEN ludi either normally or lognormally E N C not conducted due to insufficient data distri buted 7 Resu Its for both assu mptions a are calculated and provided for z lt lt Back View Report Visualize Next gt gt Help comparison See A ppendix A 6 for detailed explanations View Normal Views results calculated under the assumption that data are normally distributed View Log Views results calculated under the assumption that data are lognormally
32. MAROS E lni xj Sel ect the page Ww ith a certai n COC name Sampling Frequency Recommendation to display the recommended results for Sampling frequency is determined considering both recent and overall trends so th at C OC LI Sampling Frequency is the final recommendation Recent result is the frequency determined based on recent short period of sampling Overallt result is the frequency determined based on overall long period of sampling Sampling Frequency The final frequency BENZENE ETHVLBENZENE TOLUENE xvLENES TOTAL recommendation determined based on overall and recent trends and other The results of each monitoring well For a certain COC are listed below Well Name Sampling Frequency Recent result Overallresult 4 factors z MW 1 Annual Annual Annual 0 MW 12 Annual Annual Annual A E MAS Annus Annus Annual Recent Result The frequency determined MW 14 Biennial Annual Annual H S HINTS Berni Anus Anus based on the recent period of data E eun ina RAE S M4 Annual Annual Annual a Overal Result The frequency MW 5 Annual Annual Annual z H z determined based on the overall period a of data lt lt Back View Report Next gt gt GO Back Returns the user to the Sampling Frequency Analysis screen where the user can change Rate of Change parameters and perform a new analysis View Report Generates a report with sampling frequency recommenda
33. Step 4 Make decisions Although the confidence limits for both MW 1 and MW 2 are lower than the ACL they represent different conditions MW 1is well within compliance since all of its concentrations are below the ACL Even though all concentrations in MW 2 are above the ACL no statistically significant evidence is available to conclude its non compliance More samples are needed in the future to verify whether this is a true non compliance For MW 3 it is statistically significant that the mean TOC level at this well is out of compliance Version 2 1 A 7 17 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 9 Example TOC concentrations for confidence limits Monitoring well MW 1 MW 2 MW 3 4 19 6 97 6 97 2 91 5 20 8 64 Concentrations mg L 4 26 5 03 8 69 2 93 6 41 6 54 2 32 6 79 10 12 3 34 5 34 8 84 x 3 33 5 96 8 30 5 0 70 0 79 1 21 x14 4 5 n 2 36 4 87 6 64 ACL 5 00 399 lower confidence limit is below the ACL indicating non compliance METHOD 4 THE SEQUENTIAL T TEST IN CORRECTIVE ACTION MONITORING One purpose of groundwater corrective action monitoring is to document the effectiveness of the remedial action More specifically the groundwater contamination should be cleaned up and this attainment should be proved by appropriate statistical tests In M ethods for Evaluating the Attainment of Cleanup Standards Volume 2 Ground Water EPA 1992
34. The Linear Regression Analysis is designed for analyzing a single groundwater constituent multiple constituents are analyzed separately The MAROS software includes this test to assist in the analysis of groundwater plume stability For this evaluation a decision matrix was used to determine the Concentration Trend category for each well as presented on Table A 2 2 LINEAR REGRESSION The objective of linear regression analysis is to find thetrend in the data through the estimation of the log slope as well as placing confidence limits on the log slope of the trend Regression begins with the specification of a model to be fitted A linear relationship is one expressed by a linear equation The Linear Regression analysis in MAROS is performed on Ln COC Concentration versus Time The regression model assumes that for a fixed value of x sample Version 2 1 A 2 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE date the expected value of y log COC concentration is some function For a particular value x or sample date the predicted value for y log COC concentration is given by jy atbx The fit of the predicted values to the observed values xi yi are summarized by the difference between the observed value y and the predicted value the residual value A reasonable fit to the line is found by making the residual values as small as possible The meth
35. The plume centerline wells should be selected in the same way as in the BIOSCREEN and BIOCHLOR applications To select click on the well in the Wals for select listbox and then click the gt gt button To deselect click on the well in the Plume centerline wals listbox and then click the lt lt button At least three wells are needed for the regression analysis The selected wells do not have to be ordered Refer to Appendix A 6 for details Back Returns the user to the D ata Sufficiency Analysis M enu screen Analysis Determines the plume centerline concentration regression coefficients based on the selected plume centerline wells for the sampling events selected by the user The screen Plume Centerline Regression Results will pop up Help Provides additional information on software operation and screen specific input requirements Version 2 1 80 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Plume Centerline Regression Results This screen accessed from the Parameters for Risk Based Power Analysis screen by clicking Analysis is used to display the results for the plume centerline concentration regression grouped by COC x No of Wells The number of plume E3 Monitoring and Remediation Optimization System MAROS Plume Centerline Regression Results centerline wells that are available for The regression coefficien
36. Version 2 1 October 2004 A 4 7 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE B Empirical Data BTEX Plumes Larger Releases PLUME LENGTH DATA USED FOR STEP 3 Data from other releases besides UST sites suggests that longer BTEX plumes are possible One data set derived from a plume data compiled by Wiedemeier et al 1999 shows 18 Air Force plumes with a median BTEX plume length of 530 ft see Table A 4 1 TABLE A 4 1 LENGTH OF BTEX PLUMES FROM LARGER FUEL RELEASES DATA FROM WIEDEMEIER ET AL 1999 B c NS RE LEASES ft Elmendorf AFB AK 3 Dover AFB DE w mas m 1 amp 0 MymBem ROFaWy sc nm BaseCrk mMm 99 KingSimonAFB jak 80 Makes WI 89 PopeAFS FPTA e 72 GimsarB NY 30 MabwAre W f Semmou oonAFB NC 35 LLLI MM QN E iiie AFB os tangle Areva 15 J _ P 1 Maimm 3 0 SUME NE DhPetile po MEDIAN 25Peretie Minimum w Number ofSites 18 PLUME LENGTH CORRELATION EQUATIONS USED FOR STEP 3 A second approach to compare your plume against empirical plume data is using correlation equations One takes site data from your site applies the correlation equation and then obtains a predicted plume length Then one uses the approach outlined in Step 3 to estimate plume behavior Version 2 1 A 4 8 Air Force Center for O
37. gradual release or zi gt SCL immediate release Any such designation must be verified on the next round of sampling before further investigation is deemed necessary Once confirmed the test results indicate that groundwater at the well is contaminated As monitoring continues and no exceedence is found the combined Shewart CUSUM control charts should be updated periodically to incorporate these new data Davis 1994 suggests that every two years all new data that are in control should be pooled with the initial samples to calculate the new background mean and variance Example Step 1 The data from Gibbons 1994 Example 8 1 page 165 are used and listed in TableA 7 7 The mean and standard deviation are esti mated to be 50 ug L and 10 ug L respectively from eight previous background measurements in the same well Setp 2 The three Shewart CU SUM parameters are selected ash 5 c 1 and SCL 4 5 in units of standard deviation Step 3 The standardized value z for each new measurement is computed and presented in the fifth column of Table A 7 7 For example z3 60 50 10 1 Version 2 1 A 7 11 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 4 The quantity S is computed and presented in the seventh column of Table A 7 7 For example Ss 2 max 0 3 1 1 max 0 3 23 Step 5 The control chat is presented in Figure 1 with S plo
38. possible This will be valid and safe if the selected constituents are most likely to be different from their null hypotheses Using Appropriate Statistical M ethods This section details the statistical approaches that can be used in the data evaluation procedures for the control of false positive and false negative rates First scientifically sound statistical methods widely used for assessing the conditions of groundwater contamination and for making decisions about regulatory requirements are outlined Second procedures for dealing with problems that arise from violations of statistical assumptions will be presented In practice the procedures for dealing with violations of statistical assumptions should be performed first In this appendix the statistical methods are presented first so that assumptions of these methods are understood and strategies dealing with violations of these assumptions can be developed METHOD 1 COMBINED SHEWART CUSUM CONTROL CHART The combined Shewart CUSUM control chart ASTM 1996 EPA 1989 Gibbons 1994 is a statistical method for intra well comparisons used in detection monitoring to determine if the groundwater at the well is contaminated The combined Shewart CUSUM control chart method is sensitive to both immediate and gradual releases Also since it is an intra well comparison method problems associated with spatial variations can be avoided Version 2 1 A 7 10 Air Force Center for October 2004 Env
39. standard deviation divided by the average 165 705 Values near 1 00 indicate that the data form a S E E wx si relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of lt lt Back Next gt gt View Report Help scatter about the mean MK S The Mann Kendall Statistic S measures the trend in the data Positive values indicate an increase in constituent concentrations over time whereas negative values indicate a decrease in constituent concentrations over time The strength of the trend is proportional to the magnitude of the M ann K endall Statistic i e large magnitudes indicate a strong trend IMW 14 IMW 13 1606 50 99 9 1406 53 99 8 1 591 68 100 0 IMW 1 IMA 8 IMW 7 MW 6 1701 45 98 5 S AA 000 6 o0ooco i Note Increasing 1 Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Source Tail S T COV Coefficient of Variation MK S Mann Kendall Statistic TREND ANALYSIS Confidence in Trend The Confidence in Trend is the statistical confidence that the constituent concentration is increasing S20 or decreasing S 0 Concentration Trend The Concentration Trend for each well is determined according to the rules outlined in Appendix A 2 Results for the trend include Increasing Probably Increasing No Trend Stable Probably Decreasing D
40. the optimization will be stopped and locations eliminated in this step will be resumed The default value is 0 95 For the setting of these parameters the user is referred to the corresponding parts in chapter M AROS D amp ailed Screen Descriptions Well Sufficiency Analysis Augmentation of a monitoring network is needed when the existing network cannot achieve certain monitoring goals Augmentation in this document means more sampling locations and or more frequent sampling In this section a method for determining new sampling locations is introduced which is intended to enhance the spatial plume characterization This method utilizes the SF values obtained from the previous analysis to assess the concentration estimation error at potential areas inside which new sampling locations can be placed Among these potential areas those with a high estimation error may be designated as regions for new sampling locations Conceptually the method is to overlay a grid onto the study area and interpolate the SF values at existing sampling locations to grid cells that do not contain sampling locations These grid cells serve as potential areas for new sampling locations Those potential areas with a high estimated SF value i e high estimation error aretherefore candidate regions for new sampling locations This approach is further simplified in MAROS in order to adapt to the visualization limitations of Microsoft A ccess and Excel In the simplifi
41. well will be excluded from analysis for all sample events In this example all wells will be used in the analysis Click the Back button to return View Normal View Log N C not conducted due to insufficient data S E sample mean significantly exceeds cleanup goal lt lt Back View Report Next gt gt Help al al al al al ala A Ry ay ay ay at oj ol oj o e e amp Click the Analysis gt gt button on the Centerline Regression Projected Concentrations screen to proceed The Risk Based Power Analysis Results screen will appear In this screen the risk based cleanup status power and expected sample size for each sample event are listed in order of time The cleanup status as a function of time may reflect the progress in remediation e g from Not Attained gt Attained SAMPLING OPTIMIZATION In this example the cleanup status is Attained for all sample events at an HSCB that is 1000 ft downgradient of the network The View Normal and View Lognormal buttons allow the user to view results calculated assuming that the data are normally distributed and lognormally distributed respectively A detailed result report can be generated by clicking the View Report button Click the Next gt gt button to proceed PS and Remediation Optimization System MAROS The Risk Based Power Analysis Complete screen will appear To modify analysis parameters and
42. 084 Gibbons R D 1994 Statistical Methods for Groundwater Monitoring New York John Wiley amp Sons Gilbert R O 1987 Statistical Methods for Environmental Pollution Monitoring New York John Wiley amp Sons Kufs C T 1994 Journal of Environmental Hydrology Vol 2 No 1 pp 3 13 Loftis J C Iyer H K and Baker H J 1999 Rethinking Poisson Based Statistics for Ground Water Quality Monitoring Ground Water Vol 37 No 2 pp 275 281 Neter J Kutner M H Nachtsheim C J Wasserman W 1996 Applied Linear Statistical Models fourth edition WCB McGrow Hill Tuckfield R C 1994 Estimating an Appropriate Sampling Frequency for Monitoring Ground Water Well Contamination Westinghouse Savannah River Company WSRC MS 94 0111 Available from the National Technical Information Service U S Department of Commerce Weber E F 1995 Statistical Methods for Assessing Groundwater Compliance and Cleanup A Regulatory Perspective Groundwater Quality Remediation and Protection International Association of Hydrological Sciences IAHS Publication No 225 GSI 2000 Personal contacts Groundwater Services Inc Houston TX See also Monitoring and Remediation Optimization System MAROS User s Guide Version 2 0 November 2003 Air Force Center for Environmental Excellence Version 2 1 A 7 44 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTE
43. 25 Yes Yes Sample Event 13 12 10 1997 MW 3 1 000E 03 1155 0 4 10E 02 2 746E 24 Yes Yes Sample Event 13 12 10 1997 MW 4 3 000E 03 1135 0 4 10E 02 1 870E 23 Yes Yes Sample Event 13 12 10 1997 MW 5 3 630E 01 1194 0 4 10E 02 2 016E 22 Yes Yes Sample Event 13 12 10 1997 MW 6 1 000E 03 1267 0 4 10E 02 2 786E 26 Yes Yes Sample Event 13 12 10 1997 MW 7 1 000E 03 1277 0 4 10E 02 1 849E 26 Yes Yes Sample Event 13 12 10 1997 MW 8 1 000E 03 1245 0 4 10E 02 6 864E 26 Yes Yes Sample Event 14 6 19 1998 MW 1 1 140E 02 1177 0 2 83E 02 3 732E 17 Yes Yes Sample Event 14 6 19 1998 MW 12 1 000E 03 1090 0 2 83E 02 3 852E 17 Yes Yes Sample Event 14 6 19 1998 MW 13 1 000E 03 1125 0 2 83E 02 1 429E 17 Yes Yes Sample Event 14 6 19 1998 MW 14 1 000E 03 1088 0 2 83E 02 4 077E 17 Yes Yes Sample Event 14 6 19 1998 MW 15 1 000E 03 1000 0 2 83E 02 4 936E 16 Yes Yes Sample Event 14 6 19 1998 MW 2 1 000E 03 1192 0 2 83E 02 2 140E 18 Yes Yes Sample Event 14 6 19 1998 MW 3 2 000E 03 1155 0 2 83E 02 1 221E 17 Yes Yes Sample Event 14 6 19 1998 MW 4 1 400E 02 1135 0 2 83E 02 1 507E 16 Yes Yes Sample Event 14 6 19 1998 MW 5 6 800E 02 1194 0 2 83E 02 1 875E 16 Yes Yes Sample Event 14 6 19 1998 MW 6 1 000E 03 1267 0 2 83E 02 2 555E 19 Yes Yes Sample Event 14 6 19 1998 MW 7 1 000E 03 1277 0 2 83E 02 1 925E 19 Yes Yes Sample Event 14 6 19 1998 MW 8 1 000E 03 1245 0 2 83E 02 4 767E 19 Yes Yes Sample Event 15 12 19 1998 MW 1 1 900E 03 1177 0 7 29E 08 3 568E 07 Yes Yes Sample Ev
44. 4 8 YES 0 896 4 ji Log N C not conducted due to insufficient data S E sample mean significantly exceeds cleanup goal FigureA 6 3 Individual well deanup status results from the optional analysis Yearly averages indicates the type of data used in the evaluation yearly averages or original data without being yearly averaged If there are many years of data using yearly averages is recommended because it can reduce the influence of seasonal variation and serial correlation Distribution Assumption shows the assumption of data distribution for the results currently shown Results for both normal and lognormal assumptions are given Because normality tests for small size sample e g 20 may not be accurate presenting results under both assumptions provides a chance for comparison so that the conservative results may be used Version 2 1 A 6 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Significantly Cleanup Goal Figure A 6 3 indicates whether the mean contaminant concentration at a well is below the cleanup goal with statistical significance using the Student s t test in the optional analysis YES indicates the mean concentration is significantly below the cleanup goal supported by a power equal to or greater than 50 although may not be as high as the expected power Therefore this result is also an indication of well cleanup but secondary to th
45. 6 1 000E 03 1267 0 4 57E 02 7 589E 29 Yes Yes Sample Event 7 7 10 1991 MW 7 1 000E 03 1277 0 4 57E 02 4 807E 29 Yes Yes Sample Event 7 7 10 1991 MW 8 1 000E 03 1245 0 4 57E 02 2 072E 28 Yes Yes Sample Event 8 10 3 1991 MW 1 8 000E 01 1177 0 3 83E 02 2 234E 20 Yes Yes Sample Event 8 10 3 1991 MW 12 2 800E 02 1090 0 3 83E 02 2 181E 20 Yes Yes Sample Event 8 10 3 1991 MW 13 3 500E 02 1125 0 3 83E 02 7 145E 21 Yes Yes Sample Event 8 10 3 1991 MW 14 1 000E 03 1088 0 3 83E 02 8 407E 22 Yes Yes Sample Event 8 10 3 1991 MW 15 1 000E 03 1000 0 3 83E 02 2 436E 20 Yes Yes Sample Event 8 10 3 1991 MW 2 5 000E 03 1192 0 3 83E 02 7 867E 23 Yes Yes Sample Event 8 10 3 1991 MW 3 1 100E 01 1155 0 3 83E 02 7 127E 21 Yes Yes Sample Event 8 10 3 1991 MW 4 5 500E 02 1135 0 3 83E 02 7 659E 21 Yes Yes Sample Event 8 10 3 1991 MW 5 2 700E 00 1194 0 3 83E 02 3 935E 20 Yes Yes Sample Event 8 10 3 1991 MW 6 1 000E 03 1267 0 3 83E 02 8 930E 25 Yes Yes Sample Event 8 10 3 1991 MW 7 1 000E 03 1277 0 3 83E 02 6 091E 25 Yes Yes Sample Event 8 10 3 1991 MW 8 1 000E 03 1245 0 3 83E 02 2 072E 24 Yes Yes Sample Event 9 5 2 1992 MW 1 2 500E 01 1177 0 8 53E 02 2 216E 19 Yes Yes Sample Event 9 5 2 1992 MW 12 1 100E 02 1090 0 3 53E 02 2 106E 19 Yes Yes Sample Event 9 5 2 1992 MW 13 8 000E 03 1125 0 8 53E 02 4 449E 20 Yes Yes Sample Event 9 5 2 1992 MW 14 1 000E 03 1088 0 3 53E 02 2 054E 20 Yes Yes Sample Event 9 5 2 1992 MW 15 1 000E 03 1000 0 3 53E 02 4 596E 19 Yes Yes Samp
46. 6 27 1997 BENZENE 1 9E 02 12 12 10 1997 BENZENE 8 5E 03 12 6 19 1998 BENZENE 7 5E 03 12 12 19 1998 BENZENE 2 4E 03 12 Note Increasing I Probably Increasing Pl Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events ND Non detect Moments are not calculated for sample events with less than 6 wells MAROS Version 2 2002 AFCEE 12 1 2003 Page 1 of 1 MAROS First Moment Analysis Project Tutorial User Name Charles Newell Location Service Station State Texas COC BENZENE Distance from Source to Center of Mass Mann Kendall S Statistic 83 P S 99 5 d A Conti See ees KO rm se a in ET 100 096 e 1 2E 02 Coefficient of Variation o 1 0E 02 0 22 a3 o 1 a oO 8 0E 01 First Moment Trend 5 6 0E 01 o 5 4 0E 01 2 Q 2 0E 01 0 0E 00 Data Table Effective Date Constituent Xc ft Yc ft Distance from Source ft Number of Wells 10 4 1988 BENZENE 48 58 71 12 1117 1989 BENZENE 39 51 64 12 3 1 1990 BENZENE 49 64 81 12 5 31 1990 BENZENE 50 52 72 12 9 13 1990 BENZENE 44 64 77 12 4 3 1991 BENZENE 42 58 7i 12 7h0 1991 BENZENE 42 64 76 12 10 3 1991 BENZENE 45 65 79 12 5 2 1992 BENZENE 43 77 89 12 1 11 1994 BENZENE 46 86 98 12 5 28 1996 BENZENE 44 81 92 12 6 27 1997 BENZENE 52 100 113 12 12 10 1997 BENZENE 49 108 119 12 6 19 1998 BENZENE 58 101 116 12 12 19 1998 BENZENE 59 106 122 12
47. 7 Moment Analysis M ann Kendall Second Moment Trend Results Decreasing Trend in both Sxx and Syy both parallel and perpendicular to the plume center line no change in Mass over time Version 2 1 A 5 9 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Dissolved Mass Dissolved Mass 100 kg 100 kg Ss gt No Change in Syy Only Syy Decreases t Figure A 5 8 Moment Analysis Mann Kendall Second Moment Trend Results Decreasing Trend in Syy perpendicular to the plume center line no change in mass over time Redundant Well Removal M oment analysis can also be used to evaluate the effect of removing wells from a monitoring program The question this analysis answers is whether or not removing a well from the well network will appreciably effect future plume delineation The application of this technique involves analyzing how the moments would change if wells were removed from historical data sets Historical data used in plume delineation is evaluated for zeroth first and second moments including all wells in a monitoring program and then again excluding the wells proposed for elimination The values determined for mass center of mass and spread of mass can be compared to determine how plume delineation would change if wells are removed If removal of a well has significant impact on plume delineation then the well should be maintained in the monitori
48. A Annual Mann Kendall Trend Figure A 3 1 Decision Matrix for Determining Frequency As is shown in the above three major steps the Modified CES method is concerned not only with the magnitude of ROC but also with the direction of change The GSI style M ann Kendall analysis is adopted because it can perform distribution free test and provides us the direction of change Usually people are more concerned with increasing trend than decreasing trend assuming they have the same ROC Regulator tends to impose more stringent sampling plan if the trend is increasing An increasing trend can cause the concentration exceeding MCL and make a well non compliant On the contrast a decreasing trend may drop the concentration below MCL and turn the well into compliance All these examples indicate that attention must be paid to the direction of trend as well as the magnitude of trend As discussed above the modified CES method incorporated these considerations into the whole process of decision The final results include the 4 57 amp 84p8 0RETHANE BENZENE TOLUENE recent result based on the analysi S of recent data overal The results of each monitoring well For a certain COC are listed below result based on the analysis of Well Name Sampling Frequency Recent result Overall result overall data and the final MAUI rp Arda I recommendation after two steps wi Annual Annual Annual of adjustments As is shown in M13 Annual Ann
49. ANALY SS Select the Graph button to display the graph for MW 1 Mewtret eg are Haracdistuee Ogibrizaiisp Syston WARDS Reduced Data Plot po omy re ig Tuin d To view data for a well MW 4 click the M e ici HH dond Paes z down arrow in the first text box Wall where MW 1 is displayed A list of choices will appear Click on MW 4 to select well MW 4 Select the Graph button to display the graph for MW 4 NU H 3 Li 3 i A graph of benzene concentrations for well MW 4is displayed TREND ANA YSS Version 2 1 A 11 21 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Note If more than one COC was being used data for other chemicals can be displayed by dicking on the down arrow of the second text box Chemical The graph type can be changed from Linear to Logarithmic by selecting the Log option under Graph Type After any change dick the Graph button to display the graph 6 Usethe View Report option to print the current graph and data Mewtecteg ond sisah ii IR e iMi i Reduced Data Plot Lorena ne vad soci cem d ot Commer Try he ront tamem Ten nes ent mi s rit ne m ast mm gr we FI Enae Click here to Morrtering amd eemeetetisn Optiretration Systir MAHL ee aed s TREND ANA YS Click hereto proceed The report displays the data in graphi
50. ANALYSIS MAROS analysis allows user to weight the trend analysis data and weight well data Final suggested monitoring system categories for each COC are displayed 1 From the Plume Analysis M enu select the Step 3e MA ROS analysis option The Statistical and Plume Information Summary by Weal screen will be displayed Click hereto proceed Dat Consoldelian Dey uber vooombmo ty mtem dmn rr rm meme wj OF HOM cin 3i Plume Analysis Mersini m mm Imm Dod Drm aem bip Se Spatial Moment Anatyots stad tuus i Peso Acus Estoma F ge rtometion feme Pe on Iran m comen P m aero Cle MAROS Analysis Mory Pr eget un casita Aere vana ma REND ANAL YSS Male Vim dew 2 Statistical and Plume Information Summary by Wal allows the user to view the Mann Kendall Trend Analysis Linear Regression Analysis Modeling and Empirical results by well and constituent Results for benzene are displayed Statistical and Plume information Summary ly Well Select Next to proceed to the Statistical and Plumelnformation Summary W eighting screen T tt Dadig 9 9 Oca miae voeem LOC n gne d M ae ai mei n Vui Tov vtm oe dba Fem sat cni im rem CT ilegal im BEONE Wel hae Tere Tai Maen opsi Sepre bebeg Pew a X 4 s ne na pee jeer w 3r Vestae ii Ninh y FT eie ie p aa P pa TT MT e HINTS Ft adn MGR esa Pom Click hereto proceed
51. AROS Output Screen or by dicking on View Report Back Returns the user to the Second M oment Plot Next Takes the user to the Plume Analysis M enu Screen Help Provides information on the screen specific input requirements Meeting smi erred lection Tetirerution System WRIS Moment Analysis Complete aa Vedat Arcades l on Lomo Lt Laud ia mey ror pto bas jw at Setar Flan d cara os ba itm cedar ard o mp ni ew Contin in Sieg YY v is i 7 gt 6 8 g F q n vi 43 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE External Plume Information External Plume Information M odeling Results accessed from the Plume Analysis M enu screen allows the user to enter statistical modeling results by well and constituent or for all source or all tail wells zax External Plume Information Modeling Results Options include entering modeling trend results i based on separate modeling studies DENS for both source and tail wells ii individual wel trends based on separate modeling No separate modeling studies have been perfomed studies If there are no modeling results Fides ee choose the option No separate modeling sacewels ae a studies have been performed C Edit individual well tends based on separate modeling studies Back Returns the user to the Plume A nalysis 2 Menu Eq E Next Takes the user to the External Plume
52. Analysis Options Histo ol dne foe a COC vg these noto Low We ot ctus mzwanit ctis Got to deu Appendix A 9 for details These parameters include Low Rate M edium Rate and High Rate Here Cleanup Goal PRG Preliminary Remediation Goal mg L is used as a reference for defining the rate of change parameters By default the low rate is defined as 0 5 PRG year medium rate is defined as 1 0 PRG year and high rate is defined as 2 0 PRG year for acertain COC When Cleanup Goal of a COC is not available in the database the user is prompted to enter the value and the three rate parameters Otherwise this COC will not be analyzed The user should provide specific Rate of Change values for a specific field of study if available Refer to Back Closes this screen and returns to the Sampling Frequency Analysis screen Set to default Sets all these parameters to system default Help Provides additional information on software operation and screen specific input requirements Version 2 1 October 2004 70 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Sampling Frequency Recommendation Sampling Frequency Recommendation accessed from the Sampling Frequency Analysis screen by clicking Analysis is used to display the frequency of sampling for each sampling location and each COC amp Monitoring and Remediation Optimization System
53. B Probability 1 D Correct Conclusion False Positive Rate Probability 1 Probability o Contaminated The type of error that may cause facility wide problems TYPES OF GROUNDWATER MONITORING IN A LONG TERM MONITORING PROGRAM Thetype of groundwater monitoring program used affects the development of strategies for the control of false positive and false negative rates In long term monitoring programs three kinds of groundwater monitoring may be involved detection monitoring compliance or assessment monitoring and corrective action monitoring These three types of groundwater monitoring are mandated by the Resource Conservation and Recovery Act RCRA and the Comprehensive Environmental Response Compensation and Liability Act CERCLA or Superfund The purposes and relationships of these monitoring programs are listed in Table A 7 6 As is shown in Table A 7 6 each of the monitoring programs addresses different problems and therefore requires different statistical methods for testing the corresponding hypothesis Since each monitoring program has a different objective definitions of null hypotheses and their implications are different Table A 7 4 and Table A 7 5 illustrate the differences in definition of the two types of error for detection monitoring and corrective action monitoring In detection monitoring the false positive refers to the decision that the contamination is present in the groundwater wh
54. Both the statistical methods used to evaluate trends Mann Kendall and Linear Regression gave similar trend estimates for each well e lsourcewell appears to have a suspect data point which should be investigated further MW 2 e The dissolved mass is decreasing over time whereas the center of mass shows an increase in distance over time in relation to the source location The plume is spreading in the direction of groundwater flow and probably increasing in the direction perpendicular to groundwater flow However the trend results do show overall decreasing concentrations in individual wells e Overall plume stability results indicate that a monitoring system of Limited intensity is appropriate for this monitoring network dueto a stable Upper Aquifer plume e The well redundancy optimization tool using the Delaunay method indicates that 1 existing monitoring wells may not be needed for plume monitoring and can be eliminated from the original monitoring network of 12 wells without compromising the accuracy of the monitoring network e The well sufficiency optimization tool using the Delaunay method indicates that no new monitoring wells are needed for the existing monitoring network e The wel sampling frequency tool the Modified CES method indicates the number of samples collected over time can potentially be reduced by 56 by sampling at a less than quarterly frequency for most of the monitoring wells considering the sampling
55. Decreasing 29596 Increasing Decreasing COV Coefficient of Variation The MAROS Linear Regression Analysis Decision Matrix was developed in house by Groundwater Services Inc The user can choose not to apply one of the two statistical plume analysis decision matrices Choose N ot Used in the Trend Results weighting screen If the user would like to use another decision matrix to determine stability of the plume they would need to do this outside the software Further Considerations The results of a constituent concentration trend analysis form just one component of a plume stability analysis Additional considerations in determining the over all plume stability include e Multiple constituent concentration trend analyses e Adequate delineation of the plume e Proximity of monitoring wells with stable or decreasing constituent trends to the downgradient edge of the plume References Gilbert R O 1987 Statistical M ethods for Environmental Pollution M onitoring Van Nostrand Reinhold N ew York NY ISBN 0 442 23050 8 Berthouex P M and Brown L C 1994 Statistics for Environmental Engineers CRC Press Boca Raton FL ISBN 1 56670 031 0 Gibbons R D 1994 Statistical M ethods for Groundwater M onitoring John Wiley amp Sons New York NY ISBN 10158 0012 Fisher L D and van Belle G 1993 Biostatistics A M ethodology for theH ealth Sciences John Wiley amp Sons New York NY ISBN 0 471 58465 7 Hollan
56. Dese tere Hs ANONO Io e con Statielicn Plume Analyode tensile ptr Schon Spatial Moment Anatyois ated memi Am Acasa External F ne rdomestion bera Ig ot nr m ommo Pme n e ft iti doma m s ANALYSES Click hereto proceed The Data Reduction Part 1 of 2 menu allows the user to consolidate the data based on time intervals and parameters chosen The Period of Interest option allows the user to specify which time period will be considered The Data Consolidation option is used to define the time period to consider within the datase and to define the representative statistical dataset within the consolidated time interval For this tutorial the full dataset will be used and no data consolidation will be performed This is appropriate for the small size of the dataset Mesetert eg are evediation Dortkrisaiiar Syeterr WIS a gt E amp Data Reduction Part f of 2 Under the heading Period of Interest the two empty text boxes should be left blank This means that the full dataset will be used Under the heading Data Consolidation the first option Do Not Perform Time Consolidation should remain selected Select Next to proceed to the Data R eduction Part 2 of 2 screen Click hereto proceed N ote If the user wishes to perform time consolidation one of the options in the bottom right of the screen median average etc needs to be selected to define
57. Duration T yr monitoring guidance Monitoring Frequency 4 per yr guidance _ State Regulations Monitoring Points Comments Geochemical Evidence No specifics guidance in Parameter Evidence ss guidance for well purge Notes fpr free phase free product removal RA risk assessment mon monitoring NR not regulated gen genetic a possible MTBE 98 regulation value range in parenthesis b EPA Health Advisory MCL driving state regulation enactment ss site specific sh st pl shrinking or stable plume sc site characterization go indicators geochemical indicators shsc source hot spot control de con decreasing concentration Back Additional Data TREND ANALYSIS External Plume Information Screening Criteria accessed from the External Plume Information Empirical Evidence screen gives the user additional guidance for empirical evidence for trends by State To view information pertaining to the state of interest choose the state name from the drop down box at the top left Information on general guidelines and regulations specific for Long Term Monitoring are shown Back Returns the user to the Empirical Evidence x Additional Data Takes the user to the Screen Criteria by State Back Returns the user to the Empirical Results Help Provides information on the screen specific input requirements Sources for this information include Marti
58. E previous analyses by changing parameters or S e T E selecting a different time period ala Sacer Ani You may dec och a a D ME eo lope oec Data Sufficiency Analysis M enu Returns the user Z wisdom i eor to the D ata Sufficiency Analysis M enu screen E a lt lt Back 2 Version 2 1 79 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Parameters for Risk Based Power Analysis This screen accessed from the D ata Sufficiency Analysis M enu screen by clicking Analysis 2 is used for spedfying the parameters used in the risk based power analysis The user should ensure the correctness of the selected parameters before proceeding to further analysis E Monitoring and Remediation Optimization System MAROS x 1 Groundwater Flow Angle The Parameters for Risk Based Power Analysis preferential groundwater flow direction The following parameters are needed for the risk based power analysis The user should provide representative measu red i n degrees counter d ockw i se wells along the plume centerline for a regression of concentrations against diatance down the plume centerline from the X axis or the Easting in State coordinate systems If the angle is provided earlier in the Plume Moment Proceed Through Steps 1 4 1 Groundwater Flow Angle 3 Select Sampling Events for Analysis Counter clockwise From
59. EPA 1992 within the data and the Shapiro Wilk test is superior to most other tests for testing normality of the data EPA 2000 In normal probability plots an observed value is plotted on the x axis and the proportion of observations less than or equal to each observed value is plotted as the y coordinate The scale of the plot is constructed so that if the data are normally distributed the plotted points will approximate a straight line Visually apparent curves or bends indicate that the data do not follow a normal distribution Evaluation by means of normal a probability plot is only qualitative As a quantitative test the Shapiro Wilk test is based on the premise that if data are normally distributed the ordered values should be highly correlated with corresponding quantiles taken from a normal distribution The Shapiro Wilk test statistic W will be relatively high if the normal probability plot is approximately linear When the normal probability plot contains significant bends or curves the statistic will be relatively low If the Shapiro Wilk test is applied to data from multiple wells eg background wells the spatial variability both mean and variance differences among wells exhibited in data from these wells must be negligible Otherwise one should use the multiple group version of the Shapiro Wilk test ASTM 1998 which is suitable for the joint assessment of normality in multiple wells Details about the multiple group Shapir
60. Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MTBE PLUME TREND DATA USED FOR STEP 4 Caution should be take before using MTBE plume distributions as secondary evidence as Happel et al 1998 concluded that most of the MTBE plumes are not stable compared to the contaminant e g BTEX plumes Although our results using 1995 96 data indicate that at the majority of sites individual MTBE plumes were nearly equivalent or shorter than their corresponding benzene plumes defined by action levels of 20 and 1 ug L 1 respectively our results predict that at a portion of these sites this relationship will change over time as the contaminant plumes gradually dissociate Happel et al 1998 The Texas study arrived at the opposite conclusion however Analysis of temporal data 83 percent of wells have stable decreasing or nondetection of MTBE concentration co occurrence with benzene has remained the same for the past several years and limited plume length data shows sites with stable plumes suggests that MTBE plumes may be naturally attenuated at many sites in Texas Mace and Choi 1998 More research is needed before MTBE plume a thon data can be used as adequate secondary evidence for determining plume stability Version 2 1 A 4 11 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D Empirical Data Chlor
61. Ging where Ciayg is the geometric mean concentration of the each triangle for a particular COC i Xi Y are the spatial coordinates of the center of each triangle Vj is the volume of the triangle calculated by d A where d is the averaged saturated thickness and Aj is the area of the triangle and Xc Yc arethe coordinates of the center of mass Version 2 1 A 5 3 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Calculation of the first moment normalizes the spread by the concentration indicating the center of mass Analysis of the movement of mass should be viewed as it relates to 1 the original source location of contamination and 2 the direction of groundwater flow Spatial and temporal trends in the center of mass can indicate spreading or shrinking or transient movement based on season variation in rainfall or other hydraulic considerations No appreciable movement or a neutral trend in center of mass would indicate plume stability Distance from Source to Center of M ass To calculate the distance from the center of mass of the plume for a particular COC and sample event to the source location the following formula is used D onna al sce X y s Y y where D romcente is the distance from the source location to the center of mass for a particular COC i and sample event Xc Yc are the coordinates of the center of mass Xsource Y source ar
62. Help The functions accessed by each choice are as follows D ata Consolidation Allows reduction of data based on dates as well as consolidating duplicates based on statistical functions i e average median etc This step also allows for assigning values to non detects and J flag data Statistical Plume Analysis Perform M ann K endall Analysis and Linear Regression Analysis Spatial Moment Analysis Perform Moment Analysis Zero First and Second Moments calculated External Plume Information Enter applicable modeling data and or empirical data MAROS Analysis Allows user to weight the trend analysis data and weight well data Final suggested monitoring system categories for each COC are displayed Help Provides additional information on software operation and screen specific input requirements Version 2 1 28 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D ata Reduction Data Reduction Part 1 of 2 accessed from the Plume Analysis M enu screen allows the user to consolidate the data based on time intervals and parameters chosen Steps for use amp Monitoring and Remediation Optimization System MAROS 1 x Data Reduction Part 1 of 2 1 The box at the top of the screen indicates Ss eT the current dataset ti me range This is From 10388 Too 12 19 1988 the location to specify the date range for Spec
63. Hillgwdata TES No field names should appear in the files There is a limit on the amount of data that can be opened in Microsoft Word 2000 this will be controlled by the amount of RAM in your computer The rule of thumb for large files is that your computer should have at least 3 times the amount of RAM as the size of the file For instance if you have a 80 MB file you should have at least 256 MB of RAM to open this type of filein Word If you do not know the amount of RAM on your computer from the Start Button go to Settings and Control Panel In the control panel open the System Icon and look at the General tab This indicates the amount of RAM in your computer Toimport ERPIM S files from an Access 2000 database 1 Typeor select the name of the A ccess 2000 database 2 Ensure that the tables induded in the database file are named as follows SAM TES RES and LDI data tables The import option requires an Access file format with fields identical to those outlined in Appendix A 7 Each field must have the mandatory columns filled in Do not import files with missing data this will result in incorrect data evaluation within the software The columns must include the field names as outlined in Appendix A 6 The template file MAROS ERPMSA ccessTemplate mdb is provided with the software with example data Version 2 1 16 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SY
64. Ho Site N ot Contaminated Hx Site Contaminated Not Contaminated Correct Conclusion False Positive Probability 1 Probability a False Negative Correct Conclusion power SITIS Probability Probability 1 B Version 2 1 A 6 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The power of statistical tests is dependent upon the following design parameters 1 the false positive rate also called the significance level 2 level of sampling effort i e number of sampling points frequency and duration 3 minimum detectable difference in the effect that can be detected and 4 natural variability within the sampling environment This relationship betw een the power of a statistical test and the design parameters makes several types of power analyses possible The power of the test can be determined as a function of any of these design parameters Alternatively the value of any individual design parameter required to obtain a specified power of a statistical test can be determined as a function of the other parameters With this type of approach a relationship between the number of sampling locations sampling frequency the minimum difference that can be detected in the monitored variable and the natural variability of the monitored variable can be established and their trade offs can be studied For example Figure A 6 1 include
65. If more than one COC was being used the user would navigate the results for individual eiT 5 emi constituents by clicking on the tabs at the top of VII Marron Sane the screen The results can be printed by selecting the View Report button 9 Moment Analysis Complete screen indicates that the Spatial Moment Analysis has been performed Montlorlig and Remediation Opiinizatinn Sysiiem MAMDS Click on Plume Analysis to return to Moment Analysis Complete proceed to the External Plume ow Information The Plume Analysis M enu will appear Led sra sipoh tiae ad ae apn Click hereto proceed Version 2 1 A 11 35 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 5 EXTERNAL PLUME INFORMATION The external plume information is used when applicable modeling data and or empirical data is entered This portion of the software is an optional utility which will not be used in this tutorial 1 From the Plume Analysis M enu select the Step 3d External Plume Information option Plume Analysis Menu oi comme d Ps mum aec M trions The coer oto den iau nonet Ly Dh dg FO Vw ai Pt bns Med s Rt arte ient veg t TR va ma CT gr m Lapat Fuss vm Prati Siya da de The External Plume Information M odding 7 i 5 6 DaaCensobdelian Results screen will be displayed rie Dey e nnm He NOTI Pah ot ee C
66. LABLOTCTL Laboratory Lot Control Number 18 LABLOT SEQ Sequence Number of Lab Lot 19 CA LREFID Reference Link Between Samples and Corresponding Calibration 20 RTTYPE Remediation Technology Type 21 BASIS Basis Version 2 1 A 2 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 1 4 REQUIRED FIELD FORMAT FOR SAM IMPORT FILES SAMPLE RESULTS Column Number Field Name Description 1 SAM PLESEQ Sample Sequence N umber 2 AFIID Air Force Installation 3 ICONTRACTSEQ Contract Sequence N umber 4 LOCID Location Identifier 5 LOGDATE Log Date Note the time of sampling should NOT be included 6 MATRIX Sampling Matrix 7 SBD Sample Beginning Depth 8 SED Sample Ending Depth 9 SACODE Sample Type Code 10 SAMPNO Sample Number 1 LOGCODE Logging Company Code 12 SMCODE Sampling M ethod Code 13 WETCODE Moisture Content 14 FLD SAMPID Field Sample Identifier 15 COOLER Cooler Identifier 16 COCID Chain of Custody Identifier 17 IABLOT Ambient Blank Identifier 18 EBLOT Equipment Blank Identifier 19 TBLOT Trip Blank Identifier 20 SA PROG Program Authorization 21 REMARKS Comments A bout the Sample Version 2 1 A 2 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 1 5 REQUIRED FIELD FORMAT FOR RES IMP
67. M enu Item Tools References A pop up list of items will appear Choose the following libraries to utilize Click on the following libraries IF they are not already chosen Visual Basic for Application M icrosoft Access 9 0 Object Library M icrosoft DAO 3 6 Object Library Microsoft Graph 8 0 Object Library M icrosoft Excel 9 0 Object Library M icrosoft Office 9 0 O bject Library Click on OK when finished 4 Exit Access from the M enu Item File Exit Version 2 1 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MAROS SOFTWARE STEP BY STEP MAROS Step by step instructions will guidethe user through the most commonly used features of the software Figure 1 directs the user through the complete MAROS program flow which will assist the user in becoming familiar with the use of the software What do I need before start The MAROS Software requires a small but specific set of data in order to produce a result The data must be carefully formatted to fit the entry requirements in MAROS Data preparation is often the most difficult and time consuming part of the analysis Detailed descriptions of import file formats are presented in Appendix A 1 1 Well sampling data including the well name constituents sampled sample dates results and well locations should be entered into either Excel or Access as described in Appendix Al of this ma
68. OPTIMIZATION SYSTEM SOFTWARE 3 ADJUSTMENT BASED ON MCL If the maximum concentration in the sample is less than one half of the MCL and if the trend of COC in this well is not increasing we can reduce the sampling frequency by one level Because at such a low concentration level and with confidence that it will not increase the adjustment will not cause great risk The steps to be followed are shown in the following flow chart In addition wells that have attained cleanup standards their long term concentrations were far less than MCL can be eliminated from the monitoring network to further optimize the monitoring program Some of the empirical rules are referred to NFESC 2000 Maximum value in recent data is less than one half of COC s MCL Biennial is also made in three consecutive Annual recommendations The above determined frequency is Annual and Current CT is Stable ProbDecr or Decr Wells that have attained cleanup standards and are not critical sampling can be stopped The above determined frequency is SemiAnnual and Current CT is not Incr The above determined frequency is Quarter and Current CT is not Incr Version 2 1 A 9 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Rate of Change Linear Regression High MH Medium LM Low Sampling Frequency Q Quarterly S SemiAnnual
69. OPTIMIZATION SYSTEM SOFTWARE Steps for use 1 Start the program only if it is not automatically loaded by clicking the INIT A pply button in the W ell Locations chart sheet The Delaunay triangles are plotted by default Set the optimization and drawing control parameters in the O ptions Form Activate this form by clicking the Options button in the Well Locations chart sheet You can skip this step if you want to usethe default parameters If you do not want to see graphs in the plot area dick the Clear Resume button in the W el Locations chart sheet Clicking it again will turn on the graph output You can also achieve this by deselecting the two drawing parameters in the O ptions Form If you want to use all locations as potential locations for analysis when some of them have previously been eliminated click the Reset All button in the Well Locations chart sheet This action will reset the potential locations and redraw the graph Perform optimization by clicking the Optimize button in the chart sheet Well Locations If locations are eliminated from the network you may notice the change in the graph if the graph output is turned on Check the results in the plot area in the Well Locations chart sheet or in the Output Part in the D ataSheet If you want to change parameters and run the analysis again go back to step 2 Send results back to Access the W dl Redundancy Analysis Excel M odule screen by clicking the Back to Access butt
70. Plot screen will appear proceed 3 TheM ann Kendall Plot screen allows the user to view the Mann Kendall Trend Analysis results by well and constituent E Mevetia g ard Kard bat ley Devt hela bee Synen MAIO f per Mann Kendall Plot Graph of benzene concentrations for well MW 4 is displayed The Mann Kendall statistics are displayed for this well For example the Concentration Trend is shown to be decreasing D in the box in the left hand bottom corner Select the Next button to continue to theLinear Regression Statistics screen med brie wate wich Ro caca und pall oct o De palla dont we 73 ases EXT MP Cocco Tied gt aay rtm fe Rati Paty D wem fOr r HAAA rst P eS DARI rmt mtn dh Click hereto proceed 3423 Vere Hepar ome Note As discussed above plots of other wells and chemicals can be obtained using the Well or Chemical drop down boxes in the top of the screen followed by seleding the Graph button The graph type can be specified as Log or Linear The graph can be printed by selecting the View Report button Version 2 1 A 11 26 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 4 Linear Regression Statistics allows the user to view the Linear Regression Analysis results by well and constituent The Linear Regression analysis is another statistical m
71. REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 11 1 SAMPLING LOCATION OPTIMIZATION RESULTS BASED ON THE DELAUNAY METHOD Well Usedin MAROS Final Reasoning analysis Results Recommendation MW 1 Yes Keep Keep A downgradient well on the plume centerline providing important MW 12 Yes Eliminate Keep information for plume delineation and stability calculation It needs to be kept MW 13 Yes Keep Keep MW 14 Yes Keep Keep MW 15 Yes Keep Keep MW 2 Yes Keep Keep In the source area of the plume where MW 3 Yes Eliminate Eliminate well density is high It can be eliminated without significantly affect the plume characterization A down ross gradient well close to the plume centerline used to monitor I the lateral migration plume If the MW 4 Yes Eliminate Keep aS plume is proved to be shrinking and falls to below detection level at this area this well can be eliminated MW 5 Yes Keep Keep MW 6 Yes Keep Keep MW 7 Yes Keep Keep MW 8 Yes Keep Keep Sample events 10 to 15 were used in the above analysis The analysis parameters are 0 20 0 01 0 95 and 0 95 for Inside Node Slope Factor Hull Node Slope Factor Area Ratio and the Concentration Ratio respectively Version 2 1 A 11 61 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 11 2 SAMPLING FREQUENCY OPTIMIZATION RESULTS BASED ON THE MCE
72. S N A N A MW 7 S 15 1 5 3E 04 5 0E 04 No S S N A N A MW 6 S 15 0 5 0E 04 5 0E 04 Yes S S N A N A MW 5 S 15 15 1 2E 00 1 2E 00 No D D N A N A MW 3 S 15 12 6 9E 02 6 0E 02 No D D N A N A MW 2 S 15 7 2 0E 02 5 0E 04 No NT PD N A N A MW 1 S 15 15 1 0E 00 8 0E 01 No D D N A N A MW 4 T 15 14 5 8E 02 1 8E 02 No D D N A N A MW 15 T 15 0 5 0E 04 5 0E 04 Yes S S N A N A MW 14 T 15 7 1 1E 02 5 0E 04 No D D N A N A MW 13 T 15 10 1 8E 02 1 5E 02 No D D N A N A MW 12 T 15 11 4 7E 02 2 2E 02 No D D N A N A Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data lt 4 sampling events Source Tail S T The Number of Samples and Number of Detects shown above are post consolidation values MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 1 O MW 6 S Trend Results for BENZENE A MW 13 D Asp O MW 3 D 50 Y Coordinate O MW 7 S AFCEE Long Term Monitoring Software 50 1A0vw 12 D O MW 1 D X Coordinate ATail Wells OSource Wells rend Result Increasing I Probably Increasing PI No Trend NT Stable S Probably Decreasing PD Page 1 of 1 MAROS Sampling Location Optimization Results Project Example User Name Meng Location Service Station State Texas Sampling Events Analyzed From Sample Event 10 to Sample Event 15 1 11 1994 12 19 1998 Paramet
73. The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average Values near 1 00 indicate that the data form a relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about the mean View Report To print the Second Moment Change Plume Spread Over Time Report and analysis results click View Report to proceed Back Returns the user to the First M oment Plot Changein Location of M ass O ver Time screen Next Takes the user to the Spatial M oment Analysis Summary screen Help Provides information on the screen specific input requirements Note The information displayed in this screen can also be viewed in report form Second Moment Change Plume Spread Over Time Report from the M AROS Output Screen or by clicking on View Report see Appendix A 10 for an example report For further details on the Mann Kendall Analysis Method or Moment Analysis see Appendix A 2 and A 5 respectively Version 2 1 42 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Summarizing Spatial M oment Analysis Spatial M oment Analysis Summary accessed from the Second M oment Plot Changein Plume Spread Over Time screen allows the user to view the Moment Analy
74. Total Total coc Wells Excedences detects LEAD 12 10 10 BENZENE 12 10 10 H 4 1 2 TETRACHLOROETHANE 12 8 8 BARIUM 12 7 12 2 DICHLOROBENZENE 12 7 12 TOLUENE 12 5 12 PERCHLORATE 12 5 10 COPPER 12 4 12 ETHYLBENZENE 12 1 10 concentration for each well location at the site The total excedences valu the compound ID Insufficient Data SITE DETAILS Version 2 1 October 2004 Repr Conc 4 Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG Below PRG jj Note Top COCs by prevalence were determined by examining a representative ies above the chosen PRGs are compared to the total number of wells to determine the prevalence of COC Decision Prevalence shows a list of COC recommendations from the available dataset based on the Prevalence of the compounds Top COCs by prevalence were determined by examining a representative concentration for each well location at the site The total excedences values above the chosen PRGs are compared to the total number of wells to determine the prevalence of the compound Compounds listed first are those above the PRG and are shown on the COC Decision screen Back Returns the user to the COC Decision screen 25 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Help Provides information on the screen specific input requirements Version 2 1 26 Air Force Center for Octobe
75. U S EPA 1992 p9 12 The log transformed yearly averages are used if they are more likely to be lognormally distributed Second the likelihood ratio estimator LR is calculated as LR edo mo i Equation A 6 1 n n l t where n is the number of yearly averages or log transformed yearly averages LR is then compared with two critical statistics A and B to determine the cleanup status A and B are defined as A Pp and g d B Equation A 6 2 1 04 where a is the type error i e significance level or false positive rate and B is the type II error i e false negative rate When LR lt A cleanup standards have not been attained When LR gt B deanup standards have been attained statistically significant When LR is between A and B future tests need to be performed when more sampling data become available not statistically significant In the MAROS Data Sufficiency Analysis module a well is considered to have attained the cleanup standards only when LR gt B and the concentration trend is not Increasing as defined in the Modified CES method The sequential t test uses an easy to calculate approximation for the likelihood ratio The use of log transformed yearly averages improves the test performance with skewed data It reduces the number of samples compared to that for an equivalent fixed sample size test and has a low false positive rate and an acceptable false negative rate According to the simulation resul
76. a GIS software or export data to other software programs The database containing results can be compared against other MAROS runs for the same data set to evaluate the impact of changing parameters such as hydraulic characteristics different methods of data consolidation and data time periods M ain Menu Returns the user to the M ain M enu Help Provides information on the screen specific input requirements The Export M AROS Analysis Results to A ccess File Bin MANOS Aid GU MR accessed from the MAROS Output Reports Acoria PHA screen allows the user to export MAROS 1 analysis results to a Microsoft A ccess file Tus n roa ft e pet patey ua fw Dore mAn i ura Dew zresa toia Fires nicks that w itir ream oct dn arta rm est td Fan d em rien Syslere MAS To export results into a database 1 Enter the full file path and filename of the archived file to export or dick the browse button to find the archive file to overwrite The Folder and File name you choose will appear in thetop two boxes 2 Click Create to proceed with exporting the data to a database file DATA MANAGEMENT Back Takes the user back to the M AROS Output R eports Graphs screen Help Provides information on the screen specific input requirements Version 2 1 87 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Trend Summary Results
77. a dramatic decrease in the statistical power of an individual comparison For example if o 0 05 and n 50 a is then 0 001 This is one tenth of the regulatory performance standard of RCRA Interim Final Guidance Document EPA 1989 which requires that the comparison wise false positive rate should be no less than 0 01 Qa The use of verification resampling can solve this problem Davis et al 1987 found that the controlled use of verification resampling can control FWFPR while maintaining sensitivity to contamination There are two types of widely used verification resampling strategies 1 of m plans and California plans 1 of m plans declare a statistically significant increase when an initial sample and all of m 1 resamples indicate exceedence California plans declare a statistically significant increase when both the initial and any of the m 1 resamples indicate exceedence An example of 1 of m plans illustrates the effects of verification resampling on the control of both FWFPR and the false negative rate Assuming a 0 01 and n 50 future comparisons for one verification resample the FWFPR is a 1 Probability all wells okay Version 2 1 A 7 9 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 1 Probability one well okay 1 1 a a l a 1 1 0 01 0 01 1 0 01 50 0 005 lt 0 05 In this case the verification resample has limited the FWF
78. allows the user to choose between performing Step 3a Data Consolidation Step 3b Statistical Plume Analysis Step 3c Spatial Moment A nalysis Step 3d External Plume Information Step 3e M A ROS Analysis Version 2 1 A 11 18 October 2004 formed eg ed RevwesTetlar Opiera p Spiere AARTS Plume Analysis Menu fen Ree eripe emo neo eraa den omm i Pm mm eoe Merano The uer Pro oT elie Ria Fg ai Fr Pale deo roni D Dh Mg FO Ys ai Pe bs Rehd i mtt Fa wen Phong Step L Te Snp O DaaCeonseckdelian Dey ee homm den rn nom he FT Pj e Sratinlicn Plume Analypis thet Leen eel emm eae te Phere anm Spatial Moment Anatyors oad emt Am Achy ate Step M j External Phryne irformstion MAROS Analysis Wet otongy tr epr in coe ea erm sana m Mh Mens ew REND ANALYSES Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 2 DATA CONSOLIDATION Data consolidation allows the reduction of data based on dates as well as consolidating duplicates based on statistical functions i e average median etc This step also allows for assigning values to non detects and J flag data 1 2 From the Plume Analysis M enu select the EE Step 3a Data Consolidation option Plume Analysis Menu RR Lm seti Co Pb aM Fe api Fo won The D ata Reduction Part 1 of 2 menu will SE ere Po Re Ren PIDE UTERIS appear Data C onsolbileli ary tmm
79. analysis The Target Level is the expected mean concentration in wells after cleanup attainment it is only used in individual well celanup status evaluation The test for evaluating attainment status is from EPA 1992 Refer to Appendix A 6 of MAROS Manual for details MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 1 MAROS Risk Based Power Analysis for Site Cleanup Project Example UserName Meng Location Service Station State Texas Parameters Groundwater Flow Direction 0 degrees Distance to Receptor 1000 feet From Period Sample Event 1 to Sample Event 15 10 4 1988 12 19 1998 Selected Plume Well Distance to Receptor feet Centerline Wells MW 12 1090 0 MW 4 1135 0 MW 1 1177 0 The distance is measured in the Groundwater Flow Angle from the well to the compliance boundary Normal Distribution Assumption Lognormal Distribution Assumption Sample Event ae Sample Sample Cleanup Power Expected Celanup Power Expected Alpha Expected zie Mean Stdev Status Sample Size Status Sample Size Level Power BENZENE Cleanup Goal 0 005 Sample Event 1 12 1 30E 15 1 76E 15 Attained 1 000 lt 3 Attained 1 000 lt 3 0 05 0 8 Sample Event 2 12 144E 22 1 41E 22 Attained 1 000 lt 3 Attained 1 000 3 0 05 0 8 Sample Event 3 12 5 12E 17 7 09E 17 Attained 1 000 lt 3 Attained 1 000 3 0 05 0 8 Sample Event 4 12 3 84E 15 5 90E 15 Attained 1 000 lt 8 Attained 1 000 lt 3 0 05 0 8 Sample Event 5 12 1 14E 23 1 50
80. analysis by changing parameters or selecting a different series of sampling vents Compare Across COCs Determines the conservative all in one results by considering all COCs and then proceeds to the A ccess M odule All in one Results screen Help Provides additional information on software operation and screen specific input requirements Version 2 1 56 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Access M odule All in one Results This screen accessed from the A ccess M odule Results by COC screen by dicking Compare A cross COCs is used to display the conservative all in one sampling location optimization results A location is marked for elimination only if this location is eliminated from all COCs Here elimination of a location is equivalent to stopping sampling at this location E3 Monitoring and Remediation Optimization System MAROS Ix Eliminated Displays whether or not a Access Module All in one Results location is considered redundant and The final sampling locations after considering all COCs are determined as shown in the should be eliminated A check mark in following table A sampling location is eliminated only if it is eliminated for all COCs z i Eliminated status can be interpreted here as stopping sampling a certain well in the this field stands for the elimination of a l
81. chemicals can be obtained using the Chemical drop down box at the top of the screen followed by selecting the Graph button The graph type can be specified as Log or Linear The graph can be printed by selecting the View Report button 5 First M oment Plot Distance from Source to Center of M ass is used to view the First Moment Analysis results by constituent over time The first moment estimates the center of mass coordinates Xc and Yc for each sample event and COC The distance from the original source location to the center of mass locations indicate the movement of the center of mass over time relative to the original source The First M oment trend of the distance to the center of mass over time is determined by using the Mann Kendall Trend Methodology The First Moment trend for each COC is determined according to the rules outlined in Appendix A 1 Results for the trend include Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not Applicable Insufficient Data Other statistics displayed include the Mann Kendall Statistic S the Confidence in Trend and the Coefficient of Variation COV Refer to Appendix A 1 and A 5 for further details First Moment analysis distance from First Moment Plot _ mea rer Saame M Letter of Mie source to center of mass results are ap reme man ra Bea rt ER E displayed for benzene s tma FORTE The First Moment Trend of the distance to the cente
82. concentrations at the site are truly below the cleanup level The power values range from 0 to 1 0 N C indicates the analysis is not conducted because of insufficient data sample size 4 S E indicates the analysis is not conducted because the mean concentration significantly exceeds the cleanup goal Sampling Event Sample Cleanup Power Expected 4 Distribution si te Resu Its Cou d be A ttai n ed N ot Size Achieved ofTest Sample Size Assumption i Sample Event 7 1 Not Attained SE SE Lognormal A ttai ned or N IC not cond ucted due to Z A insufficient data This evaluation is k Sample Evet 12 12 Atained 7 000 3 Views based on the estimated concentrations i Sample Event 13 12 Attained 1 000 3 f E Saige edd 127 eA VG 3 projected to the compliance boundary m pc oom E 2 4 pen and therefore is a risk based evaluation e S ve not conducted due to insufficient data S E sample mean significantly exceeds cleanup goal Pow er of T est Th e pro babi ity th at th e a i2 lt lt Back View Report Next Expected Sample Size The number of data required to achieve the expected power with the observed variability of the projected concentrations lt 3 indicates a very small data variability leading to a high power 2100 indicates the opposite N C indicates the analysis is not conducted because of insufficient data sample size 4 S E indicates the analysis is not conducted because the mean concent
83. data are strongly correlated methods like collapsing and averaging data over a longer interval Ward et al 1988 and 1990 can be applied to remove serial correlation Version 2 1 A 7 31 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 15 Example calculation for test of serial correlation Quarter Data mg L Residuals e i i ej e e 1 1 4 69 0 28 0 076 2 5 51 0 54 0 151 0 297 0 674 3 5 55 0 58 0 318 0 342 0 002 4 5 09 0 13 0 074 0 016 0 210 5 6 08 1 11 0 141 1 241 0 974 6 4 59 0 37 0 414 0 138 2 206 7 3 73 1 24 0 460 1 535 0 753 8 3 10 1 87 2 314 3 486 0 394 9 4 93 0 04 0 068 0 001 3 352 10 5 28 0 32 0 012 0 102 0 126 11 3 23 1 74 0 555 3 024 4 236 12 5 94 0 97 1 692 0 946 7 354 13 6 22 1 26 1 224 1 584 0 082 14 5 52 0 56 0 704 0 313 0 488 15 6 01 1 04 0 583 1 086 0 233 16 3 97 0 99 1 035 0 986 4 142 Mean 4 97 y 2 03 15 17 25 23 0 13 D 1 66 7 00 6 00 d 5 00 o E E 4 00 9 3 00 t co 9 2 00 o o 1 00 0 00 0 1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Time No Quarter Figure A 7 3 Time series plot of quarterly manganese concentrations Version 2 1 A 7 32 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE METHOD 9 METHODS FOR DEALING WITH NONDETECTS If data generated from chemical analysis contains nondetects i e measure
84. error N k Version 2 1 A 7 27 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 5 Consult an F distribution table with the desired significance level o k 1 numerator degrees of freedom and N k denominator degrees of freedom to find the critical F value If f is greater than F reject the assumption of equal variances Example Step 1 The arsenic data presented in Table A 7 13 are used again in Table A 7 14 second column to illustrate the Levene s test Assuming MW 1 and MW 2 are upgradient wells and MW 3 and MW 4 are downgradient wells we want to test the assumption of equal variances before using a parametric AN OVA test Each group mean is calculated and presented in the fourth column of Table A 7 14 For example a tY 7 13 964 12 77 9 66 8 46 11 21 Step 2 The absolute residuals are listed in the fifth column of Table A 7 14 Group means of these absolute residuals are 2 15 2 46 0 27 and 1 55 and the overall mean is 1 61 presented in the sixth and seventh columns of Table A 7 14 respectively For example za 7 x x 5 15 6 97 1 82 1 82 nj ie z Y 2 75 1 56 0 70 2 39 1 61 z 29 57 rea Step 3 The sums of squares for the absolute residuals are SS rora k n i l zy NZ 2 75 1 56 2 397 J 16x1 61 15 95 j l nz Nz Y x215 44x26 4x0 27 4 4x1 55
85. exceeds the specified value of the unsigned S when no trend is present So the confidence in the trend is calculated as 1 minus this probability Step 4 The mean and standard deviation of this sample are 0 024 and 0 010 respectively Therefore the coefficient of variation is COV 20 010 0 024 0 435 Step 5 From Table A 7 11with S lt 0 the confidence in the trend lt 0 and the COV lt 1 the concentration trend is Stable Two meanings are thus indicated 1 the slope of the times Version 2 1 A 7 23 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE series is not statistically significantly different from zero and 2 the fluctuation of benzene concentrations within the time period in this monitoring well is quite small Table A 7 12 Benzene data and computation of the Mann Kendall statistic Time 2 5 98 3 9 98 4 6 98 5 15 98 6 29 08 7 17 98 9 1 98 10 8 98 Data mg L 0 026 0 028 0 034 0 018 0 027 0 036 0 011 0 008 1 1 1 1 1 1 1 1 1 1 1 1 1 Sign of difference 1 1 1 1 1 between consecutive 1 1 1 1 measurements 1 1 1 1 1 1 Total No of signs 1 2 0 2 5 0 0 10 No of signs 0 0 3 2 0 6 7 18 Mann Kendall statistic 10 18 8 METHOD 6 METHODS FOR TESTING NORMALITY There are many methods available for checking the normality of data among which the normal probability plot is particularly useful for spotting irregularities
86. f s e b The t result along with the degrees of freedom n 2 are used to find the confidence in the trend by utilizing a t distribution table found in most statistical textbooks e g Fisher L D and van Belle G 1993 The resulting confidence in the trend is utilized in the linear regression trend analysis as outlined in Table A 2 2 RESULTS AND INTERPRETATION OF RESULTS LINEAR REGRESSION ANALYSIS The Constituent Trend Analysis Results are presented in the Linear Regression Analysis Screen accessed from the M ann K endall Analysis screen The software uses the input data to calculate the Coefficient of Variation COV and the first order coefficient Ln Slope for each well with at least four sampling events A Concentration Trend and Confidence in Trend are reported for each well with at least four sampling events If there is insufficient data for the well trend analysis N A Not Applicable will be displayed in the Concentration Trend column Figure A 2 2 BENZENE ETHYLBENZENE 1 2 DICHLOROBENZENE TOLUENE xvLENES TOTAL Statistical Analysis Results Last column is the result for the trend Confidence Concentration S T Average Ln Slope COV S 6 8E 04 8 5E 05 8 8E 01 S 546 04 3 1E 05 2 5E 01 S MWN 14 S S95E 03 1 0E 03 1 6E 00 99 6 D MWN 13 S 17E 02 1 E 03 1 1E 00 100 0 D MV 1 2 S 35E 02 1 7E 03 1 6E 00 100 0 D MwWN 1 5 3 6E 01 1 4E 03 1 7E 00 83 596 D MVY 6 T 5 0E 0
87. for determining the sampling frequencies Step 1 Set frequency based on recent trends Based on the trends determined by rates of change from linear regression analysis a location is routed along one of four paths The lowest rate 0 10 ppb per year leads to an annual frequency schedule The highest rate 30 ppb per year leads to a quarterly schedule Rates of change in between these two extremes are qualified by variability information with higher variability leading to a higher sampling frequency Variability is characterized by a distribution free version of the coefficient of variation the range divided by the median concentration with 1 0 as the cut off Step 2 Adjust frequency based on overall trends If the long term history of change is significantly greater than the recent trend the frequency may be reduced by one level If this is not so no change could be made Step 3 Reduce frequency based on risk Since not all compounds in the target list are equally harmful frequency is reduced by one level if recent maximum concentration for compound of high risk isless than one half of theMCL It was stated that the evaluation by CES should be performed at the end of each year s monitoring All the target chemicals should be evaluated to finally make the decision Latest updates by LLNL indude biennial sampling of the well if three successive annual recommendations are made and the cut off value of variability at high concentrations
88. for this tutorial has 4 COCs benzene ethylbenzene toluene and xylenes of which benzene will be used as the indicator compound for the plume All boxes will initially be blank Click on the arrow to the right of the top box to display alist of COCs Select benzene by clicking on benzene The top box should now have benzene displayed as shown Click Next to proceed Click hereto proceed Note The other drop down boxes can be used to change the COC or to select up to 4 additional COCs To view a list of suggested COCs click on the button Recommended COCs This will result in a summarized list of COC recommendations from the available dataset as well as a detailed view of the process used to make the COC recommendation For example the user can choose a preliminary remediation goal PRG to screen representative concentrations from the dataset The user can either select the appropriate dean up standard Region 3 Region 9 or TCEQ or custom goals can be specified There is also an option to have a detailed view of the process used to make the COC recommendation according to toxicity prevalence or mobility Version 2 1 A 11 16 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 7 VIEWING DATA The Initial D ata Table allows the user to view the initial data table with the COCs chosen as well as the sample events de
89. frequency reduction only e The MAROS Data Sufficiency Power Analysis application indicates that the monitoring record has sufficient statistical power at this time to say that the plume will not cross a hypothetical statistical compliance boundary located 1000 feet downgradient of the most downgradient well at the site With the progress of Version 2 1 A 11 69 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE remediation this hypothetical statistical compliance boundary will get doser and doser to the downgradient wells of the monitoring system The MAROS optimized plan consists of 11 wells 1 sampled semiannually 5 sampled annually and 5 sampled biennially The MAROS optimized plan would result in 9 5 samples per year compared to 24 samples per year in the original monitoring program Implementing these recommendations could lead to a 60 reduction in samples from the original plan in terms of the samples to be collected per year The recommended long term monitoring strategy based on the analysis of the original monitoring plan results in a moderate reduction in sampling costs and allows site personnel to develop a better understanding of plume behavior over time A reduction in the number of redundant wells an increase in the number of wells in areas with inadequate information as well as reduction in sampling frequency is expected to results in a moderate cos
90. gt 0s hey te fe SUNVILLE STREET FIGURE A 11 2 SERVICE STATION BENZENE TREND RESULTS Note If extraction or recovery wells are present in a well network these well trend results need to be treated differently for the purpose of individual trend analysis interpretation primarily due to the different course of action possible for the two types of wells For monitoring wells strongly decreasing concentration trends may lead the site manager to decrease their monitoring frequency as well look at the well as possibly attaining its remediation goal Conversely strongly decreasing concentration trends in extraction wells may indicate ineffective or near asymptotic contamination extraction which may in turn lead to either the shutting down of the well or a drastic change in the Version 2 1 A 11 43 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE extraction scheme Other reasons favoring the separation of these two types of wells in the trend analysis interpretation is the fact that they produce very different types of samples On average the extraction wells possess screens that are twice as large and extraction wells pull water from a much wider area than the average monitoring well Therefore the potential for the dilution of extraction well samples is far greater than monitoring well samples Version 2 1 A 11 44 Air Force Center for October 2004 Environmental
91. in data over time Gilbert 1987 The Mann Kendall test can be viewed as a nonparametric test for zero slope of the first order regression of time ordered concentration data versus time The AFCEE MAROS Tool indudes this test to assist in the analysis of groundwater plume stability and plume Version 2 1 A 5 1 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE changes over time The M ann K endall test does not require any assumptions as to the statistical distribution of the data e g normal lognormal etc and can be used with data sets which include irregular sampling intervals and missing data The Mann Kendall test is designed for analyzing a single groundwater constituent multiple constituents are analyzed separately For more details on the M ann Kendall Trend Analysis refer to Appendix A 1 ZEROTH MOMENT SHOWS CHANGE IN MASS OVER TIME The zeroth moment is the sum of concentrations for all monitoring wells and is a mass estimate The zeroth moment calculation can show high variability over time largely due to the fluctuating concentrations at the most contaminated wells as well as varying monitoring well network Plume analysis and delineation based exdusively on concentration can exhibit a fluctuating degree of temporal and spatial variability The mass estimate is also sensitive to the extent of the site monitoring well network over time Therefore the plume sho
92. into three levels Low Medium and High They represent the degree of change or how fast the concentration of COC change Ww i l l ap pear over the time period The unit for Cleanup Goal is mg L The units for rate of change parameters are ma L vear Monitoring and Remediation Optimization System MAROS COC name Cleanup Goal LowRate MediumRate High Rate BENZENE 0 005 0 0025 0 005 0 01 The Cleanup Goal is generally the site specific or risk based cleanup goal for a COC If the user does not provide this value the software will use the maximum contaminant level MCL for that COC instead The Low Rate Medium Rate and High Rate are threshold values used to dassify the rate of change i e the linear slope from concentrations vs time regression By default the Low Rate is defined as 50 change of the Cleanup Goal per year the Medium Rate as 100 change of the Cleanup Goal per year and High Rate as 200 change of the Cleanup Goal per year In this example the default values will be used Click the Back button to return to the Sampling Frequency Analysis screen Set to default Help Version 2 1 A 11 54 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Note The ROC parameters should be modified according to site specific conditions and needs For example higher ROC parameters can be appli
93. kept no significant information loss To ensure that the elimination of sampling locations from monitoring network will not cause significant information loss two indicators are developed to measure the information loss One is Average Concentration Ratio CR and the other is Area Ratio AR which are defined as CR EE C on AR A urreni avr Original Original where Cavr Current average plume concentration estimated after elimination of locations in the current step of optimization Cavr Origina average plume concentration estimated from the potential locations before elimination of any locations Acurrent Triangulation area based on locations after elimination of locations in the current step of optimization Aorigina Triangulation area based on potential locations before any optimization before elimination of any locations Version 2 1 A 4 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The average plume concentration is taken as the area weighted average of the average concentrations of all Delaunay triangles C Y rc TA i where TA l re I num 1 N number of all Delaunay triangles in the triangulation TA area of each Delaunay triangle i 1 2 N TC average concentration of each Delaunay triangle i 1 2 N TC is computed as refer to Figure A 2 3 rc NCCA amp NC A NC A A A
94. mean concentration in a well is significantly below the cleanup goal 2 the power associated with this test and 3 the expected sample size in order to achieve the desired power Because power analysis is difficult to perform for the sequential t test but easy for the Student s t test the optional power analysis is provided in MAROS as an alternative for assessing data sufficiency associated with the cleanup status evaluation To determine if the mean concentration is statistically below the cleanup goal a significance test based on the following statistic is used lt Equation A 6 3 ysin where c is the cleanup goal e g MCL m and s are the sample mean and standard deviation respectively n is the number of concentration data in the sample and t is the test statistic following the Student s t distribution with n 1 degrees of freedom When log transformed data are used i e under lognormal distribution assumption c is the logarithmic cleanup goal and m and s are the mean and standard deviation of the log transformed data respectively Same as in the sequential t test both yearly averages and original data can be used in the optional analysis The significance of the test is found by comparing the test statistic t with the critical t value under significance level o If t is less than the critical t value and both of them are negative the test is Significant indicating the mean concentration is below the cleanup goal Otherwi
95. needing consideration is the facility wide false positive rate FWFPR also called site wide or experiment wise false positive rate This happens when the monitoring status of a facility or site depends on the probability of obtaining a false positive with any parameter at any well at the facility or site For example in detection monitoring when any of the constituents in any monitoring well indicates an exceedence over the background the site is declared contaminated and must enter into more extensive compliance monitoring Even if the comparison wise false positive rate is very low the FWFPR associated with a large program can be greatly exaggerated and can often lead to a dedaration of contamination For instance assuming the significance level or false positive rate of an individual comparison is o and all comparisons are independent the probability of at least one of n comparisons being significant by chance alone is given by a l o Eq 1 If a 0 05 and n 60 the FWFPR at is 0 95 This indicates the site is almost certain to be declared contaminated when in fact no contamination is present The most effective way to control FWFPR is the combined use of Bonferroni inequality and verification resampling Davis et al 1994 Gibbons 1994 Bonferroni inequality works by inversely specifying the comparison wise significance level a with a fixed FWFPR o a x x Eq 2 n However when n is large a becomes too small and results in
96. not performed dueto insufficient data Cleanup Achieved Figure A 6 2 indicates whether the mean contaminant concentration at a well is below the cleanup goal with statistical significance using the sequential t test Attained indicates the mean concentration is significantly below the cleanup goal and has achieved the Targ amp Levd Attained is always supported by a sufficient power equal to or greater than the expected power Therefore the deanup goal has been attained and the well may be eliminated from the monitoring network Not Attained indicates the mean concentration is higher than the cleanup goal Cont Sampling indicates although the mean concentration is below the cleanup goal it is not statistically significant because 1 the mean concentration does not achieve the Targ amp Levd or 2 the existence of large data variability prevents the test from resulting in significance The latter case corresponds to an inadequate power in the test In the case of Cont Sampling more samples areto be collected for a future re evaluation Results shown are based on yearly averages NO BENZENE ETHYLBENZENE TOLUENE xvLENES TOTAL Well Name Sample Significantly Power Expected 4 Distribution Size Cleanup Goal of Test Sample Size Assumption MWI 1 5 YES 1 000 2 3 Lognormal MVv 12 8 YES 0 953 5 MVv 13 8 NO 0 089 100 MW 14 7 NO 0131 100 View NIV 5 4 NO S S Normal MVv 2 8 NO S S MUS g YES 1 000 3 View MW
97. of Hydrocarbon Plumes from Leaking Petroleum Storage Tank Sites in Texas Bureau of Economic Geology University of Texas at Austin Austin Texas Geologic Circular 97 1 1997 Mace R E and W Choi 1998 Size and Behavior of MTBE Plumes in Texas Conference on Petroleum Hydrocarbons and Organic Chemical in Groundwater NGWA 1998 http www api org ehs mtbelink htm McNab W W D W R J Bear R Ragaini C Tuckfield and C Oldenburg 1999 Historical Case Analysis of Chlorinated Volatile Organic Compound Plumes Lawrence Livermore Laboratory University of California Livermore Ca 1999 http searchpdf adobe com proxies 0 5 69 6 html Rice D W R D Grose J C Michaelsen B P Dooher D H MacQueen SJ Cullen W E Kastenberg L G Everett M A Marino California Leaking Underground Fuel Tank LUFT Historical Case Analysis Environmental Protection Dept Nov 16 1995 Groundwater Services Inc Florida RBCA Planning Study Prepared for Florida Partners in RBCA Implementation Groundwater Services Inc Houston Texas 1997 713 522 6300 or www GSl net com Wiedemeier T H Rifai H S Newell C J and Wilson J W 1999 Natural Attenuation of Fuels and Chlorinated Solvents John Wiley amp Sons New York Version 2 1 A 4 16 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 5 SPATIAL MOMENT ANALYSIS Authors Aziz J J an
98. of contaminants exceeds assimilative capacity of aquifer Stable Insignificant changes Active or passive remediation processes are controlling plume length Shrinking Residual source nearly exhausted and active or passive remediation processes significantly reducing plume mass Exhausted Average plume concentration very low e g 1 ppb and unchanging over time Final stages of source zone dissolution over a relatively small area at a site As shown in the conceptual plume lifecycle figure below see Figure A 4 4 of the nearly 500 sites addressed by this analysis nearly 7596 were found to be in either a stable or shrinking condition based on analyses of both plume length and concentration Plume concentrations were predominantly shrinking 47 to 59 whereas lengths were frequently stable 42 to 61 These results suggest that dissolved hydrocarbon plumes tend to reduce more rapidly in concentration than in length Version 2 1 A 4 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE FIGURE A 4 4 Title 2236 API PIC D REV eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers TEMPORAL TRENDS FOR PLUME LENGTH TOP AND AVERAGE PLUME CONCENTRATION BOTTOM FOR BTEX PLUMES SMALL RELEASES
99. of the groundwater monitoring network should be reassessed for reducing the scope of the system or to stop monitoring altogether Users can compare the projected duration of the sampling at their site to the suggested duration of monitoring evaluated based on the decision matrix below The matrix was developed based on engineering judgment and experience of the authors It is not based on any kind of statistical analysis If their site has groundwater monitoring planned for a significantly longer time period then some reduction in the monitoring duration could be applied subject to local and federal regulations TABLE A 8 4 DURATION DETERMINATION FOR SITES WITH CURRENT SITE TREATMENT Source or Tail Trend Category or PI Trends NT or N A S Trends D or PD Trends Remediate indefinitely or Insufficient Stop treatment if PRG Consider stopping consider increasing Data met Consider stopping treatment if performance or continue treatment if plumehas decreasing trends remediation mechanism sampling been stable for extended have been period occurring for extended period The sampling duration at the site is determined by the Source and Tail Stability results Sites with both decreasing Source and Tail trends are suggested to consider stopping treatment if Version 2 1 A 8 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE decreasing trend
100. positive rate o that needs to be controlled for the site For example a 0 05 0 005 etc If N constituents need to be tested simultaneously use Bonferroni inequality to obtain o the overall significance level for a single constituent as a a N Step 2 Determine the resampling plan that will be used One of Two Samples in Bounds plan exceedence is declared when both the initial sample and the resample exceed the limit One of Three Samples in Bounds plan exceedence is declared when both the initial sample and the two resamples exceed the limit or First or Next Two Samples in Bounds plan exceedence is declared when both the initial sample or any of the two resamples exceed the limit Usually One of Two Samples in Bounds plan and One of Three Samples in Bounds plan are used ASTM 1998 Step 3 Compute the mean x and standard deviation s of then background samples at least 4 for the single constituent Step 4 Determine k the number of monitoring wells that will be sampled for the single constituent Step 5 Consult the tables in Davis and McNichols 1987 or Gibbons 1994 with n a and k to locate the factor K Version 2 1 A 7 13 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 6 Calculatethe prediction limit as follows x Ks Step 7 Make decisions For any downgradient well if its initial sample exceeds the prediction li
101. removing the seasonal means from the data While the methods of calculating and testing the significance of the serial correlation coefficient are given in METHOD 68 procedures for adjusting serial correlation are not provided In this section approaches from EPA guidance 1992b for dealing with seasonal effects and serial correlation are presented These adjustments aid in determining the standard error of the mean and degrees of freedom associated with it when seasonal effects and or serial correlation exist Standard error of the mean is crucial in constructing confidence limits that are widely used in compliance monitoring and corrective action monitoring Recall in METHOD 3 thelower confidence limit is calculated as _ Lea x where I is the standard error of a sample mean and the degrees of freedom Yn vn associated with EE isn 1 vn Now we denote the standard error of the mean as s and thus a lower confidence limit is generally in the form of x t ipf 5 and an upper confidence limit in the form of x 1 5 5 Version 2 1 A 7 36 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE where Df is the degrees of freedom associated with s The following procedures provide for calculating s in the presence of seasonal effects or serial correlation or both assuming that no obvious trend exists in the data In the presence of trends in the data detren
102. results by well and constituent si Monitoring and Remediation Optimization Setem aa ax To navigate the results for individual Spatial Moment Analysis Results constituents dick on the tabs at the top of the The Moment Analysis is used for analyzing a single groundwater constituent multiple constituents Screen are analyzed separately Each tab below shows the statistics for one constituent See manual text or Help for description of moment analysis method Zeroth M oment Estimated M ass The zeroth HE ETHYLBENZENE TOLUENE XYLENES TOTAL plume as coordinates Xc and Yc for each Bak Next gt gt View Report Help sample event and COC The center of mass locations indicate the movement of the center NIS eae moment is a mass estimate for each sample a cetimatea Moment Center of Mass 2nd Moment Spread 2 event and COC The estimated mass is used to Effective Date Mass Kg c Yc Sux sq ft sq E a ues eMI S 1 3 T oz I 6 evaluate the change in total mass of the plume E ue axe e o s oem eem over time ES 5 31 1990 22E 01 2 24 2 885 103 397 pd 9n3n9s0 3 0E 02 50 141 4 260 301 034 z m sen E s ues First Moment Center of Mass The first TEENS aay a eae eRe moment estimates the center of mass of the E Ed oe of mass over time Second Moment Spread of Plume The second moment indicates the spread of the contaminant about the center of mass Sxx and Syy or the distance of con
103. results for the site and COC benzene are displayed Tul et onc 2 Mmi Com etremens m 4 Moins Select View Report to obtain a summary of the results for the analysis Select Next to proceed to the M AROS A nalysis Complete screen Click hereto View Mantloring and Rewedlatios Optimization Sysiaw D MOROS APO CAMPIE At this point the data has been analyzed and design category suggestions are complete Proceed to the M ain M enu and PM Con yh Skud Scene D Oa Prey cO e qo Port Pais Fabia Fiat Air Force Center for Environmental Excellence WA Costes te Step A or Step i gt gt AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE choose to either perform Sampling Optimization Analysis or choose MAROS Output Select Continueto Step 4 or Step 5 to proceed Version 2 1 A 11 41 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Typical Overview Statistics Plume Trend Analysis Results Atthis point in the software the user should assess the overall statistical trend results for the site where they can gain information on the plume stability as well as the distribution of individual well trends Again the goal of the tutorial is to show the user tips and pitfalls when applying MAROS at a typical site The tutorial example has been used only to illustrate the utilities o
104. samples ND Notes 1 Consolidation of data included non detect N D values set to the detection limit 0 001 mg L and the maximum value of duplicate data was applied 2 MW Monitoring Well 3 Source Source Zone Well Tail Tail Zone Well 4 Decreasing D Probably Decreasing PD Stable S No Trend NT Probably Increasing PI and Increasing 1 5 Overall Trend is calculated from a weighted average of the Linear Regression and Mann K endall Trends For further details on this methodology refer to Appendix A 8 Version 2 1 October 2004 A 11 42 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The monitoring well trend results show that 5 out of 5 source wells and all 7 tail wells have a Probably Decreasing Decreasing or Stable trends Both methods gave similar trend estimates for each well MAROS Trend Analysis Well Type yp PD D S I PI 5 of 5 100 0 of 5 0 7 of 7 100 0 of 7 0 Note Decreasing D Probably Decreasing PD Stable S Probably Increasing PI and Increasing I When considering the spatial distribution of the trend results Figure A 11 2 map created in ArcGIS from MAROS results the majority of the decreasing or stable trend results are located near the Tanks source area indicating a decreasing source LEGEND Trend Result Stable Non detect E Decmasing Trend A Stable Trend Ceacton at o
105. seasonal means or medians as e x H where the observation at time i must be in the season j Step 3 Estimate p the observed serial correlation as Step 4 Calculate D the Durbin Watson statistic as N Ye E D i 2 2 e t Mm ii IL Step 5 Consult the Durbin Watson table Neter et al 1996 for test bounds with desired significance level usually 0 05 and the number of observations n Use the first column p 1 1 in this table to find du the upper critical value for the test If D lt du conclude that there is a significant serial correlation and keep 6 the observed serial correlation for future use If D gt du conclude that there is no serial correlation or a serial correlation that is negligible Version 2 1 A 7 30 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Example Step 1 A hypothetical dataset containing quarterly measurements of manganese concentration from a monitoring well during a four years period is presented in Table A 7 15 The time series plot of this dataset is given in Figure A 7 3 from which no obvious trend or seasonal effects can be inferred Therefore we calculate the sample mean of this time series which is 4 97 as the estimate of the overall mean from which the residuals can be obtained Step 2 Calculate residuals by subtracting the overall mean 4 97 from each observation They a
106. significance of spatial variation 5 Choosethe appropriate statistical approaches from the correct category based on considerations from the first four steps Refer to Table A 7 17 This may need to be done on a well by well basis Note If any of the above conditions change during the process of long term monitoring re evaluate the above steps For example if the monitoring requirement in the site changes from compliance monitoring to corrective action monitoring all above steps should be re evaluated Considerations Methods that control FWFPR or FWFNR should be used if FWFPR or FWFNR is critical to making a monitoring decision Sensitivity or power of a statistical approach should always be evaluated and in some cases compared to EPA references When more than one method is eligible the one with highest power at the range of observed variability is preferred as long as it meets the requirement of the false positive rate Table A 7 18 Methods for testing statistical assumptions Objective Method Description Method 6 To test the normality of data The method is a superior alternative to the Shapiro Wilk test Chi Square test and is widely used EPA 1992a 1992b Method 7 To test the homogeneity of The method is a more formal procedure than Levene s test variances between data from Box Plots visual method It has a high power different wells that Bartlett s test for non normal data Method 8 To test if there is si
107. site works best If you would like to view a list I z dire edd of suggested COCs dick on the button E st Recommended COCs This will result in a m summarized list of COC recommendations from the available dataset as well as a criteria ranking system toxicity prevalence or mobility used to make the COC recommendation see below Back Next gt gt Next Takes the user to the Initial D ata Table screen Back Returns the user back to the W ell Coordinates screen Help Provides information on the screen specific input requirements Version 2 1 23 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Risk Leva Assessment accessed from the COC Decision screen allows the user to choose a preliminary remediation goal PRG used to screen representative concentrations from the dataset E Monitoring and Remediation Optimization System MAROS E ni x Choose from the list of generic Preliminary Remediation Goal Risk Level Assessment your own PRG s Click on the appropriate standard to be used in database comparisons for COC recommendations Enter your own modifications to cleanup goals under custom goals in mg L Note User entered cleanup standards will supersede chosen standards PRG recommendations below or you can enter SITE DETAILS lt lt Back
108. statistical methods used to determine the sampling location and sampling frequency MAROS Output Allows the user to view print site specific summary reports and graphs Quit Closes the database program and Access When the database is closed any data that you are currently working on will be erased It is suggested that you Archive the current database if necessary before exiting Help Provides additional information on software operation and screen specific input requirements Version 2 1 13 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D ata M anagement The Data M anagement Menu accessed from the M ain Menu is used to perform database operations such as importing manual data addition and archiving These operations are used to import initial site data as well as additional data into the software Import file formats are discussed in detail in Appendix A 1 amp Monitoring and Remediation Optimization System MAROS 0 x Select One Option Import New Data Manual Data Addition analysis EN ON future MAROS analyses Main Menu B z rr E rr 9 z E s a Data Management Menu Choose between importing ERPIMS files or and Excel file in the standard MAROS format import previously archived data files To add individual Records to the MAROS underlying database for Import MAROS Archive File Import previou
109. the Access database tools to develop customized queries or reports which provide more detailed analysis and presentation of the dataset Version 2 1 90 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDICES A l Data Import File Formats A 2 Statistical Trend Analysis M ethods A 3 Well Redundancy Sufficiency Analysis D elaunay M ethod A 4 Qualitative Evidence Empirical D ata M ethod A 5 Spatial Moment Analysis M ethod A 6 Data Sufficiency Analysis A 7 False Positive N egative M inimization M ethodology A 8 MAROS SiteResults M ethod A 9 Sampling Frequency Analysis M odified CES M ethod A 10 Sample MARO S Reports A 11 MAROS Tutorial Version 2 1 85 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 1 DATA IMPORT FILE FORMATS A few words on data management As a general rule assembling site data for the analysis is the most difficult and time consuming step in the optimization protocol The simple input file required by MAROS is the culmination of years of painstaking sampling and documentation Often the necessary data are not in database format and may need to be entered into electronic format manually Assembling the information with fidelity and clarity is the most important step in reaching an optimum well network solution The following two
110. the center of mass over time relative to the original source E Monitoring and Remediation Optimization System MAROS E Dox C hoose the chemi cal of i nterest from the na MN Pona dropdown boxes at the top of the screen cit chee to rs iet mne ca Mame red I x boa NM ct Choose the graph type i e Log or Linear on Chemical BENZENE 7000 Click Graph on graph to proceed First Moment Trend The First Moment trend of the distance to the center of mass over time is determined by using the Mann Kendall n G ce ea Trend Methodology The First Moment trend MUST EN Confidence in for each COC is determined according to the g EE rules outlined in Appendix A 2 Results for the First Moment Treni WT IE trend include Increasing Probably Increasing Decen nolDI No fend NT Not appleale A due to Puheen da No Trend Stable Probably Decreasing lt Back Nets VewRepon Hei Decreasing or Not Applicable Insufficient Data 2 5E 02 2 0E 02 lance ft 4 5E 02 Source Dist 9 wo gt Z E ka W z e Ed x s wo MK S The Mann Kendall Statistic S measures the trend in the data Positive values indicate an increase in the distance from the source to the center of mass over time whereas negative values indicate a decrease in the distance from the source to the center of mass over time The strength of the trend is proportional to the magnitude of the Mann Kendall St
111. the date range during which one episode in the monitoring program was carried out For example if all wells were sampled between 3 1 2002 and 3 5 2002 the sample event could be defined for all the wells as occurring on 3 3 2002 amp Monitoring and Remediation Optimization System MAROS Sample Events Sample events need to be lumped in order to correctly consolidate groundwater data Choose a sample event name from the drop down box or type in the name you would like to use Then enter a date range for the sample event and an effective date The effective date will be used for plotting purposes as well as later data consolidation To edit sample events choose the sample event name and change the range Sample Event Name Date Range Effective Date OK C Te A Sample Events in Database Effective Date E 10 4 1988 11 7 1989 341990 sanso Sample Date 10 04 1988 Sample Event Sample Event 1 Sample Event 2 Sample Event 3 Sample Event 4 Back Next gt gt 1117988 0301 1990 05 31 1990 SITE DETAILS Steps for use 1 Choose a sample event name from the drop down box or type in the name you would like to use Enter a date range for the sample event e g 10 04 1998 to 10 06 1998 and an effective date e g 10 04 1998 The effective date will be used for plotting purposes and further data consolidation Select OK to update the sample event information
112. the reason for unequal variances in different spatial locations Bartlett s test and Levene s test EPA 1992 EPA 2000 are most widely used for checking the assumption of equal variances Levene s test is less sensitive to departures from normality than Bartlett s test and has power superior to Bartlett s test for non normal data In addition Levene s test has power nearly as great as Bartlett s test for normally distributed data Therefore we introduce Levene s test in this section An exploratory method worth mentioning is the Box Plots through which one can visualize the spread or dispersion within a dataset and compare across groups to see if the assumption of equal variances is reasonable Details for Box Plots can be found in EPA guidance EPA 1992 Procedures Levene s test Step 1 For each of the k groups calculate the group mean as 1 n X x wherei 1 2 K nj number of data in group i Nj ja Step 2 Compute the absolute residuals as z and calculate the group means of these Xi Xj absolute residuals as 1 Nj Zi rdi i j l Also calculate the overall mean of he absolute residuals as k nj Z Y ey wheren n N2 Nk i l j l Step 3 Compute the following sums of squares for the absolute residuals k ni k SS rorar Y Y ES EE Nz SS Groups Y ng Nz and SS ERROR SS rora 7 SS Grours id j l i l Step 4 Compute the F statistic as SS Groups Kk 1 f SS
113. theory The idea behind using empirical data as a line of evidence is summarized by one of the conclusions from an extensive chlorinated solvent plume study performed by the Lawrence Livermore National Laboratory Statistical methods such as general linear models and comparison of probability distributions of plume length indices are useful to quantify expected relationships between plume length and site and CVOC variables within a population of CVOC plumes In addition they provide population statistics that may be used to bound the uncertainty inherent in expected plume behaviors McNab et al 1999 The empirical data for groundwater plumes has been derived from a series of multiple site statistical studies sometimes called plume a thon studies These include plumea thon studies of e BTEX plumes in California Texas Florida and nationwide four studies e MTBE plumesin California and Texas two studies e Chlorinated solvent plumes nationwide two studies In the MAROS system the user has the option but not the requirement to use the body of empirical data on plume behavior to help design and optimize a monitoring system Key Points Caveats Key points regarding the use empirical data as a secondary line of evidence are summarized below e Useof empirical data as a line of evidence is optional to the user e The empirical data if used should be considered secondary evidence and not weighted as much asthe primary evi
114. to the Centerline Regression ia H Projected Concentrations screen MW 15 EI E e E Use All Wells Selects all wells for analysis MW 4 el Mes z Help Provides additional information on software e WA 4 operation and screen specific input requirements z MW 8 vw x a Note if a well is selected deselected here it will be d c s e ssected deselected for all COCs Version 2 1 83 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Risk Based Power Analysis Results This screen accessed from the Centerline Regression Projected Concentrations screen by clicking Analysis is used to display the results for risk based site cleanup evaluations grouped by COC amp 3 Monitoring and Remediation Optimization System MAROS 3 xj Sample Size The nu mber of projected Risk Based Power Analysis Results concentrations i e wells available for sped petal ie iln alei leyendo aliae cta analysis in the current sampling event site cleanup goal based on the projected concentrations is met at the compliance boundary Data are assumed to be normally or lognormally distributed and results under both assumptions are given for comparison Cleanup Achieved Indicates whether the cleanup goal is achieved for the entire BENZENE ETHYLBENZENE TOLUENE xYLENES TOTAL correct conclusion can be made when the average projected
115. trends in the analysis when there are several non detect samples TREND ANALYSIS Trace TR Choose the number value you would like to represent a Trace result in the data The TR flag is equivalent to the J flag used by most labs to indicate a result that is reported but is below the method detection limit Back Returns the user to the D ata Reduction Part 2 of 2 screen Next Takes the user to the R educed D ata T able Screen Help Provides information on the screen specific input requirements Version 2 1 30 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D ata Reduction Reduced D ata Table accessed from the D ata Reduction Part 2 of 2 screen allows the user to view the reduced data table with the COCs chosen as well as the data consolidation performed This table is not available for editing E Monitoring and Remediation Optimization System arasy a Back Returns the user to the Data Reduction Reduced Data Table Part 2 of 2 screen Below is the data table with all specified data reduction operations performed Next Takes the user to the Reduced Data Plot Source Result Number il Well Hame Tail Dato coc mat Flag screen IMWw 14 s 10 3991 TOLUENE 5 5E 01 IMUN 14 s 1219998 XYLENES TOTAL 2 0E 03 IMW 13 s 93 990 ETHYLBENZENE 5 0E 04 ND s 2 sans s suns pve Er Help Provides information on the screen
116. values are classified into four levels S Small 0 3 M Moderate 0 3 0 6 L Large 0 6 0 9 and E Extremely large 0 9 1 0 A colored label around the center centroid of each triangle is used to indicate the SF level at a potential area Back to Access Switches to the MAROS interface in Microsoft A ccess The user loads and enters this module from the Wd Sufficiency Analysis New Locations screen by dicking the Analysis button The data will be transferred from Microsoft Access and the Version 2 1 59 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE analysis is completed once the xIsN ewLocation interface shows up The user can proceed with the following steps 1 Search the potential areas for new sampling locations with L large or E extremely large labels New wells could be placed inside these regions e g at the centroid of a triangle 2 Click Back to Access to return to Microsoft Access The xlsN ewLocation worksheet will remain open until the user closes it The user can save the file with another name to prevent the current results from being overridden by a new analysis 3 To perform another analysis choose a COC from the Wal Sufficiency Analysis New Locations screen and then click the Analysis button to enter the xIsN ewLocation module then go to step 1 WARNING Do not change the name of worksheet xlsN ewLocation or move it to other fold
117. varieties of prediction limits not only control the facility wide false positive rates at specified level but also minimize false negative rates for a certain number of n and k For a fixed number of k increasing n will increase the power of the test For a fixed number of n decreasing k will increase the power of the test Example Step 1 A set of hypothetical data representing a single constituent is presented in Table A 7 8 for demonstration As in most cases of groundwater quality data the eight independent background measurements are transformed by taking their natural logarithm Then the background mean and standard deviation of the transformed data are computed as 1 029 and 0 672 respectively Version 2 1 A 7 14 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 2 The facility wide false positive rate a of 0 05 is chosen and the One of Two Samples in Bounds plan exceedence is declared when both the initial sample and the resample exceed the limit is specified Since only one constituent is considered a a Step 3 Assume there are ten monitoring wells k 10 for future comparison Step 4 Consulting the tables in Davis and M cN ichols 1987 or Gibbons 1994 with n 8 a 0 05 and k 10 and wefind K 2 03 eg in Gibbons 1994 page 23 Table 1 5 Step 5 Calculate the prediction limit as x Ks 1 029 2 03 x 0 672 2 393 Step 6 M
118. well in coarse granular or highly slowly to sample effectively bedrock well if consistent amounts of water fractured medium with study objectives Redesign new well 2 Well silts up Select screen opening size and sand pack Water is murky or bottom of Redevelop well after installation gradation to be compatible with geologic well feels mushy when periodically materials to be screened Add a sump sounded below the well screen 3 Sand pack Specify a well sorted poorly graded Wel recharges much more Redevelop well becomes clogged coarse grained w ashed quartz sand or gravel consistent with the aquifer material slowly than expected periodically Redesign new well 4 Well seals leak Design seals to be compatible with projected use of well and site hydrogeology and geochemistry Monitor installation of seals closely by repeatedly measuring the depth to the seal Water elevation and quality on either side of the seal are more similar than expected Abandon leaking wells to prevent inter aquifer leakage and replace well 5 Well materials are degraded by contaminants or fail structurally Specify stainless steel for areas of high organic contamination and PVC or Teflon in areas of extreme pH Specify appropriate material strength based on expected loads Screen or overdrill highly fractured bedrock wells Obstructions found in the well Aquifer materials that are larger than screen slots
119. wells eliminated area ratio and concentration ratio WARNING Before dicking the Back to A ccess button the user should have performed the optimization by using the W ell Locations chart sheet see instructions on the next page If not the original set of data will be returned Do not make changes in this sheet Furthermore it is recommended that theuser operate in the W ell Locations chart sheet Version 2 1 64 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The Well Locations chart sheet is shown below A plot area is located in the center where the sampling locations are plotted in the EAST NORTH coordinate system or relative coordinates system The legend is in the upper right side The middle right side contains the command buttons used to control the optimization process d xlsD elaunay m Removable 150 0 Irremovable INIT Apply Reset All GlearResum Terminate TK TEN Well Locations Datasheet 7 INIT Apply Initializes the program in order to begin an analysis This is a starting point Reset All Allows all potential locations to be selected This is very helpful when you have eliminated some locations and then want to recover them Clear Resume To dear resume all the lines drawn on the plot area It is only a switch for graphic output Data will not be altered Terminate Clears memory and stop
120. wells is less than 3 no regression is performed and all values are set to 0 lt lt Back View Report Next gt gt SAMPLING OPTIMIZATION After inputting the above information click the gt gt Analysis button to proceed The Plume Centerline Regression Results screen will appear The regression coefficient ZEEE x from the exponenti al reg ression of Centerline Regression Projected Concentrations center i ne concentrati ons vs d i stance Concentrations from each sampling location are projected to the compliance boundary at or upgraidient to the downgradient receptor using the regression results obtained in the previous step The projected a below fe h si li lassified by COC The d ill down centerline and the confidence erea e yy Neue E a A N e ger pe i f mage want to use in the risk based power analysis in the next step associated with the coefficient are BEBE displayed for each sample event The x Well Projected Below Usein 4 regression analysis is performed only e m for events in which at least three z EL Lp me z centerline wells were sampled Note z IEEE 4 x the regression coefficients are all ES 3s 4 negative indicating the concentration a a US a Version 2 1 A 11 5i z lt lt Back View Report Analysis gt gt Help October 2004 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE is decaying along the plume centerline
121. 0 PD Mw 3 T 69 02 13E 03 105 99 9 D zl EI Note Increasing 1 Probably Increasing FI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Source T ail S T COV Coefficient of Variation TREND ANALYSIS Slope The slope of the least square fit through the given data indicates the trend in the data Positive values indicate an increase in constituent concentrations over time whereas negative values indicate a decrease in constituent concentrations over time Confidence in Trend The Confidence in Trend is the statistical confidence that the constituent concentration is increasing slope gt 0 or decreasing slope lt b Concentration Trend The Concentration Trend for each well is determined according to the rules outlined in Appendix A 2 Results for the trend include Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or N ot A pplicable Insufficient Data The information displayed in this screen can also be viewed in report form Linear Regression Statistics Report from the M AROS Output Screen see A ppendix A 10 for an example report Back Returns the user to the M ann Kendall Plot Screen Next Takes the user to the Linear Regression Plot Screen Help Provides information on the screen specific input requirements Version 2 1 34 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATIO
122. 0 0 000 x M14 102 0 20 0 0 000 0 000 0 000 sampl l ng events Um MAAS 190 0 125 0 0 446 0 446 0 446 3 we 2 Max SF Displays the maximum SF E wa so ao 055 085 085 value of a location obtained from one of MW 5 40 70 0 0 681 0 681 0 681 zi the sampl i ng events F4 MM M E a Back Returns the user to the Access lt lt Back Optimize by COC gt gt n x o Ss E M odule Potential Locations Setup screen Optimize by COC Performs optimization for each COC by eliminating redundant sampling locations in each COC and then proceeds to the A ccess M odule Results by COC screen Help Provides additional information on software operation and screen specific input requirements N ote the Slope Factor in MAROS is a parameter indicating the relative importance of a location in the monitoring network and is not related to toxicological values for a particular COC i e carcinogenic risk Version 2 1 55 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Access M odule Results by COC This screen accessed from the A ccess M odule Slope Factor V alues screen by clicking Optimize by COC is used to display the sampling location optimization results for each COC Redundant locations that are eliminated are marked The remaining locations are unmarked and are recommended for the next round of sampling Elimination of a location from a COC only means
123. 00 1177 0 4 25E 02 3 522E 22 Yes Yes Sample Event 2 11 17 1989 MW 12 4 600E 02 1090 0 4 25E 02 3 445E 22 Yes Yes Sample Event 2 11 17 1989 MW 13 2 600E 02 1125 0 4 25E 02 4 397E 23 Yes Yes Sample Event 2 11 17 1989 MW 14 2 600E 02 1088 0 4 25E 02 2 120E 22 Yes Yes Sample Event 2 11 17 1989 MW 15 1 000E 03 1000 0 4 25E 02 3 437E 22 Yes Yes Sample Event 2 11 17 1989 MW 2 2 700E 01 1192 0 4 25E 02 2 645E 23 Yes Yes Sample Event 2 11 17 1989 MW 3 1 800E 01 1155 0 4 25E 02 8 501E 23 Yes Yes Sample Event 2 11 17 1989 MW 4 1 200E 01 1135 0 4 25E 02 1 326E 22 Yes Yes Sample Event 2 11 17 1989 MW 5 1 700E 00 1194 0 4 25E 02 1 530E 22 Yes Yes Sample Event 2 11 17 1989 MW 6 1 000E 03 1267 0 4 25E 02 4 039E 27 Yes Yes Sample Event 2 11 17 1989 MW 7 1 000E 03 1277 0 4 25E 02 2 640E 27 Yes Yes Sample Event 2 11 17 1989 MW 8 3 000E 03 1245 0 4 25E 02 3 087E 26 Yes Yes Sample Event 3 3 1 1990 MW 1 2 200E 00 1177 0 3 14E 02 1 947E 16 Yes Yes Sample Event 3 3 1 1990 MW 12 1 400E 01 1090 0 3 14E 02 1 904E 16 Yes Yes Sample Event 3 3 1 1990 MW 13 4 900E 02 1125 0 3 14E 02 2 220E 17 Yes Yes Sample Event 3 3 1 1990 MW 14 3 400E 02 1088 0 3 14E 02 4 924E 17 Yes Yes Sample Event 3 3 1 1990 MW 15 1 000E 03 1000 0 3 14E 02 2 297E 17 Yes Yes Sample Event 3 3 1 1990 MW 2 1 000E 03 1192 0 3 14E 02 5 526E 20 Yes Yes Sample Event 3 3 1 1990 MW 3 1 000E 03 1155 0 3 14E 02 1 766E 19 Yes Yes Sample Event 3 3 1 1990 MW 4 2 200E 01 1135 0 3 14E 02 7 282E 17 Yes Yes M
124. 04 A 11 9 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Step 2 Site Details allows initial definition of site specific data induding choosing the Source and Tail wells sample events and providing site specific Constituents of Concern COC s STEP I MAIN MENU Select Site Details from the M ain M enu The Site Information screen will be displayed E pans tes gp nd Rerpesl tae D p ro ater fre WATS Dato Managemarc Meer mate op na ami FRE es atten cree te etter Sis etai 1 w P at de QA t OI ets e cn Piums Anateiis Fer urw Den Gaiansta Cosmia fiina ariin Anra a Ci trm Ds rmn Ppt teed oF Sampling Optimization ee UI Da ee er paa wert in amuebo lhe sunan netin dui mary naars smpi 1 MAROSOput arb dE Fh mc mmn Lett STEP 2 SITE INFORMATION Site Information is the first step in defining the site type as well as parameters unique to the site Site details were outlined at the start of the tutorial t Merr petog atc irre dicun Lyereteatto Systems WANK iJ The following information will need to be Site Information entered on the Site Information screen Fuit adn uev vache rh sored ira 2C pies aam aen saa Under General enter e Location Service Station ewe tbe tuse esta IT v Gens met M Enter details State Texas U
125. 05 Expected Power 0 8 MW 1 6 3 90E 01 5 44E 01 NO S E S E NO S E S E MW 12 6 7 61E 03 1 01E 02 NO S E S E NO S E S E MW 13 6 6 57E 03 1 07E 02 NO S E S E NO S E S E MW 14 6 1 00E 03 1 50E 11 YES 1 000 lt 3 YES 1 000 3 MW 15 6 1 00E 03 1 50E 11 YES 1 000 3 YES 1 000 3 MW 2 6 2 89E 03 3 53E 03 NO 0 391 19 NO 0 418 17 MW 3 6 3 53E 02 4 91E 02 NO S E S E NO S E S E MW 4 6 1 87E 02 1 20E 02 NO S E S E NO S E S E MW 5 6 1 11E 00 8 68E 01 NO S E S E NO S E S E MW 6 6 1 00E 03 1 50E 11 YES 1 000 3 YES 1 000 3 MW 7 6 1 00E 03 1 50E 11 YES 1 000 3 YES 1 000 3 MW 8 6 1 00E 03 1 50E 11 YES 1 000 3 YES 1 000 3 Note N C refers to not conducted because of insufficient data N 4 S E indicates the sample mean significantly exceeds the cleanup level and thus no analysis is conducted Sample Size is the number of concentration data in a sampling location that are used in the power analysis Expected Sample Size is the number of concentration data needed to reach the Expected Power under current sample variability The Target Level is the expected mean concentration in wells after cleanup attainment it is only used in individual well celanup status evaluation The Student s t test on mean difference is used in this analysis Refer to Appendix A 6 of MAROS Manual for details MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 1 Regression of Plume Centerline Concentrations Project E
126. 155 0 5 06E 02 6 567E 27 Yes Yes Sample Event 6 4 3 1991 MW 4 1 600E 02 1135 0 5 06E 02 1 925E 27 Yes Yes MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 2 of Project Example U ser Name Meng Location Service Station State Texas Observed Regression Projected Below Sampling Effective Well Concentration Distance Down Coefficient Concentration Detection Used in Event Date mg L Centerline ft 1 ft mg L Limit Analysis BENZENE Sample Event 6 4 3 1991 MW 5 9 100E 01 1194 0 5 06E 02 5 547E 27 Yes Yes Sample Event 6 4 3 1991 MW 6 1 000E 03 1267 0 5 06E 02 1 521E 31 Yes Yes Sample Event 6 4 3 1991 MW 7 1 000E 03 1277 0 5 06E 02 9 177E 32 Yes Yes Sample Event 6 4 3 1991 MW 8 1 000E 03 1245 0 5 06E 02 4 627E 31 Yes Yes Sample Event 7 7 10 1991 MW 1 1 700E 00 1177 0 4 57E 02 7 852E 24 Yes Yes Sample Event 7 7 10 1991 MW 12 3 000E 02 1090 0 4 57E 02 7 854E 24 Yes Yes Sample Event 7 7 10 1991 MW 13 2 900E 02 1125 0 4 57E 02 1 438E 24 Yes Yes Sample Event 7 7 10 1991 MW 14 1 000E 03 1088 0 4 57E 02 2 686E 25 Yes Yes Sample Event 7 7 10 1991 MW 15 1 000E 03 1000 0 4 57E 02 1 492E 23 Yes Yes Sample Event 7 7 10 1991 MW 2 1 000E 03 1192 0 4 57E 02 2 329E 27 Yes Yes Sample Event 7 7 10 1991 MW 3 1 100E 01 1155 0 4 57E 02 1 387E 24 Yes Yes Sample Event 7 7 10 1991 MW 4 1 400E 02 1135 0 4 57E 02 4 399E 25 Yes Yes Sample Event 7 7 10 1991 MW 5 2 500E 00 1194 0 4 57E 02 5 314E 24 Yes Yes Sample Event 7 7 10 1991 MW
127. 30 Saturated Thickness Uniform 12 ft Mann Kendall Trend test performed on all sample events for each constituent Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events Note The Sigma XX and Sigma YY components are estimated using the given field coordinate system and then rotated to align with the estimated groundwater flow direction Moments are not calculated for sample events with less than 6 wells MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 2 of 2 MAROS Zeroth Moment Analysis Project Tutorial User Name Charles Newell Location Service Station State Texas coc BENZENE Change in Dissolved Mass Over Time Date Porosity 0 30 of v s A os gw a e Pi amp Pa Saturated Thickness wc UU TUS hes AMEMEMEI EN 9 12E01 E pom o y U UyUyUy Confidence in a 1 0E 01 Trend 3 8 amp 0E 02 100 09 4 0E 02 Coefficient of Variation 0 0E 00 Zeroth Moment Trend D Data Table Estimated Effective Date Constituent Mass Kg Number of Wells 10 4 1988 BENZENE 1 7E 01 12 11 17 1989 BENZENE 1 3E 01 12 3 1 1990 BENZENE 1 1E 01 12 5 31 1990 BENZENE 6 8E 02 12 9 13 1990 BENZENE 6 6E 02 12 4 3 1991 BENZENE 5 1E 02 12 7 10 1991 BENZENE 5 9E 02 12 10 3 1991 BENZENE 8 0E 02 12 5 2 1992 BENZENE 2 9E 02 12 1 11 1994 BENZENE 2 8E 02 12 5 28 1996 BENZENE 3 2E 02 12
128. 35 EPA FLAGS EPA Data Qualifier Codes Version 2 1 A 2 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 1 6 EXAMPLE MAROS CONSTITUENT NAME CONVENTION Abreviation MAROS CAS or ERPIMS Constituent Constituent Number Constituent Synonym Code Name Type BTEX AND MTBE 71 43 2 Benzene B BZ BENZENE ORG 100 41 4 Ethylbenzene E EBZ ETHYLBENZENE ORG 108 88 3 Toluene 1r BZME TOLUENE ORG 1330 20 7 Xylene mixed isomers X XYLENES XYLENES TOTAL ORG 108 30 3 Xylene m X XYLEN ES1213 XYLENES o amp m ORG 95 47 6 Xylene o X XYLEN ES1213 XYLENES o amp m ORG 1634 04 4 Methyl t Butyl Ether MTBE TBUTMEE tert BUTYL METHYL ETHER ORG CHLORINATED COMPOUNDS 75 27 4 Bromodichloromethane BDCME BROMODICHLOROMETHANE ORG 56 23 5 Carbon tetrachloride CT CTCL CARBON TETRACHLORIDE ORG 108 90 7 Chlorobenzene CLBZ CHLOROBEN ZENE ORG 75 00 3 Chloroethane CLEA CHLOROETHANE ORG Trichlorometh 67 66 3 Chloroform ane TCLME CHLOROFORM ORG Methyl 74 87 3 Chloromethane Chloride CLME CHLOROMETHANE ORG 95 57 8 Chlorophenol 2 CLPH2 2 CHLOROPHENOL ORG 124 48 1 Dibromochloromethane DBCME DIBROMOCHLOROMETHANE ORG Dichlorobenzene 1 2 95 50 1 o DCBZ12 12 DICHLOROBENZENE ORG Dichlorobenzene 1 4 106 46 7 p DCBZ14 14 DICHLOROBENZENE ORG Dichlorodifluoromethan 75 71 8 e FC12 DICHLORODIFLUOROMETHANE ORG 75 34 3 Dichloroethane 1 1 11DCA DCA11 11 DICHLOROETHANE ORG 107 06 2 Dichlo
129. 4 0 0E 00 0 0E 00 100 0 2 xj Note Increasing I Probably Increasing Pl Stable 5 Probably Decreasing PD Decreasing D No Trend NT Not Applicable N 4 Source T ail S T COV Coefficient of Variation FIGURE A 2 2 LINEAR REGRESSION ANALYSIS RESULTS e The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value Values less than or near 1 00 indicate that the data form a relatively close group about the mean value Values larger than 1 00 indicate that the data show a greater degree of scatter about the mean e The Log Slope Ln Slope measures the trend in the data Positive values indicate an increase in constituent concentrations over time whereas negative values indicate a decrease in constituent concentrations over time e The Confidence in Trend is the statistical confidence that the constituent concentration is increasing In slope gt 0 or decreasing In slope lt 0 Version 2 1 A 2 9 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE e The Concentration Trend for each well is determined according to the following rules where COV isthe coefficient of variation TABLE A 2 2 MAROS LINEAR REGRESSION ANALYSIS DECISION MATRIX Confidence in Positi Ln Slope Trend H NY Negative COV 1 Stable SR iilud COV 1 No Trend 9096 9596 Probably Increasing Probably
130. 700 gu Analysis Complete Excel M odule screen z MNA 550 370 m Lr Z E i a 2 a View Report Generates a report with 9 Eliminated whether or not the well is abandoned from the monitoring the all in one sampl Ing location network as a redundant well optimization results The user can then a Help Provides additional information on software operation and screen specific input requirements Sampling Location Analysis Complete Excel M odule This screen accessed from the Excel M odule All in one Results screen by clicking Next is a message screen indicating that the sampling location determination by the Excel module has been completed and the user can proceed to other analyses t Merrtarwvg and Rerslatian Dytiesission ByM DAANAN A Back Returns the user to the Excel M odule All in z Sampling Location Analysis Complete one Results screen The user can go back to re run the E Excel Module analysis by changing parameters or by selecting a E POTEST STRE PEN EINEN E UNIUS different series of sampling vents o1 xanwole fone daed Dy seii Yo np OA Digo Vo over goor OF Bepi rirunatan Yusa din gu tect is chase settet sogis A RO NIIT Go to Sampling Optimization Returns the user to rios es C the Sampling O ptimization screen a z i Beck py ps e Version 2 1 68 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM S
131. 8 Benzo a Pyrene BZAP BEN ZO a PY RENE ORG 218 01 9 Chrysene CHRYSENE CHRYSENE ORG 53 70 3 Dibenzo a h Anthracene DBAHA DIBENZ ah ANTHRACENE ORG 206 44 0 Fluoranthene FLA FLUORANTHENE ORG 86 73 7 Fluorene FL FLUORENE ORG 193 39 5 Indeno 1 2 3 c d Pyrene INP123 INDENO 1 2 3 c d PY RENE ORG 91 20 3 Naphthalene NAPH NAPHTHALENE ORG 85 01 8 Phenanthrene PHAN PHENANTH RENE ORG 129 00 0 Pyrene PYR PYRENE ORG OTHER COMPOUNDS 67 64 1 Acetone ACE ACETONE ORG 65 85 0 Benzoic acid BZACID BENZOIC ACID ORG 71 36 3 Butanol n BTOH n BUTANOL ORG 75 15 0 Carbon disulfide CDS CARBON DISULFIDE ORG 107 21 1 Ethylene glycol ETEGLY ETHYLENE GLYCOL ORG 110 54 3 Hexane n C6N n HEXANE ORG 67 56 1 Methanol MEOH METHANOL ORG METHYL ETHYL KETONE 2 78 93 3 Methyl ethyl ketone MEK MEK BUTANONE ORG 108 95 2 Phenol PHENOL PHENOL ORG Version 2 1 A 2 8 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 2 STATISTICAL TREND ANALYSIS METHODS Authors Newell C J and Aziz J J Groundwater Services Inc This appendix details the data evaluation and remedy selection procedures employed by the Monitoring and Remediation Optimization System MAROS Software The procedures outlined below were developed to assess appropriate response measures for affected groundwater plumes based on scientifically sound quantitative analyses of current and historical site groundwater conditions In
132. 997 MW 14 1 000E 03 1088 0 4 44E 02 1 090E 24 Yes Yes Sample Event 12 6 27 1997 MW 15 1 000E 03 1000 0 4 44E 02 5 405E 23 Yes Yes Sample Event 12 6 27 1997 MW 2 1 000E 03 1192 0 4 44E 02 1 080E 26 Yes Yes Sample Event 12 6 27 1997 MW 3 3 000E 03 1155 0 4 44E 02 1 673E 25 Yes Yes Sample Event 12 6 27 1997 MW 4 2 800E 02 1135 0 4 44E 02 3 792E 24 Yes Yes MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 4 of Project Example User Name Meng Location Service Station State Texas Observed Regression Projected Below Sampling Effective Concentration Distance Down Coefficient Concentration Detection Used in Event Date Well mg L Centerline ft 1 ft mg L Limit Analysis BENZENE Sample Event 12 6 27 1997 MW 5 6 230E 01 1194 0 4 44E 02 6 158E 24 Yes Yes Sample Event 12 6 27 1997 MW 6 1 000E 03 1267 0 4 44E 02 3 876E 28 Yes Yes Sample Event 12 6 27 1997 MW 7 1 000E 03 1277 0 4 44E 02 2 487E 28 Yes Yes Sample Event 12 6 27 1997 MW 8 1 000E 03 1245 0 4 44E 02 1 029E 27 Yes Yes Sample Event 13 12 10 1997 MW 1 3 600E 02 1177 0 4 10E 02 4 013E 23 Yes Yes Sample Event 13 12 10 1997 MW 12 1 000E 03 1090 0 4 10E 02 3 944E 23 Yes Yes Sample Event 13 12 10 1997 MW 13 5 200E 04 1125 0 4 10E 02 4 885E 24 Yes Yes Sample Event 13 12 10 1997 MW 14 1 000E 03 1088 0 4 10E 02 4 280E 23 Yes Yes Sample Event 13 12 10 1997 MW 15 1 000E 03 1000 0 4 10E 02 1 578E 21 Yes Yes Sample Event 13 12 10 1997 MW 2 1 000E 03 1192 0 4 10E 02 6 027E
133. A NIA MA 8 T S D NIA N A IMw 7 T s D NA NIA ves s s NIA NIA Mw 5 T D D N A NIA MA 4 T D D NIA NA v Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Source Tail S T TREND ANALYSIS Back Next View Report To navigate the individual constituent results dick on thetabs at the top of the screen The information displayed in this screen can also be viewed in report form Lines of Evidence Summary Report from the M AROS Output Screen see Appendix A 10 for an example report Back Returns the user to the Plume Analysis M enu Next Takes the user to the Statistical and Plume Information Summary W eighting Screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 47 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MAROS Analysis Statistical and Plume Information Weighting Statistical and Plume Information Summary Weighting accessed from the Statistical and Plume Information Summary by Wal screen allows the user to weight the individual lines of evidence i e Mann Kendall Trend Analysis Linear Regression Analysis Modeling and Empirical results Monitoring and Remediation Optimization System MAROS Plume Information Consolidation Step 1 Statistical and Plume informat
134. AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well Redundancy Analysis Excel M odule Well Redundancy Analysis Excel M odule accessed from the Wal Redundancy Analysis Delaunay M ethod screen by clicking Excel M odule is a control screen for applying the Delaunay method in a stand alone Microsoft Excel module It is used for 1 setting up the properties of potential locations 2 proceeding to the Excel module for optimization and 3 displaying the results transferred back from the Excd Module The stand alone Excel module xlsDelaunay2K is explained below The data table is similar to that in the Well Redundancy Analysis Excel Module Screen A ccess M odule Potential Locations Setup 2 Meratenng enti Voresesdialinss Ciptarendtion Syster Gillie Perfor nelradextccy atys Dry siman iocsoon fh as irist the graph ns vere utt mi tr aad iim thee a poa al cd oen of these pcm oul eun ex abes o one angle event Oy COC Seleds the COC you want to n 3 ETE Beer analyze from the dropdown list Lacs EXCuar WSCoore Seria Hameuwis Eieisatnd E Analysis Runs the Excel module The x m 4 m xIsD daunay2K worksheet will be opened w I s 4 J and becomes the current screen The Jo a V eoo analysis is performed for the currently wes 3 a x B selected COC and for one sampling zs CE T Fi a d T event only 2 is a 4 4 zi a Reset Sets all the sampling locations in a 3 c ack View
135. AROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of Project Example U ser Name Meng Location Service Station State Texas Observed Regression Projected Below Sampling Effective Well Concentration Distance Down Coefficient Concentration Detection Used in Event Date mg L Centerline ft 1 ft mg L Limit Analysis BENZENE Sample Event 3 3 1 1990 MW 5 1 200E 00 1194 0 3 14E 02 6 227E 17 Yes Yes Sample Event 3 3 1 1990 MW 6 1 000E 03 1267 0 3 14E 02 5 242E 21 Yes Yes Sample Event 3 3 1 1990 MW 7 1 000E 03 1277 0 3 14E 02 3 829E 21 Yes Yes Sample Event 3 3 1 1990 MW 8 1 000E 03 1245 0 3 14E 02 1 046E 20 Yes Yes Sample Event 4 5 31 1990 MW 1 2 300E 00 1177 0 2 77E 02 1 655E 14 Yes Yes Sample Event 4 5 31 1990 MW 12 1 900E 01 1090 0 2 77E 02 1 517E 14 Yes Yes Sample Event 4 5 31 1990 MW 13 5 200E 02 1125 0 2 77E 02 1 577E 15 Yes Yes Sample Event 4 5 31 1990 MW 14 4 400E 02 1088 0 2 77E 02 3 714E 15 Yes Yes Sample Event 4 5 31 1990 MW 15 1 000E 03 1000 0 2 77E 02 9 634E 16 Yes Yes Sample Event 4 5 31 1990 MW 2 1 000E 03 1192 0 2 77E 02 4 750E 18 Yes Yes Sample Event 4 5 31 1990 MW 3 1 400E 01 1155 0 2 77E 02 1 851E 15 Yes Yes Sample Event 4 5 31 1990 MW 4 1 600E 02 1135 0 2 77E 02 3 679E 16 Yes Yes Sample Event 4 5 31 1990 MW 5 1 300E 00 1194 0 2 77E 02 5 843E 15 Yes Yes Sample Event 4 5 31 1990 MW 6 1 000E 03 1267 0 2 77E 02 5 963E 19 Yes Yes Sample Event 4 5 31 1990 MW 7 1 000E 03 1277 0 2 77E 02 4 522
136. Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE RESULTS AND INTERPRETATION OF RESULTS The results of risk based site cleanup evaluations are presented in the Risk Based Power Analysis Results screen introduced in the MAROS Detailed Screens Description chapter The site cleanup status power and expected sample size for each sampling event with at least six projected concentrations are calculated under both normal and lognormal assumptions Figure A 6 6 When a sampling event has less than six projected concentrations insufficient data or the mean projected concentration is higher than the cleanup goal N C or S E respectively are displayed in result fields indicating the analysis is not conducted Cleanup Achieved presents the risk based site cleanup status at the compliance boundary at the time when the sampling event was taken The result indicates whether the mean projected concentration at the compliance boundary is below the cleanup level with statistical significance Results could be Attained cleanup goal achieved Not Attained cleanup goal not achieved or NC not conducted due to insufficient data The results may be different over time i e over sampling events selected The results as a function of time can be used to evaluate the effectiveness of site remedial actions Power of Test is the probability that the site is confirmed to be clean when the pr
137. C recommendations from the available dataset The choices at the bottom of the screen allow a view of the process used to make the COC recommendation below Enter up to 5 COCs for Mobility based COCs LEAD LEAD 1 12 TETRACHLOROETHA BENZENE 13 DICHLOROBENZENE TOLUENE TOLUENE BARIUM PERCHLORATE COPPER COPPER For more information Toxicity Prevalence Region 9 PRG criteria used User specified cleanup Back BENZENE 1 1 1 2 TETRACHLOROETHANE PERCHLORATE BARIUM 12 DICHLOROBENZENE PERCHLORATE BENZENE TOLUENE 1 1 12 TETRACHLOROETHAN 1 2 DICHLOROBENZENE LEAD BENZENE z BARIUM potus ETHYLBENZENE COCs for site goals included in PRG criteria lolx Enter up to 5 COCs for the site in the boxes to the left If you would like a detailed view of the process used to make the COC recommendation click on Toxicity Prevalence or Mobility at the left side of the screen The information displayed in this screen can also be viewed in report form COC Assessment Report from the MAROS Output Screen see Appendix A 8 for an example report Back Returns the user to the Risk Leva Assessment screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 24 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails t Monitoring and Remediation Optimization System MAROS
138. Coefficient of Variation COV and the M ann K endall statistic S for each well with at least four sampling events see Figure A 2 1 A Concentration Trend and Confidence in Trend are reported for each well with at least four sampling events If there is insufficient data for the well trend analysis N A Not Applicable will be displayed in the Concentration Trend column BENZENE ETHYLBENZENE 1 2 DICHLOROBENZENE TOLUENE XYLENES TOTAL Statistical Analysis Results Last column is the result for the trend MK S Confidence in Trend Concentration Trend MS S 0 985 11 70 50 S MAC S 0 249 7 62 60 S MAC 4 S 1606 50 99 90 D M13 S 14106 53 99 80 D MAI 12 S 1 581 68 100 00 D MAC S 1 701 15 98 50 D MB T 0000 0 47 8096 S MVS T 0 851 31 99 80 D Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Source T ail S T COV Coefficient of Variation MK S Mann Kendall Statistic FIGURE A 2 1 MANN KENDALL ANALYSIS RESULTS e The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value Values less than or near 1 00 indicate that the data form a relatively dose group about the mean value Values larger than 1 00 indicate that the data show a greater degree of scatter about the mean e TheMann Kendall statistic MK S measures the trend in the data Pos
139. E 19 Yes Yes Sample Event 4 5 31 1990 MW 8 1 000E 03 1245 0 2 77E 02 1 096E 18 Yes Yes Sample Event 5 9 13 1990 MW 1 1 500E 00 1177 0 4 45E 02 2 707E 23 Yes Yes Sample Event 5 9 13 1990 MW 12 3 000E 02 1090 0 4 45E 02 2 598E 23 Yes Yes Sample Event 5 9 13 1990 MW 13 1 500E 02 1125 0 4 45E 02 2 737E 24 Yes Yes Sample Event 5 9 13 1990 MW 14 1 400E 02 1088 0 4 45E 02 1 325E 23 Yes Yes Sample Event 5 9 13 1990 MW 15 1 000E 03 1000 0 4 45E 02 4 749E 23 Yes Yes Sample Event 5 9 13 1990 MW 2 1 000E 03 1192 0 4 45E 02 9 258E 27 Yes Yes Sample Event 5 9 13 1990 MW 3 6 000E 02 1155 0 4 45E 02 2 882E 24 Yes Yes Sample Event 5 9 13 1990 MW 4 3 800E 02 1135 0 4 45E 02 4 444E 24 Yes Yes Sample Event 5 9 13 1990 MW 5 1 500E 00 1194 0 4 45E 02 1 270E 23 Yes Yes Sample Event 5 9 13 1990 MW 6 1 000E 03 1267 0 4 45E 02 3 290E 28 Yes Yes Sample Event 5 9 13 1990 MW 7 1 000E 03 1277 0 4 45E 02 2 109E 28 Yes Yes Sample Event 5 9 13 1990 MW 8 1 000E 03 1245 0 4 45E 02 8 757E 28 Yes Yes Sample Event 6 4 3 1991 MW 1 1 900E 00 1177 0 5 06E 02 2 735E 26 Yes Yes Sample Event 6 4 3 1991 MW 12 2 200E 02 1090 0 5 06E 02 2 575E 26 Yes Yes Sample Event 6 4 3 1991 MW 13 1 900E 02 1125 0 5 06E 02 3 790E 27 Yes Yes Sample Event 6 4 3 1991 MW 14 1 000E 03 1088 0 5 06E 02 1 295E 27 Yes Yes Sample Event 6 4 3 1991 MW 15 1 000E 03 1000 0 5 06E 02 1 108E 25 Yes Yes Sample Event 6 4 3 1991 MW 2 1 000E 03 1192 0 5 06E 02 6 744E 30 Yes Yes Sample Event 6 4 3 1991 MW 3 1 500E 01 1
140. E 23 Attained 1 000 3 Attained 1 000 3 0 05 0 8 Sample Event 6 12 1 52E 26 3 16E 26 Attained 1 000 3 Attained 1 000 lt 3 0 05 0 8 Sample Event 7 12 8 25bE24 4 71E 24 Attained 1 000 3 Attained 1 000 3 0 05 0 8 Sample Event 8 12 1 09E 20 1 30E 20 Attained 1 000 3 Attained 1 000 lt 3 0 05 0 8 Sample Event 9 12 1 82E 19 3 12E 19 Attained 1 000 lt 3 Attained 1 000 lt 3 0 05 0 8 Sample Event 10 12 2 01E 23 4 38E 23 Attained 1 000 3 Attained 1 000 lt 3 0 05 0 8 Sample Event 11 12 3 64E 34 1 24E 33 Attained 1 000 lt 3 Attained 1 000 lt 3 0 05 0 8 Sample Event 12 12 5 62E 24 1 54E 23 Attained 1 000 3 Attained 1 000 3 0 05 0 8 Sample Event 13 12 1 61E 22 4 50E 22 Attained 1 000 3 Attained 1 000 3 0 05 0 8 Sample Event 14 12 773E47 144E 16 Attained 1 000 lt 3 Attained 1 000 3 0 05 0 8 Sample Event 15 12 4 58E 07 6 70E 07 Attained 1 000 lt 8 Attained 1 000 253 0 05 0 8 Note N C means not conducted due to a small sample size N lt 4 or that the mean concentration is much greater than the cleanup level Sample Size is the number of sampling locations used in the power analysis Expected Sample Size is the number of concentration data needed to reach the Expected Power under current sample variability MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 11 MAROS TUTORIAL Authors Aziz J J Groundwater Se
141. Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MOMENT ANALYSIS Moment Trend results from the Zeroth First and Second Moment analyses for the Upper Aquifer monitoring well network were varied Moment Trend results from a selected Upper Aquifer monitoring well dataset are given in the Moment Analysis Report Appendix 10 All 12 wells were used in the moment analysis Mann Kendall Trend Analysis Comment Zeroth Decreasing The decrease in dissolved mass maybe due to the extraction system moving high concentration groundwater from source zones to nearby monitoring wells or the change in monitoring wells sampled over the sampling period analyzed Increasing The center of mass is moving down gradient relative to the approximate source location MW 1 through time in a South Westerly direction perpendicular to groundwater flow Probably The plume shows a probably increasing trend in the direction of groundwater flow Increasing and increasing perpendicular to groundwater flow This indicates that wells representing areas on the tip and the sides of the plume are increasing Increasing Note The zeroth moment or dissolved mass estimate can show high variability over time due to the fluctuating concentrations at the most contaminated wells as well as a varying monitoring well network This may result in an unexpected increasing trend of mass over time To investigate the influence of fluct
142. FA No further action SW Surface water lt lt Back TREND ANALYSIS Help Provides information on the screen specific input requirements Meere hag and ferre fld ior Cut man en Serm omelets Continue t Dig c 55 4 E g x Q F4 ul a Version 2 1 October 2004 46 At this point the Modeling and Empirical Trend results have been entered You may now proceed to the Step 3e LTM Long Term Monitoring Analysis to weight the Plume Information and analyze the trends in the groundwater data Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MAROS Analysis Statistical and Plume Information Summary by W dl accessed from the Plume Analysis M enu screen allows the user to view the Mann Kendall Trend Analysis Linear Regression Analysis Modeling and Empirical results by well and constituent amp Monitoring and Remediation Optimization System MAROS JENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL Statistical and Plume Information Summary by Well The results from the statistical modeling or empirical analysis for each COC are shown in the data tables sheets below To view the data from each well for individual COC s clicking on the tabs at the top Well Name Source Tail Mann Kendall Regression Modeling Empirical 4 Mw 15 s s s N A NIA IMA 14 s D D N A N A avr s D D N A N A IMw 12 s D D NA NIA avn s D D N
143. Fer Pon erem mro erm iier corner d Pe imm iege cm venies Te cnet ra er iai feno ined Dy Pb A MI V i n pn Mee ey eet A BRA A Rab citm AL chy Ae MUR B4 Da Ceonsokdelian RE Dey e me Ete ry rere be OME Pap OF wet CON amy Ac nd Shabelizal Plume Analyede et mimm mm Tet rmn on EX Spatal Moment Anatyoic saai tuus di hm Aras External ftue rtomestion vw PURI oo Inti mane m arm le ets ain ments MAROS Analysin 2 Moment Analysis Site D etails allows the user to enter the additional site data required in the Moment Analysis Data required includes porosity groundwater flow direction approximate contaminant source location and aquifer saturated thickness Thefollowing parameters are to be entered for this tutorial Groundwater flow direction East i e along the x axis shown on Figure A 11 1 enter 0 degrees as groundwater flow direction is defined in degrees from the x axis in a counter clockwise direction Porosity 3096 enter as 0 3 Source Location X20 and Y 20 i e in middle of Tank Field near MW 1 see Figure A 11 1 Uniform Saturated A quifer 12 ft thick Click here for choices feme ue e t hl Sot agame it Coe uada amegas lledoais select Version 2 1 October 2004 A 11 29 To enter the groundwater flow direction dick on the down arrow in the first text box under Groundwater Flow Direction A list of choices will appear
144. Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 1990 1 5 67 1990 2 4 65 1990 3 2 62 4 25 1990 4 4 07 1991 1 3 02 1991 2 7 99 1991 3 3 17 4 41 1991 4 3 44 1992 1 4 53 4 58 0 107 1992 2 6 60 1992 3 3 71 4 66 1992 4 3 80 1993 1 4 41 1993 2 4 90 1993 3 5 96 5 00 1993 4 4 73 Mean of the yearly averages PVariance of the yearly averages Step 5 Calculate t and Cstl 425 544 5 t 1 04 s 0 10 m 4 gaha 498 nos s 0 107 m 4 Step 6 The likelihood ratio is LR exp pie m 7 exp 3 06x 4 2 x 1 04 x MED MN 4 83 m m 1l 4t 4 4 1 4 1 04 Step 7 SinceA LR lt B 0 11 4 83 lt 9 more data need to be collected to perform the test again Inn the above test we use p 0 10 which represents stringent control of error rates especially in terms of false negative rate If in the above test a p 0 20 or a 0 1 and B 0 6 then we will get B 2 4 This may lead to a different condusion since LR is greater than B in these two cases Therefore the test result is dependent on the levels of false error rates which we hope to control METHOD 5 MANN KENDALL TEST FOR TRENDS IN CONCENTRATION DATA The Mann Kendall test is a non parametric statistical procedure that is well suited for analyzing trends in data over time Gilbert 1987 The Mann Kendall test can be viewed as a Version 2 1 A 7 21 Air Force C
145. G WELL LOCATIONS Version 2 1 A 11 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Geology Hydrogeology The shallow geologic unit under the Service Station known as the Upper Aquifer consists primarily of sand and gravel The Upper Aquifer has an approximate saturated thickness of 12 feet The groundwater flow direction is predominantly toward the East and the groundwater seepage velocity is approximately 92 ft r Parameter Value Seepage V elocity 92 ft yr Effective Porosity 3096 Approximate Zone A Source Location Near Well MW 1 Approximate Saturated Thickness 12ft General Groundwater Flow Direction East TABLE A 11 1 EXAMPLE SITE SERVICE STATION SITE PARAMETERS Remedial Action and Long Term Monitoring A site investigation of the service station was performed in 1986 and the results showed that the groundwater plume in the shallow Upper Aquifer principally contains benzene and is 270 ft long approximately 150 feet wide The plume also contains ethylbenzene toluene and xylenes at concentrations above the MCL level According to the results of the site investigation a leaking underground storage tank is the source of benzene Nonaqueous phase liquid NAPL was found in the source area and the leaking tank was removed along with excavating the contaminated soil The area that extends from the edge of the property across Sunnyville Street M
146. Graphing accessed from EESE the MAROS Output Reports screen allows the Trend Summary Results Graphing user to view print graphical Trend summary arent emet it neuen manta E04 De are o hm Tomi veers results in Excel Excel Graph s Takes the user to the Excel Graph screens Back Returns the user to the M AROS Output Reports screen Help Provides information on the screen specific input requirements TREND ANAL YSS Trend Summary Results Graph accessed from the Trend Summary Results Graphing screen allows the user to view print graphical Trend summary results in Excel This will open Excel on your computer to provide the trend result T Source Wells Poin Trend Results for BENZENE 40 A MNI PD o0 O Mw O cti 0 A Mw 3 D A MP 5 y 150 100 50 50 AB ww reqp 150 200 250 Print Chart 20 O Mw 1 D 40 A MWw 4 0 ATail Wells O Source Wells Y Coordinate 60 Almw s 0 Back to A MW S Access 80 A Mw 8 5 Trend Result 100 Increasing I 120 z 4 140 1 X Coordinate I4 4 DI BENZENE ETHYLBENZENE Air Force LIMP lal bf graphs Excel G raph s Takes the user to the Excel Graph screens Print Chart Prints the current summary graph Back to Access Returns the user to the Trend Summary Results Graphing screen Note Do not change the name or content of the worksheet xIsLO ETr
147. HOWS CHANGE IN CENTER OF MASS OVER TIME The first moment esti mates the center of mass coordinates Xc and Yc for each sample event and COC The changing center of mass locations indicate the movement of the center of mass over time Whereas the distance from the original source location to the center of mass locations indicate the movement of the center of mass over time relative to the original source The 2 D coordinates for the center of mass of the plume for a given sample event can be calculated from Fn 0000400 Qus Meme yg ffia POE f M e 00 00 00 nC dxdyaz nC dxayaz oo 00 oo o00 oo oo where Ci is the concentration of the COC n is the porosity and x y are the spatial coordinates Similar to the Zeroth Moment calculation the data are spatially discontinuous therefore a numerical approximation to this equation is required To conduct the numerical integration the horizontal plane x y was divided into contiguous triangular regions with the apex of each triangle defined by a well sampling location with an associated COC concentration at each sample location A spatial interpolation method over these triangles allows the first moment calculations using Delaunay Triangulation see A ppendix A 2 for methodology The Delaunay triangulation is a rough way to discretize the domain The following formulas represent the 2 D approximation of the center of mass QALXVCS r DOM 2 s VV
148. IMwW 13 s 344990 ETHYLBENZENE 4 7E 02 a s IMW 14 s 4 31991 TOLUENE 5 5E 01 sped fi C l nput requi rements IW 14 12194998 TOLUENE 1 2E 00 IMW 14 697998 TOLUENE 8 0E 01 M 14 s 12104997 TOLUENE 8 5E 01 I2 IMW 14 s 6 27 1997 TOLUENE 6 5E 01 7 IW 14 s 5 28996 TOLUENE 7 DE 01 zs IMW 14 s 5 2992 TOLUENE 8 5E 01 Ed IMW 13 S 710991 ETHYLBENZENE 5 0E 04 ND IMA 14 S 11 47 989 XYLENES TOTAL 1 9E 01 a M 13 s 10 3 9891 ETHYLBENZENE 20E 03 zi W ee lt lt Back Help Reduced Data Plot accessed from the Reduced Data Table screen allows the user to view the reduced data in graphical form amp 3 Monitoring and Remediation Optimization System MAROS xj Choose the Well and Chemical of interest from Reduced Data Plot the dropdown boxes at the top of the screen Ec i aka a seraa i a bes bale Tee alee Choose the graph type i e Log or Linear Click Sele wen HVS Z chemical FENZENE EI Graph on graph to proceed 9 P Sf SP Seeur ry EY x S es Lo View Report To print the current graph and AM ee eee E E data dick View Report to proceed jeep os i 3 0E 02 x Graph Back Returns the user to the Reduced D ata Table owe os screen pes ese N ext Takes the user to the M ain M enu screen TREND ANALYSIS lt lt Back Next gt gt Help Provides information on the screen specific input requirements Marti ritmo sad aeree lach rm Dytimisatisa Syrien MODA Gate Gon
149. Lag 1 adjustment based on AR 1 model first order autoregressive model Seasonal patterns but no Sub Method 2 The method makes inferences out of serial correlation Seasonal adjustment seasonally adjusted residuals remove seasonal mean Seasonally adjusted Sub Method 3 The method adjusts for serial correlation of residuals exhibit serial Combined adjustment seasonally adjusted residuals correlation References ASTM 1998 Standard Guide for Developing Appropriate Statistical Approaches for Ground Water Detection Monitoring Programs D6312 98 American Society for Testing and Materials ASTM Pennsylvania Barcelona M J Wehrman H A Schock M R Sievers M E and Karny J R 1989 Sampling Frequency for Ground Water Quality Monitoring Nevada Office of Research and Development U S Environmental Protection Agency Clayton C A Hines J W and Elkins P D 1987 Detection Limits with Specified Assurance Probabilities Analytical Chemistry Vol 59 No 20 Version 2 1 A 7 43 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Cohen J 1988 Statistical Power Analysis for the Behavioral Sciences New Jersey Lawrence Erlbaum Associates Davis C B and McNichols R J 1987 One sided Intervals for at Least p of m Observations from a Normal Population on Each of r Future Occasions Technometrics Vol 29 pp 359 70 Davis C
150. M SOFTWARE APPENDIX A 8 MAROS SITE RESULTS Authors N ewell C J and Aziz J J Groundwater Services Inc The preliminary monitoring system optimization results are based on site classification source treatment and monitoring system category Figure A 8 1 The decision matrices below are heuristic rules based on the judgment of the authors Users are expected to review and modify as necessary to reflect site specific hydrogeology contaminants risks and regulatory considerations General recommendations by more rigorous statistical methods can be obtai ned by using the more detailed optimization approaches outlined in Appendices A 2 and A 3 General site results are outlined by for Sampling Frequency Well Sample Density and Duration of Sampling These criteria take into consideration plume stability type of plume and groundwater velocity The results are specific to only one COC Each COC considered in the MAROS software is assigned a result based on the criteria outlined here Tail Source FIGURE A 8 1 DECISION MATRIX FOR ASSIGNING MONITORING SYSTEM CATEGORIES MODERATE M EXTENSIVE E LIMITED L PLUME STABILITY INCREASING I PROBABLY INCREASING PI NO TREND NT STABLE S PROBABLY DECREASING PD DECREASING D Weighted Average Two types of weighting are available within the MAROS Analysis software i e LOE weighting and well weighting The weighting for these analyses follow a simple weighted average defi
151. Mass Report and analysis results dick View Report to proceed Distance from Sourceto Center of M ass screen Next Takes the user to the Second M oment Plot screen Help Provides information on the screen specific input requirements Note The information displayed in this screen can also be viewed in report form First Moment Change in Location of Mass Center Over Time Report from the MAROS Output Screen or by clicking on View Report see Appendix A 10 for an example report Version 2 1 October 2004 41 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Spatial M omentAnalysis Change in Plume Spread Over Time Second M oment Plot Change in Plume Spread Over Time accessed from the First M oment Plot Change in Location of M ass Over Time screen allows the user to view the Second Moment Analysis results by constituent over time The second moment indicates the spread of the contaminant about the center of mass Sxx and Syy or the distance of contamination from the center of mass The Second Moment represents the spread of the plume over time in both the x and y directions lonitoring and Remediation Optimization System MAROS zox Choose the chemical of interest from the change Plame Spd Oe dropdown boxes at the top of the screen RE lect cu n mical below to graph The second moment trend result in the box below reflects the chemical chosen
152. N OPTIMIZATION SYSTEM SOFTWARE Statistical Plume Analysis Linear Regression Plot accessed from the Linear Regression Statistics screen allows the user to view the linear regression data in graphical form Monitoring and Remediation Optimization System MAROS E ni xj Linear Regression Plot Select a well and chemical below to graph The concentration trend result in the box below reflects the chemical and well chosen to be graphed Select wen Mw 12 Chemical ETHYLBENZENE E ud Graph Type SO db fh P SMS Log 3 EY P S FN s c SOS 8 d SY s Linear 2 0E 01 T 18E01 4 6E 01 Graph E 1 4E 01 e S 12E01 tf Heel View Data t 6 0E 02 4 0E 02 Ln Slope o MRE i oom 7 Confidence in gt Trend z 99 2 N cov E Linear Regression Trend D Cz SE Now hereaehg Q Probably horashg p Stble 9 Probably Decreasing PD Decrashg D No Treid TT NT NotApplicabie Qui Moke nt cata Next gt gt View Report Help Version 2 1 October 2004 Choose the Well and chemical of interest from the dropdown boxes at the top of the screen Choose the graph type i e Log or Linear Click Graph on graph to proceed View Report To print the current graph click View Report to proceed Back Returns the user to the Linear Regression Statistics screen Next Takes the user to the Statistical Plume Analysis Summary screen Help Provides information on the screen 35 Air Force C
153. N SYSTEM SOFTWARE Spatial M oment Analysis Change in Location of M ass Center Over Time First M oment Plot Change in Location of M ass Over Time accessed from the First M oment Plot Distance from Source to Center of M ass screen allows the user to view the First Moment Analysis results by constituent over time The first moment estimates the center of mass of the plume coordinates Xc and Y c for each sample event and COC The center of mass locations indicate the movement of the center of mass over time E Monitoring and Remediation Optimization System MAROS nl xi First Moment Plot Change in Location of Center of Mass Over Time Select a chemical below to graph The first moment trend result in the box below reflects the chemical chosen to be graphed Select Chemical BENZENE v Ye ft Xe ft in 2 gt 4 5 Z W z e z x lt a 2 Help lt lt Back Next gt gt View Report Back Returns the user to the First M oment Plot Choose the chemical of interest from the dropdown boxes at the top of the screen Choose the graph type i e Log or Linear Click Graph on graph to proceed The source location coordinates are shown on the screen left To view the data for the graph choose View Data this shows a table with the Xc Yc and Source Distance for all sample events View Report To print the First Moment Change in Location of Center of
154. OFTWARE Sampling Frequency Analysis Sampling Frequency Analysis accessed from the Sampling O ptimization screen by clicking Sampling Frequency Analysis is the control screen to determine the frequency of sampling at each location The Modified CES method adopted from Cost Effective Sampling by Ridley et al 1998 is applied This method is based on the analysis of recent and overall trends of COC concentrations Details of the method are available in Appendix A 9 E Meeatonng and Hnesedialue iplrmeratien yukin PAIN The term recent period refers to the time period in which the latest series of sampling events occurred It is used to differentiate for example the latest two years of sampling from the history of sampling all sampling events The etna mam ree mm Quim recent period could contain any series edi of continuous sampling events ending with the latest sampling event Darei ering he beonning nci ending coming evert TION fiw sues Dose 1 j Te f Eao Ever T5 zi From Selects a sampling event from the dropdown list as the beginning of the Medo eg recent period CES To Selects a sampling event as the end Hua eem EI of the recent period SAMPLING OPTIMI Q Confirm Confirms the recent period defined by the user Options Shows the Sampling Frequency Analysis Options screen where the Rate of Change parameters for analyzing the concent
155. OPPER 1 7E 00 1 3E 00 30 4 PERCHLORATE 1 2E 01 9 2E 02 27 6 Note Top COCs by toxicity were determined by examining a representative concentration for each compound over the entire site The compound representative concentrations are then compared with the chosen PRG for that compound with the percentage excedence fron the PRG determining the compound s toxicity All compounds above exceed the PRG Prevalence Total Total Percent Total Contaminant of Concern Class Wells Excedences Excedences detects LEAD MET 12 10 83 3 10 BENZENE ORG 12 8 66 7 10 BARIUM MET 12 7 58 3 12 TOLUENE ORG 12 5 41 7 12 COPPER MET 12 4 33 3 12 1 2 DICHLOROBENZENE ORG 12 4 33 3 12 1 1 1 2 TETRACHLOROETHANE ORG 12 3 25 0 9 PERCHLORATE INO 12 2 16 7 10 Note Top COCs by prevalence were determined by examining a representative concentration for each well location at the site The total excedences values above the chosen PRGs are compared to the total number of wells to determine the prevalence of the compound Mobility Contaminant of Concern Kd PERCHLORATE BENZENE 0 0984 TOLUENE 0 347 1 1 1 2 TETRACHLOROETHANE 0 857 1 2 DICHLOROBENZENE 1 91 LEAD 10 BARIUM 11 COPPER 40 Note Top COCs by mobility were determined by examining each detected compound in the dataset and comparing their mobilities Koc s for organics assume foc 0 001 and Kd s for metals MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of Project User Name
156. ORT FILES RESULTS Column Number Field Name D escription 1 ITESTSEQ Tests Sequence N umber 2 RESULTSEQ Results Sequence N umber 3 PARLABEL Parameter Label 4 PRCCODE Parameter Classification Code 5 PARVQ Parameter Value Qualifier 6 PARVAL Parameter Value 7 PARUN Parameter Value Uncertainty 8 PRESICION Parameter Value Precision 9 EXPECTED Expected Parameter Value 10 EVEXP Integer Value of Expected Value 11 EVMAN Decimal Value of Expected Value 12 EVPREC Predsion of Expected Value 13 MDL Method Detection Limit 14 RL AFCEE Reporting Limit 15 UNITS Units of Measure 16 VQ 1C 1st Column Value Qualifier 17 VAL 1C 1st Column Value 18 FCVA LEXP 1st Column Value Integer Value 19 FCVALMAN 1st Column Value Decimal Value 20 FCVALPREC Precision of 1st Column Value 21 VQ CONFIRM 1st Column Value Qualifier 22 VAL CONFIRM Confirm Column Value 23 CN FVA LEXP Confirming Value Integer Value 24 CNFVALMAN Confirming Value Decimal Value 25 CNFVALPREC Precision of Confirming Value 26 DILUTION Dilution Value 27 DILEXP Dilution Value Integer Value 28 DILMAN Dilution Value Decimal Value 29 DILPREC Precision of Dilution Value 30 UNCVALEXP Uncorrected Value Integer Value 31 UNCVALMAN Uncorrected Value Decimal Value 32 ICRVALEXP Corrected Value Integer Value 33 CRVALMAN Corrected Value Decimal Value 34 DQTYPE Data Qualifier Type
157. POC well in the monitoring network The default values for all potential locations are True Version 2 1 A 4 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE H ull node Slope Factor The SF threshold for nodes locations located on the edge convex hull of the triangulation domain When SF of a hull node is less than this threshold and if the node is Removable it will be eliminated from the monitoring network The current default value for this parameter is 0 01 The threshold for hull node is usually more stringent than that of the inside node because the elimination of a hull node may cause reduction in the triangulation area thereby causing greater information loss reduction in AR For contrast the elimination of an inside node will only affect the average concentration ratio CR Area Ratio AR The ratio of triangulation area represents the area of a contaminant plume at current optimization step to the original triangulation area before optimization If the AR value in an optimization step is less than the threshold the optimization will be stopped and locations eliminated in this step will be resumed The default value is 0 95 Concentration Ratio CR The ratio of average concentration of a contaminant plume at current optimization step to that of the original value before optimization If the CR value in an optimization step is less than the threshold
158. PR to within 5 In using a 0 01 instead of o 0 001 0 05 50 0 001 with other conditions unchanged the sensitivity of the individual comparison will be significantly increased However in corrective action monitoring for evaluating the attainment of cleanup standards the FWFPR no longer poses a threat Assuming the site is declared clean only when all constituents of concern in all wells attain the cleanup standards the FWFPR will always be less than or equal to the maximum comparison wise false positive rate For example if the maximum comparison wise false positive rate is Omax 0 2 for n 10 independent future comparisons the FWFPR is given by x Probability all wells dean lt Omax i 0 27 1x 107 lt lt omax However the facility wide false negative rate FWFNR may now cause problems if cleanup of the site is declared only when all constituents at all wells attain cleanup standards In this case and for large monitoring programs even if all wells have attained the cleanup standard a non attainment decision could still be reached due to FWFNR The cause of the problem associated with FWFNR can be analyzed in the same way as that of detection monitoring Regardless of the strategy used simultaneous analysis of more wells and more constituents will increase facility wide false error rates either FWFPR or FWFNR Therefore a non statistical suggestion for reducing the FWFPR and FWENR is to choose as few constituents and wells as
159. Probably Increasing PI Stable 5 Probably Decreasing PD 236 z Decreasing D No Trend NT Not Applicable N A due to insufficient data lu ps Next View Report Help Version 2 1 October 2004 Choose the Well and chemical of interest from the dropdown boxes at the top of the screen Choose the graph type i e Log or Linear Click Graph on graph to proceed View Report To print the current graph click View Report to proceed Back Returns the user to the M ann Kendall Statistics screen Next Takes the user to the Linear Regression Statistics screen Help Provides information on the screen specific input requirements 33 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Statistical Plume Analysis Linear Regression Statistics accessed from the M ann K endall Plot screen allows the user to view the Linear Regression A nalysis results by well and constituent For further details on the Linear Regression Analysis Method see A ppendix A 2 E Monitoring and Remediation Optimization System MAROS E lini xi To nav i g ate th e resu ts for i n d i V i d u al Cingar Regression Statisties constituents click on the tabs at the top of the The Linear Regression Analysis is used for analyzing a single groundwater constituent multiple screen constituents are analyzed separately Each tab below shows the statistics for one constituent
160. RE MAROS Analysis Weighting Wells Plume Information by Wal Weighting accessed from the Results of Information Weighting screen allows the user to weight the individual wells by all chemicals or by constituent To weight wells select Weight Wells on the right side of the screen Then choose to either enter the weight of each well within individual COC datasets by clicking on Individual Chemicals difficult approach and then entering the weights in the column to the right of the results on each tab Or choose to weight the data from each well for all COC s by clicking on All Chemicals easy approach and then entering the data on the front tab Choices for weighting methods range from High to Low If you choose to weight trend methods select Do Not Weight TREND ANALYSIS Monitoring and Remediation Optimization System MAROS Anl x Plume Information Consolidation Step 2 Plume information by Well Weighting The results from the statistical modeling or empirical lines of evidence for each COC are shown in the sheets below Choose to either enter the weight of each Well within individual COC datasets by clicking on Individual Chemicals dificult appoach and then entering the weights in the column to the right of the results on each tab Or choose to weight the data from each wel for all CD s by clicking on All Chemicals easy approach and then entering the data on the front tab If you choose no
161. REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MAROS DETAILED SCREEN DESCRIPTIONS Start Screen The Start Screen gives the user access to the software system Enter the User N ame and Project Name in the boxes to the left of the Start Button The User N ame and Project N ame will appear in the headings of MAROS output reports Click Start to proceed to use the database software E Monitoring and Remediation Optimization System MAROS HQ Air Force Center for Environmental Excellence Monitoring and Remediation Optimization System MAROS Software Tool Version 2 0 Julia J Aziz Neng Ling Jim Gonzaks Charles Newell Ph D P E Hanadi S Rifai Ph D P E Javier Santilfan Groundwater Services Ine University of Houston AFCEE Data management tool for analyzing and optimizing groundwater monitoring programs User Hame f Project Name Start Copynght 2002 Air Force Center for Environmental Excellence Utilizing the MAROS software is analogous to a train trip You begin the expedition by importing your raw groundwater data that has been collected over several sampling periods from the field site of interest As you journey through the software you can get off at any station along the way The results that you are presented with at each stop whether graphical or in a report will be based on increasingly more consolidated data These data consolidation steps will lead to a higher degree of assumptions being used in order to r
162. RING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well construction is not adequately documented Field data collection procedures are inadequate Sample collection procedures are inappropriate Sample analysis procedures are inadequate or undocumented Examples of program implementation problems include missing or ambiguous data different results for duplicated samples and presence of chemicals in blanks The remedies for program implementation problems include resampling using improved protocols employing more experienced personnel and employing a reputable laboratory for analysis Details for the prevention recognition and correction of typical program implementation problems are presented in Table A 7 3 A monitoring system with the above problems will appear to be functioning properly but will actually be producing data that are misleading uninterpretable or incorrect The qualitative evaluations described above should be the initial steps used to reduce false positive and false negative rates These steps should be performed before any of the data analyses or statistical approaches presented later in this appendix are employed Version 2 1 A 7 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 1 Prevention Recognition and Correction of sample space problems Problem Prevention Recognition Correction 1 Wells not Use basic hydrogeol
163. Repert Wes Hey COC as potential locations The Selected status of each location will be set to True Back Returns the user to the W ell Redundancy Analysis D daunay M ethod screen View Report Generates a report with sampling location optimization results for the one sampling event selected by the user The user should analyze COCs before viewing the report After getting feedback from the report the user can go back to rerun the analysis by changing parameters or selecting a different series of sampling vents Next Proceed to the Excel M odule All in one Results screen Help Provides additional information on software operation and screen specific input requirements Steps for use 1 Choose the COC for analysis by selecting from the COC dropdown list or typing in the name 2 Set the Sdected check box of a location to decide whether this location is included in the analysis Set Removable check box to decide whether a location can be eliminated by the optimizing process 2 Set the Sdected and Removable status of a location by using the Shortcut Menu in worksheet xIsD daunay2K This can be performed only when the worksheet xIsD elaunay2K is running 3 Press button Analysis and the screen will switch to worksheet xIsD elaunay2K The data will be transferred to worksheet xlsD elaunay2K Version 2 1 62 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYS
164. S METHOD Well Original MAROS Final Reasoning Frequency Recommende Recommendation d Frequency MW 1 Semiannual since 96 Annual Annual MW 12 Semiannual since 96 Annual Semiannual Most downgradient along the plume centerline and serve as a sentry well MW 13 Semiannual since 96 Biennial Biennial Non detects or below MCL since 94 amp in the central part of the plume MW 14 Semiannual since 96 Biennial Annual Non detects since 91 but monitors the downgradient part of plume MW 15 Semiannual since 96 Biennial Biennial All historical concentrations are Non detects and far from plume MW 2 Semiannual since 96 Biennial Annual Non detects or below MCL since 94 but it monitors the lateral migration of plume near the source MW 3 Semiannual since 96 Annual Annual Recommended for elimination MW 4 Semiannual since 96 Annual Annual MW 5 Semiannual since 96 Annual Annual MW 6 Semiannual since 96 Biennial Biennial All historical concentrations are nondetects amp an upgradient well MW 7 Semiannual since 96 Biennial Biennial All historical concentrations are nondetects or below MCL amp an upgradient well MW 8 Semiannual since 96 Biennial Biennial All historical concentrations are nondetects or below MCL amp a cross gradient well in the upgradient section of the plume The default ROC parameters were used in the above analysis i e 0 5MCL ear 1 0MCL year and 2 0MCL ear for the Low Medium
165. SIS METHOD on scssssssssssssssssssssssssssssee A 5 1 A 6 DATA SUFFICIENCY ANALYSIS ccsssssssssssssssssssssssssssseessessssssssssssses A 6 1 A 7 FALSE POSITIVE NEGATIVE MINIMIZATION sssrin A71 A 8 MAROS SITE RESULTS METHOD rnnt A 81 A 9 SAMPLING FREQUENCY ANALYSIS MODIFIED CES METHOD snnnnnnnnnnnnnnnttnntttntttttnnnnnnnnno A 9 1 A 10 SAMPLE MAROS REPORTS errrrrnrnnnrnrntnnntttnttntttnnnnnnnnn A 10 1 AH MAROS TUTORIAL ote eee A 11 1 Disclaimer MAROS is made available on an as is basis without guarantee or warranty of any kind express or implied The United States Government Groundwater Services University of Houston the authors and reviewers accept no liability resulting from the use of MAROS or its documentation Version 2 1 i Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Implementation of M AROS and interpretation of the results are the sole responsibility of the user List of Acronyms ACRONYM DEFINITION AFCEE Air Force Center for Environmental Excellence AR Area Ratio ASTM American Society for Testing and Materials CERCLA Comprehensive Environmental Response Comprehensive and Liability Act CES Cost Effective Sampling COC Constituent of Concern CR Concentration Ratio CT Concentration Trend DL Detection Limit ERPIMS Environmental Re
166. SOFTWARE Click the Data Sufficiency Analysis Menu button to return to the Data Sufficiency Analysis M enu screen 14 Select Analysis 2 from the Data Sufficiency Analysis M enu screen to perform the risk based site scale power analysis The Parameters for Risk Based Power Analysis screen will appear amp 3 Monitoring and Remediation Optimization System MAROS x Parameters for Risk Based Power Analysis Proceed through steps 1 to 4 to input the required information In step 1 input the groundwater flow angle Since the flow direction is towards The following parameters are needed for the risk based power analysis The user should provide representative wells along the plume centerline for a regression of concentrations against diatance down the plume centerline Proceed Through Steps 1 4 East i e 0 degree counter clockwise 1 Groundwater Flow Angle 3 Select Sampling Events for Analysis from the X axis input O to confirm fom oe s Fem Senge Fret tl Analysis Site Details Te Sample Event 15 z For the Distance to Receptor input z are ee 4 Select Plume Centerline Wells 1000 to confirm the distance This Ed 2 veee am wv distance is then used to locate the 5 QUARE wd sy hypothetical statistical compliance 3 pedal ME boundary HSCB In this example the EI m isle En HSCB is 1000 ft downgradient from IJ
167. STEM SOFTWARE Title 2236 API PIC B FX2 eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers FIGURE A 4 2 LIMIT OF MIGRATION OF PETROLEUM HYDROCARBON PLUMES BASED ON COMBINED RESULTS FROM FOUR STUDIES NEWELL AND CONNOR 1998 FOUR STUDIES INCLUDED THE LAWRENCE LIVERMORE STUDY RICE ET AL 1996 TEXAS BEG STUDY MACE ET AL 1997 FLORIDA RBCA STUDY GSI 1997 AND UNPUBLISHED DATA FROM THE HGDB DATABASE NEWELL ET AL 1990 Title 2236 API PIC C REV eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers FIGURE A 4 3 SUMMARY STATISTICS FOR INDIVIDUAL PLUME A THON STUDIES MOST STUDIES FOCUSED ON BENZENE OR BTEX RELEASES FROM SMALL FUEL RELEASES SUCH AS UNDERGROUND STORAGE TANKS USTS AT SERVICE STATIONS Version 2 1 A 4 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE PLUME TREND DATA USED FOR STEP 4 Two studies California and Texas evaluated the trends in dissolved petroleum hydrocarbon plumes Rice et al 1995 developed the following classification system to evaluate BTEX plume trends Expanding Residual source present Mass flux
168. STEM SOFTWARE Import MAROS Archive File Import Archive File accessed from the Data Managenent Menu screen is used to import previously archived data files as follows E3 Monitoring and Remediation Optimization System MAROS To i mport archived data i nto the ful database import MAROS Archive Fife 1 Enter the full file path and filename of the ac ie beet aa oe a cer pai archived file to import or dick the browse button to find the import file The Folder and File name you choose will appear in the top two boxes 2 Choose the import option that corresponds to the import data Choose Replace if all the data for the analysis are in the file to be imported Replace should be chosen for a new analysis you are replacing an empty lt lt Back her file After you choose Replace a dialog box will ask if you really want to replace the data select Yes Choose Append if the file represents additional data to those already present in the database Appended data may be data for a new sample event or additional well data 3 Click Retrieve to proceed with importing the archived file to the existing database A dialog box will inform you if the data have been successfully imported Folder p GS USERS 2236 LTMP archives Filename archivesiteB mdb DATA MANAGEMENT Back Takes the user back to the D ata M anagement screen after the data have been
169. See manual text or Help for description of trend determination method COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average Values near 1 00 indicate that the data form a relatively dose group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about lt lt Back Next gt gt View Report Help the mean Residuals COV The Coefficient of Variation COV of the residuals is a statistical measure of how the residuals the difference between the predicted values and observed values vary about the mean value Values near 1 00 indicate that the data form a relatively close group about the mean value and that the Linear Regression statistics can be relied upon more strongly Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about the mean and therefore the M ann K endall analysis should be relied upon more strongly BENZENE ETHvLBENZENE TOLUENE XYLENES TOTAL Statistical Analysis Results Last column is the result for the trend nc Well ST mgl LnSlope cov inTrend Trend 38 01 14E 03 170 99 6 D 3 6E 02 1 TE 03 159 100 0 D 17E 02 15E 03 14 100 0 D 9 5E 03 1 0E 03 161 99 6 D 5 0E 04 0 0E 00 0 00 100 0 s 23E 02 5 8E 04 331 33
170. TEM SOFTWARE 4 Run worksheet xlsD elaunay2K by following the instructions shown in screen xIsD elaunay2K introduced shortly 5 After finishing analysis in worksheet xlsD elaunay2K send results back by pressing Back to Access button The screen will switch back and locations that have been eliminated will be shown in field Eliminated Selected and Removable fields will also be updated if any change has been made in module xIsD elaunay2K 6 Select other COCs and go back to step 1 until all the COCs have been analyzed Version 2 1 63 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE XIsD elaunay2K xlsD elaunay2K accessed from the Wal Redundancy Analysis Excel M odule screen by clicking Analysis is a stand alone Microsoft Excel worksheet used to perform well redundancy analysis by the Delaunay method This worksheet contains two parts a chart sheet Well Locations and a datasheet D ataSheet The user can click the sheet tab on the lower left corner of the worksheet to Switch between the two parts The Well Locations chart sheet is shown on the next page The figure below shows the D ataSheet d xlsD elaunay 8 e 59 1 F Source Data Part Output Part i 6 Num of locations E32 7 mt Point selected status 9 ordinates Triangle Area Trian
171. TION OPTIMIZATION SYSTEM SOFTWARE e Basic output 1 page Sampling Plan that is intended to be used as a strawman or basis for discussion not as an authoritative detailed statistically based product The user can apply additional tools in MAROS to refinethis basic plan An important premise for the report is knowledge of historical trends for each COC and each well However the software is not a kriging tool atthis time Sample data reduction and data analysis tools result in summary reports Note For kriging available software products include GEOEAS or GEOPack from the U S EPA Also some commercial software for kriging include GS Geostatistics for the Environmental Sciences GMS Groundwater Modeling System ArcGIS 8 x and EarthVision These software products include variograms and kriging for the purpose of interpolation but are not specifically geared toward groundwater well network optimization A higher level of statistical knowledge and background would be required to implement these geostatistical tools The AFCEE MAROS Software should be used in Access 20006 along with Excel 20008 in order to analyze the trends in groundwater data as well as perform statistical optimization of well location sampling frequency and duration The software can be used to export data to an Access archive file for future software use Groundwater data can be imported from Excel or ERPIMS files as well as entered manually Version 2 1 2 Air Force Ce
172. TORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Start Using MAROS STEP 1 INSTALLATION If the MA ROS software is not already installed on your computer follow the instructions on page 3 of the User M anual STEP 2 START MAROS To start using the software go to the subdirectory where MAROS is installed e g CAAFCEE MAROS and double click on the afcee MAROS v2 mdb file TheStart Screen will be displayed STEP 3 ENTER USER INFORMATION TheStart Screen gives the user access to the software system i act ty att Lemon Comm JT b HQ Abe Foreo Canter Fi Visirommental Fseelbenee e Monitoring and Remediation bh o Ww Optimization System MAROS Sotiware Tout Enter the following information as User Name and Project Name in the boxes to the left of the Start Button e User Name Enter your name e Projet Name Tutorial imss 28 Click the Start button when finished Hert leen a Jap ud OA be Rea Cu o Lc ada Diac Click here Version 2 1 A 11 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Importing Data STEP I MAIN MENU The M ain screen serves at the center of the user interface The user progressively steps through the Compliance Monitoring Trend Analysis and Optimization Evaluation process by navigating through the options displayed As individual steps of the process are completed options to se
173. The power values range from O to 1 0 N C indicates the analysis is not conducted because of insufficient data sample size lt 4 S E indicates the analysis is not conducted because the mean concentration significantly exceeds the cleanup goal Expected Sample Size The amount of data required to achieve the expected power with the variability shown in the data lt 3 indicates that the data have a very small variability resulting in a high power gt 100 indicates the opposite N C indicates the analysis is not conducted because of insufficient data sample size 4 S E indicates the analysis is not conducted because the mean concentration significantly exceeds the cleanup goal To facilitate the power analysis concentration data are assumed to be either normally or lognormally distributed Results for both assumptions are calculated and provided for comparison In most cases they agree with each other See Appendix A 6 for detailed explanations View Normal Views results calculated under the assumption that data are normally distributed View Log Views results calculated under the assumption that data are lognormally distributed Back Closes this screen and returns to the Individual Well Cleanup Status Results screen View Report Generates a report with optional power analysis results for the type of data and time period selected by the user The user can go back to rerun the analysis by selecting a different type of data or by s
174. Trend NT Not Applicable N A Due to insufficient Data 4 sampling events ND All Samples are Non detect MAROS Version 2 2002 AFCEE 12 1 2003 Page 1 of 1 MAROS Linear Regression Statistics Summary Project User Name Location Service Station State Texas Time Period 10 4 1988 to 12 19 1998 Consolidation PeriodNo Time Consolidation Consolidation Type Median Duplicate Consolidation Average ND Values 1 2 Detection Limit J Flag Values Actual Value Average Median All Source Conc Conc Standard Samples Coefficient Confidence Concentration Well Tail mg L mg L Deviation ND LnSlope_ of Variation in Trend Trend BENZENE MW 8 S 6 7E 04 5 0E 04 6 5E 04 No 9 6E 05 0 97 83 596 S MW 7 S 5 3E 04 5 0E 04 1 3E 04 No 8 2E 05 0 24 79 3 S MW 6 S 5 0E 04 5 0E 04 0 0E 00 Yes 0 0E 00 0 00 100 0 S MW 5 S 1 2E 00 1 2E 00 8 2E 01 No 8 6E 04 0 67 100 0 D MW 3 S 6 9E 02 6 0E 02 7 3E 02 No 1 3E 03 1 05 99 9 D MW 2 S 2 0E 02 5 0E 04 6 9E 02 No 5 2E 04 3 52 92 6 PD MW 1 S 1 0E 00 8 0E 01 9 7E 01 No 1 6E 03 0 92 100 096 D MW 4 T 5 8E 02 1 8E 02 8 6E 02 No 8 5E 04 1 47 99 7 D MW 15 T 5 0E 04 5 0E 04 0 0E 00 Yes 0 0E 00 0 00 100 0 S MW 14 T 1 1E 02 5 0E 04 1 6E 02 No 1 1E 03 1 50 99 9 D MW 13 T 1 8E 02 1 5E 02 1 9E 02 No 1 5E 03 1 03 100 096 D MW 12 T 4 7E 02 2 2E 02 7 0E 02 No 1 7E 03 1 48 100 0 D Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT N
175. Use the scroll bar to see all the choices Click O E to select The groundwater flow direction O is displayed indicating the flow is toward the East Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE To enter the porosity value 3096 type 0 3 in the text box next to the heading MS Moment Analysis Site Details Porosity s Age Taba teni Matran Enter the x and y coordinates of the source in using the text boxes adjacent to Single Source Location In the X ft box type 0 and in the Y ft box type 0 Enter thickness here Peewee T te Too om gms iei te apttd ib prae edu The men Matron Enter the overall saturated thickness of the aquifer in the text box next to Uniform Saturated Thickness Type in the value 12 Maxtworg ext lare tiae v Creta acsi dyetere PAULIE Moment Agalysis Site Det L sands iba larwe Coris t Cov ammi To continue select Next The Spatial M oment Analysis Results screen will be displayed Click hereto proceed Note Where the thickness of the saturated aquifer varies according to well location representative saturated thickness of the aquifer at each well can be entered by clicking on Variable Saturated Thickness and then entering the data for each well B ommum bereed Iso fi smes dete matt teh err
176. VENT 8 1991 Version 2 1 A 11 46 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 15 3 35 55 4 T ww S 100 55 102 190 FIGURE A 11 4 SERVICE STATION BENZENE CONCENTRATIONS FOR SAMPLE EVENT 10 1994 For a generic plume the MAROS software indicates to e Continue semi annual sampling frequency e May need up to 15 wells These MAROS results are for a generic site and are based on knowledge gained from applying the MAROS Overview Statistics The frequency recommendation is for the whole monitoring network and the number of wells seems high Therefore a more detailed analysis for both the well redundancy and sampling frequency utilizing the detailed statistics analysis in the MAROS 2 0 software is needed to allow for reductions and recommendations on a well by well basis These overview statistics were also used when evaluating a final recommendation for each well after the detailed statistical analysis was applied Version 2 1 A 11 47 Air Force Center for October 2004 Environmental Excellence Sampling Optimization AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 4 Sampling Optimization allows the user to perform detailed sampling optimization with modules to optimize sampling location by the Delaunay method and sampling frequency by the Modified CES method or to evaluate data sufficiency by statistical power analysis Sele
177. Version 2 1 A 9 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The adoption of minimum frequency of quarterly sampling is referred to Barcelona et al 1989 The use of sampling intervals at Quarterly Semi Annual Annual and Biennial is very common in long term groundwater monitoring AFCEE 1997 NFESC 2000 and is adopted in MAROS Version 2 1 A 9 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D etails of the M odified CES M ethod Based on the CES method we made some modifications to it and developed the so called Modified CES method The M odified CES method has three major steps that are similar to those of the CES method The details of the decision procedures for the three major steps are given in the sub sections starting from the next page In the Modified CES method Concentration Trend CT by GSI Groundwater Services Inc style Mann Kendall analysis is used instead of the distribution free version of the coefficient of variation for the characterization of the variability The GSI style M ann K endall trend results fall into 6 categories Decreasing Decr Probably Decreasing ProbDecr Stable No Trend Probably Increasing ProbIncr and Increasing Incr The result of nonparametric M ann Kendall analysis is judged with Coefficient of Variation standard deviati
178. W 15 is designated as the down gradient area The land used for this area is retail businesses Well Well Well Type Category 5 MW 1 MW MW 2 MW MW 3 MW MW 4 MW MW 5 MW MW 6 MW 7 MW 8 MW 9 MW 10 MW 11 MW 12 MW 13 MW 14 MW 15 JA ataiaia A iMi o m alulla Version 2 1 A 11 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Note MW Monitoring Well S Source Zone Well T Tail Zone Well TABLE A 11 2 EXAMPLE SITE SERVICE STATION WELL CATEGORIES The regulatory agency involved with the site concluded that Monitored Natural Attenuation would be an adequate remediation remedy for the site due to the size of the plume and its distance from any receptors The original groundwater long term monitoring plan was completed in 1998 It consisted of compliance monitoring with the goal of plume reduction monitoring to verify progress toward achieving cleanup goals over a 30 year period The number of monitoring wells that were sampled in the original Upper Aquifer monitoring network is 15 Figure A 11 1 All monitoring wells have been sampled semi annually in the Upper Aquifer for BTEX since the implementation of the original long term monitoring plan Between 1988 and 1998 15 sampling events had been carried out at the site Version 2 1 A 11 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONI
179. Yes Yes Sample Event 10 1 11 1994 MW 5 1 200E 00 1194 0 4 33E 02 4 139E 23 Yes Yes Sample Event 10 1 11 1994 MW 6 1 000E 03 1267 0 4 33E 02 1 460E 27 Yes Yes Sample Event 10 1 11 1994 MW 7 1 000E 03 1277 0 4 33E 02 9 467E 28 Yes Yes Sample Event 10 1 11 1994 MW 8 1 000E 03 1245 0 4 33E 02 3 786E 27 Yes Yes Sample Event 11 5 28 1996 MW 1 3 540E 01 1177 0 6 76E 02 9 695E 36 Yes Yes Sample Event 11 5 28 1996 MW 12 1 000E 03 1090 0 6 76E 02 9 824E 36 Yes Yes Sample Event 11 5 28 1996 MW 13 1 000E 03 1125 0 6 76E 02 9 216E 37 Yes Yes Sample Event 11 5 28 1996 MW 14 1 000E 03 1088 0 6 76E 02 1 125E 35 Yes Yes Sample Event 11 5 28 1996 MW 15 1 000E 03 1000 0 6 76E 02 4 317E 33 Yes Yes Sample Event 11 5 28 1996 MW 2 1 000E 03 1192 0 6 76E 02 9 933E 39 Yes Yes Sample Event 11 5 28 1996 MW 3 1 000E 02 1155 0 6 76E 02 1 212E 36 Yes Yes Sample Event 11 5 28 1996 MW 4 3 700E 02 1135 0 6 76E 02 1 734E 35 Yes Yes Sample Event 11 5 28 1996 MW 5 6 780E 01 1194 0 6 76E 02 5 883E 36 Yes Yes Sample Event 11 5 28 1996 MW 6 1 000E 03 1267 0 6 76E 02 6 233E 41 Yes Yes Sample Event 11 5 28 1996 MW 7 1 000E 03 1277 0 6 76E 02 3 170E 41 Yes Yes Sample Event 11 5 28 1996 MW 8 1 000E 03 1245 0 6 76E 02 2 759E 40 Yes Yes Sample Event 12 6 27 1997 MW 1 4 600E 02 1177 0 4 44E 02 9 666E 25 Yes Yes Sample Event 12 6 27 1997 MW 12 1 000E 03 1090 0 4 44E 02 9 970E 25 Yes Yes Sample Event 12 6 27 1997 MW 13 1 000E 03 1125 0 4 44E 02 2 110E 25 Yes Yes Sample Event 12 6 27 1
180. ____ SampleEvent Ej j fene ac E Analysis it will be shown in the gray Corfim or redefre textbox as a default value This angle is tege epe E 4 Select Plume Centerline Wells very important for the risk based ae mue analysis and errors in this value may cause erroneous results Distance from edge of tail to receptor feet from Site Details Distance from the most downgradient p gt well to compliance 1000 2 Distance to Receptor The distance in bound Mw feet from the most downgradient well to the compliance boundary delineated ES _ gt anaieis according to the nearest downgradient receptor The value shown in the gray textbox is the distance from plume tail to receptors provided in Site Details and is used as a reference only The compliance boundary can be at or upgradient of the nearest downgradient receptor See A ppendix A 6 and Figure A 6 4 for details about this parameter z 6 e N Q s o Z a Ed o 3 Select Sampling Events for Analysis Selects the starting and ending sampling events from the From and To dropdown lists respectively The user can choose to analyze one or more sampling events 4 Select Plume Centerline Wells Selects the representative wells along the plume centerline from source to tail Data from these wells will be used in the regression of plume centerline concentrations against the distance down the plume centerline
181. a Information screen Note If Edit individual Mea Dee well trends based on separate modeling studies is chosen the next screen will allow this data entry Help Provides information on the screen specific input requirements External Plume Information M odding Results allows the user to enter modeling results obtained by methods different from M ann K endall or Linear Regression Results of alternative statistical analyses can be entered by well and constituent Enter the results from modeling studies e g Monitoring and Remediation Optimization System MAROS E Api xl Increasing I Stable S etc in the blanks External Plume Information Modeling Results provided next to the well name To navigate Edessa Mg etse bil A ers beh the results for individual constituents click on et EE Y S CET le aa UO GIUM the tabs at the top of the screen If you would Arenas vital Cherie like to weight all chemicals the same choose BENZENE ETHYLBENZENE TOLUENE xYLENES TOTAL the button All Chemicals Otherwise enter Sowceal Modeling TE 1 the results for each COC and each well when you choose Individual Chemicals Ata later step in this program you will be able to weight these lines of evidence Note Increasing l Probably Increasing PI Stable S Probably Modeling results should be taken from fate Rea A TAE Aere A and transport models that take site specific Net gt gt Help data and
182. a from each sampling event can be reduced in Data Reduction Yearly averages are recommended if there are more than 4 years of data At least 4 data yearly averages or original data are required for the analysis Click on the option box to select the type of data you want to use 2 Select thetime period for evaluation Concentration data from an individual well over the time period specified will be used in the analysis Selecting a different time period may lead to different results From Selects the starting year from the dropdown list To Selects the ending year from the dropdown list Back Returns the user to the D ata Sufficiency Analysis M enu screen Analysis Calculates the cleanup status power and expected sample size for each individual well for each COC for the time period selected by the user The Individual Wal Cleanup Status Results screen will pop up Help Provides additional information on software operation and screen specific input requirements Version 2 1 October 2004 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Individual Well Cleanup Status Results This screen accessed from the Individual Wal Cleanup Status screen by clicking Analysis is used to display the results for individual well deanup status evaluations grouped by COC x Sample Size The number of data records amp Monitoring and Remediation Optimization System
183. a result of the limiting effects of biotransformation processes Thus the differences in apparent CVOC plume growth rates provides an independent line of evidence to support the conclusion that reductive dehalogenation influences plume length behavior at sites where vinyl chloride plumes are present References Aziz C E CJ Newell A P Smith C J Newell P A Haas and J Gonzales 2000 BIOCHLOR Database Natural Attenuaton Decision Support System Air Force Center for Environmental Excellence Brooks AFB Texas June 2000 www gsi net com Aziz C E CJ Newell J R Gonzales P Haas T P Clement and Y Sun 1999 BIOCHLOR Natural Attenuaton Decision Support System vers 1 0 User s Manual Air Force Center for Environmental Excellence Brooks AFB Texas www gsi net com Happel A M E H Beckenbach and R U Halden 1998 An Evaluation of MTBE Impacts to California Water Resources Lawrence Livermore National Laboratory University of California UCRL AR 130897 Livermore California June 11 1998 http www api org ehs mtbelink htm Newell C J and J A Connor 1998 Characteristics of Dissolved Hydrocarbon Plumes Results of Four Studies American Petroleum Institute Washington D C December 1998 ww gsi net com Newell C J L P Hopkins and P B Bedient A Hydrogeologic Database for Groundwater M odding Ground W ater September 1990 Mace R E R S Fisher D M Welch and S P Parra Extent M ass and Duration
184. ably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data lt 4 sampling events ND Non detect MAROS Version 2 2002 AFCEE 12 1 2003 Page 1 of 1 MAROS Mann Kendall Statistics Summary Project User Name Location Service Station State Texas Time Period 10 4 1988 to 12 19 1998 Consolidation Period No Time Consolidation Consolidation Type Median Duplicate Consolidation Average ND Values 1 2 Detection Limit J Flag Values Actual Value All Source Numberof Numberof Coefficient Mann Kendall Confidence Samples Concentration Well Tail Samples Detects of Variation Statistic in Trend ND Trend BENZENE MW 8 S 15 1 0 97 12 70 4 No S MW 7 S 15 1 0 24 8 63 3 No S MW 6 S 15 0 0 00 0 48 0 Yes S MW 5 S 15 15 0 67 55 99 7 No D MW 3 S 15 12 1 05 69 100 0 No D MW 2 S 15 7 3 52 27 89 9 No NT MW 1 S 15 15 0 92 90 100 0 No D MW 4 T 15 14 147 59 99 9 No D MW 15 T 15 0 0 00 0 48 0 Yes S MW 14 T 15 7 1 50 68 100 0 No D MW 13 T 15 10 1 03 62 99 9 No D MW 12 T 15 11 1 48 82 100 0 No D Note Increasing I Probably Increasing Pl Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events Source Tail S T The Number of Samples and Number of Detects shown above are post consolidation values MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 1 MAROS Spatial Momen
185. ach well Or choose to enter the overall saturated thickness of the aquifer by clicking on Uniform Saturated Thickness easy approach and then entering the overall saturated thickness in the blank provided Also enter the Groundwater Flow direction degrees away from the X axis and the approximate x and y coordinates of the source 1 Groundwater Flow Direction SU Direction from X axis counterclockwise 2 Porosity 3 Note See any groundwater textbook for suggested soil porosities 4 Aquifer Saturated Thickness Uniform Aquifer Saturated Thickness 1 ft C Variable Aquifer Saturated Thickness 3 Approximate Center of Contaminant Source Location Single Source Location X ft ft 12 12 Source Saturated Variable Source Location Mail Thickness ft Well Name s 10 Xsource Y source a Constituent ft M S 10 BENZENE 12 12 I s 10 ETHYLBENZENE 12 12 MwW 12 s 10 TOLUENE 12 12 MAC s 10 XYLENES TOTAL 12 12 I E MAL T a E lt lt Back Help Next Takes the user to the M oment A nalysis Statistics Screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 37 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Spatial Moment Analysis M oment Analysis Statistics accessed from the M oment Analysis Site D etails screen allows the user to view the Spatial Moment Analysis
186. ailed Water quality data are confusing If necessary resample the collection protocols A dapt the protocols to usually volatile chemicals are at well using improved procedures are the geologic conditions and lower concentrations than protocols and or more inadequate contaminants expected expected and other chemicalsare experienced personnel present when they were not projected especially in blanks 4 Sample Work closely with a reputable Documentation is poor duplicate If necessary resamplethe analysis laboratory to design an appropriate samples yield varied results well and have analyses procedures are analytical program laboratory blanks are severely conducted by a reputable inadequate or are contaminated spike recoveries are laboratory undocumented poor Adapted from Table 3 in Kufs 1994 Statistical Concerns Regarding False Positive and False N egative Rates As mentioned previously false positives and false negatives are the two types of errors existing in any statistical tests concerning the null hypothesis denoted as H o the alternative hypothesis is denoted by H 1 From the statistical definition false positive refers to the decision that the null hypothesis is rejected when in fact it is true false negative is failing to reject the null hypothesis when it does not hold Correspondingly the false positive rate type error rate a is the probability of incorrectly declining the null hypothesis and fal
187. ain M enu select MAROS Output by clicking the button next to the label This action will take the user to the M ARO S Reports G raphs screen Version 2 1 8 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 16 MAROS Reports G raphs Select the report or graph you would like to view then click on the button next to the list This action will take the user to the report or graph chosen To print select the print icon on the tool bar or select Print from the file menu Click Close to exit the Report What COCs should I choose for my site The MAROS Tool can help the use to choose the Constituents of Concern for your site Up to five COCs can be analyzed at one time by the MAROS software However the tool works best when one to three representative COCs are chosen To receive input from the software on how to rank or choose COCs 1 Follow directions for Importing Entering Data above 2 Main Menu From the M ain M enu select Site Details by clicking the button next to the label This action will take the user to the Site Information screen 3 Site Details In each screen select the information that describes the site click on Next to continue to the next screen First enter the site details on the Site Information screen N ext define sample events on the Sample Events screen Then select the representative wells in the Source and Tail zones on
188. ake decisions Comparisons are made in logarithmic scale Since the initial exceedence Xe is not confirmed by the verifying resample x which is within the prediction limit the conclusion of contamination cannot be drawn Version 2 1 A 7 15 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 8 Example dataset for constructing prediction limits Original Ln Mean Standard Prediction Original Verifying background transformed deviation limit exceedence resample data n 8 data x x s Xe X 2 339 0 850 12 606 9 107 1 435 0 361 3 071 1 122 5 146 1 638 Ln x Ln x 4 949 1 599 1 029 0 672 2 393 2 534 2 209 6 466 1 867 0 912 0 092 2 418 0 883 Assume there is only one exceedence in the 10 future measurements METHOD 3 CONFIDENCE LIMITS IN COMPLAINCE MONITORING Confidence limits are statistical estimates of the minimum or maximum population parameter e g mean concentration or both that will include the true parameter value with a specified level of confidence eg 99 confidence based on a sample of n measurements In groundwater compliance monitoring concern is with the lower confidence limit being exceeded by a predetermined standard such as the alternate concentration limit A CL If the lower confidence limit built from a compliance well exceeds the ACL it may indicate that the groundwater contamination is significant and correctiv
189. ake decisions If thelower confidence limit of any constituent in any compliance well exceeds the ACL then there is statistically significant evidence of contamination Otherwise the site is within compliance Gibbons 1994 argues that when the ACL is estimated from the background mean and since the compliance monitoring is conditional on prior demonstration of a significant increase over background the test is in fact a two sample t test instead of the above confidence limit method which is a one sample t test The dependence due to repeated comparisons of multiple compliance well means to a single pooled background mean should also be considered This suggests that a Dunnett type test should be used Readers can consult Gibbons 1994 and Dunnett 1955 for details Example Step 1 Example TOC concentration data Table A 7 9 in three compliance wells are used to construct lower confidence limits and the result is compared to the ACL that is estimated from background samples as 5 00 mg L Step 2 Calculate the mean and standard deviation of the concentrations for each monitoring well They are shown in Table A 7 9 Step 3 From EPA guidance EPA 1989 thet value for 5 degrees of freedom and significance level of 1 is 3 365 The lower 99 confidence limit for each monitoring well is computed and presented in Table A 7 9 For example the lower 99 confidence interval for MW 1is X lq 100557 J6 3 33 3 365 x0 70 2 45 2 36
190. akes the user to the Results of Information Weighting screen Help Provides information on the screen specific input requirements Results of Information Weighting accessed from the Statistical and Plume Information Summary by Well screen allows the user to view the weighted statistical modeling and empirical lines of evidence for each COC To navigate the results for individual constituents click on the tabs at the top of the Screen Back Returns the user to the Statistical and Plume Information Summary W eighting screen Next Takes the user to the Plume Information by Well Weighting screen E Monitoring and Remediation Optimization System maros I Results of information Weighting 051 x The results from the weighted statistical modeling or empirical lines of evidence for each COC are shown in the sheets below ETHYLBENZENE TOLUENE XYLENES TOTAL Help Provides information on the screen specific input requirements Version 2 1 October 2004 Well Name SIT Trend Result a M 15 s Li IMW 14 s D IMW 13 s D IMW 12 s D IMACT S D M 8 T PD MACT T PD zl 2 oil Note Increasing l Probably Increasing Pl Stable S Probably Decreasing PD lt Decreasing D No Trend NT Not Applicable N A Source Tail S T e Pd Z lt lt Back Next gt gt Help 48 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWA
191. alysis for Individual Well Cleanup Status Before testing the cleanup status for individual wells one important issue must be considered the stability or trend of the contaminant plume Only after the plume has reached or is reaching steady state can we conduct a test to examine the cleanup status of wells Applying the analysis to wells in an expanding plume may cause incorrect conclusions and is less meaningful Although in long term monitoring the site may require many years to attain site cleanup individual wells become dean gradually beginning with the tail wells and followed by the source wells If we can show that the average concentration in a well is below the cleanup level with statistical significance we can eliminate it from the monitoring network or at least reduce its frequency of sampling If the average concentration is lower than the cleanup level but is not significant we can find out by power analysis how many more samples need to be collected to confirm the cleanup with data variability unchanged For cleanup status evaluation a modified sequential t test for assessing attainment of cleanup standards based on the mean contaminant levels is adopted U S EPA 1992 The test procedures involve several steps First two statistics 6 and t need to be calculated based on the yearly averages i e the annual mean concentrations When calculating 5 and t the untransformed yearly averages are used if they follow normal distribution
192. amples for a quarterly sampling while a three year period is needed to generate six samples with a semiannual sampling The analysis will still proceed with less than six samples but the recommended results may be inaccurate Also six sampling events do not necessarily lead to six samples because sampling could be skipped at certain events for some wells amp 3 Monitoring and Remediation Optimization System MAROS Sampling Frequency Analysis Determine the sampling frequency for sampling locations by the Modified CES method which is based on the Isl from Lawrence Livermore National Lab Click to for Select Sample Event 15 confirm Define the recent period Confirm by selecting the begi inning and ending sampling event Sample Event 10 Event 10 Sample Event 19 Rate of Change parameters Options o sions lt lt Back Analysis Help SAMPLING OPTIMIZ In this example select Sample Event 10 from the From dropdown list and Sample Event 15 from the To dropdown list Click the Confirm button to confirm the selection Notice that the Analysis gt gt button is now activated 9 View or modify the Rate of Change ROC 1 parameters by selecti ng O pti ons The Sampling Frequency Analysis Options S am pl i n g F requ en Cy A n al ysi S e O pti on S screen Classify the rate of change for a COC
193. ampling data and more complex geology For instance if a site has co mingled plumes typically the plume networks should be analyzed separately for the different constituents If the site has more than one aquifer affected by contaminants the well networks for each should be analyzed separately In general the MAROS method applies to 2 D aquifers that have relatively simple site hydrogeology However for a multi layered 3 D system the user should apply the statistical analysis layer by layer Version 2 1 A 11 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails The primary constituent of concern at the Service Station site is benzene which is analyzed at 12 monitoring wells in the Upper Aquifer well network Figure A 11 1 The site has 10 years of approximately semi annual sampling data All monitoring wells have been sampled semi annually in the Upper Aquifer for BTEX since the implementation of the original long term monitoring plan The historical benzene data for all or in some cases a subset of wells were analyzed using the MAROS 2 1 software in order to 1 assess plume stability and 2 recommend changes in sampling frequency and sampling locations without compromising the effectiveness of the long term monitoring network ww Source Area wwe SUNVILLE STREET Mw A au EE EST FIGURE A 11 1 EXAMPLE SITE SERVICE STATION MONITORIN
194. analysis the reults could be a little different To better understand the influence of parameters on optimization results the user can try several runs with different parameters each time amp Monitoring and Remediation Optimization System MAROS 6 The Wel Sufficiency Analysis New Locations screen allows the user to perform a sufficiency analysis to determine potential new sampling locations This analysis utilizes the SF values obtained from the Well Redundancy analysis to predict the concentration estimation uncertainty at unsampled regions The regions where uncertainty is high are the Well Sufficiency Analysis New Locations represented by Slope or this analysis An coc BENZENE A E nalysis Reset SAMPLING OPTIMIZATION potentail locations for adding new ws a ee Uem sampling points vs no so o2 4 Back Next gt gt Help Only one COC can be analyzed each time using this analysis For example if there are three COCs this process has to be repeated three times The Selected column shows the status of whether a well is used in the analysis The user can exclude wells from analysis by unchecking the checkbox following a well In this example all wells are used in the analysis The Reset button can be used to reselect all wells Select the COC from the COC dropdown list In this example select benzene Then dick the Analysis button to p
195. and High thresholds respectively Version 2 1 A 11 62 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MAROS Output At this point in the software the user has gone through all of the optimization utilities and can begin to consolidate the knowledge of the site with the MAROS analysis results to make a final determination of the site optimization The goal of the tutorial is to show the user tips and pitfalls when applying MAROS at a typical site The tutorial example has been used only to illustrate the utilities of the MAROS software and it is by no means a complete site analysis Step 5 MAROS Output Reports Graphs allows the user to view print reports and graphs from the site trend analyses as well as a preliminary Site Recommendation Report This allows production of standard Reports induding the one page heuristic approach to sampling optimization based on plume stability and site parameters with results for sampling frequency duration and density Sample Reports are located in A ppendix A 10 The reports can be used to assess the project objectives defined at the start of the tutorial The MAROS output results should also be reviewed before proceeding to optimization of the monitoring network to ensure that the trends in the data are fully understood Spend time reviewing the data and trend results both spatially and temporally Try to identify any sp
196. and mean groundwater velodity at each site Large daughter product plumes do not commonly extend a large distance downgradient of the parent product plumes PLUME LENGTH CORRELATION GRAPHS USED FOR STEP 3 AFCEE Study Aziz et al 2000 also evaluated correlations to chlorinated solvent plume lengths In general the best correlation to log plume length in ft was log Plume Width x Maximum Concentrations as shown in Figure A 4 5 Title logPCElogPWPCE_Orig eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers Title logTCElogPWTCE eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers Title logDCElogPWDCE eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers Title logVClogPWVC eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers FIGURE A 4 5 CORRELATION OF LOG PLUME LENGTH WITH LOG PLUME WIDTH X MAXIMUM CONCENTRATION
197. ariance of distribution varies with changing units of measurement Therefore we suggest that the Poisson prediction limits should be used only for counts of analytical hits usually for VOCs In the case of 100 nondetects i e detection frequency equals zero one can also use the laboratory specific quantitation limit or limits required by the applicable regulatory agency ASTM 1998 as the nonparametric prediction limits In this case one should question whether the constituent is a useful indicator of contamination and if not statistical testing of the constituent should not be performed Statistical independence of data is still the underlying assumption of all of the above suggested procedures Note that the above suggested percentages are not hard and fast rules and should be based on judgement EPA 2000 Version 2 1 A 7 33 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Procedures Cohen s Adjustment Step 1 Let n be the total number of measurements and denote them as xi i 1 2 n among which m measurements are above detection limit DL Thus there are n m measurements that are below the DL nondetects Step 2 Compute the sample mean from the data above the DL as 1 m xX X m iz Step 3 Compute the sample variance from the data above the DL as m 1 Step 4 Compute two parameters h and y as n m s gt and y x
198. ased on two large databases of historical plume data were considered when evaluating the minimum well density reflecting both BTEX and chlorinated solvent plume information Mace 1997 and McNab 1999 Mace 1997 used data from 138 BTEX plumes while McNab 1999 presented data from 37 the chlorinated solvent plumes These data were combined plotted and then used to develop the following equation sampling density number of wells 1 5 plumelength where plume length is in units of feet and the sampling density is the number of wells for the entire plume In other words this equation indicates the monitoring well density actually in use at the sites in the database and is based on plumes of different sizes roughly 50 ft to 5000 ft MAROS uses this equation to indicate a well density that is typical at many sites Based on recommendations developed by ASTM 1998 a minimum of four wells is specified for all plumes User should also consider the well density in light of adequately defining characterizing the plume through gathering sufficient site information Current Site Treatment Sites currently undergoing site treatment i e pump and treat system etc have separate site suggestions for sampling frequency duration and density applied FREQUENCY No recommendation is given for the sampling frequency at a sitethat is currently undergoing remediation DURATION MAROS uses a simple decision matrix to assess when the design
199. ata collected so far as x xx k21 2 m Dy j l where x is the jth measurement at year k Step 4 Compute the mean x and variance s of the yearly averages as 5 m 1 Therestrictions of using yearly averages and at least four samples a year can be eased as long as there are no seasonal effects and no significant serial correlation between samples For example this test can be used for cases in which there are only two samples per year or there are only a series of annual or biennial samples Step 5 Calculatethet and 6 for the likelihood ratio as Cs ga STH t 2 Version 2 1 A 7 19 Air Force Center for October 2004 m Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE g u S 2 S x m Step 6 Calculatethe likelihood ratio as incest m m m l t Step 7 Compare LR with parameters A and B to make decisions If LR lt A conclude that the groundwater at this well or site does not attain the cleanup standard Reconsider treatment effectiveness If LR gt B conclude that the mean groundwater concentration in this well is less than the cleanup standard If the yearly averages in this well do not show a statistically significant increasing trend conclude that the groundwater at this well attains the cleanup standard Otherwise conclude that the groundwater at this well does not attain the cleanup standard and reconsider treatment effectiven
200. ater than the sample size more samples are needed to confirm the cleanup status Risk Based Power Analysis for Site Cleanup Evaluation The use of risk based goals in managing contaminated sites requires that cleanup standards be met at the compliance boundary In order to perform a sufficiency analysis at the compliance boundary a strategy was developed as follows First select monitoring wells along the plume centerline and regress concentrations from these centerline wells against their distances down the plume centerline with an exponential model Second for each monitoring well project its concentration to the compliance boundary using the exponential model with its distance to the compliance boundary Third these projected concentrations at the compliance boundary constitute a group of estimated concentrations that can be evaluated by statistical power analysis The result from this type of power analysis provides a statistical interpretation of whether the risk based site cleanup goal has been met The exponential regression model is y A EXP Bx Equation A 6 13 where A and B are regression coefficients x is the distance from a plume centerline well to the plume source and y is the concentration at this well This regression follows the concept of bulk attenuation rate in natural attenuation which assumes that the spatial change in plume concentrations can be modeled as exponentially decaying with distance downgradient from the s
201. atistic i e large magnitudes indicate a strong trend Confidence in Trend The Confidence in Trend is the statistical confidence that the distance to the from the source to the center of mass is increasing S20 or decreasing S 0 COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average Values near 1 00 indicate that the data form a relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about the mean View Report To print the First Moment Distance from Source to Center of Mass Report and analysis results dick View Report to proceed Back Returns the user to the Z eroth M oment Plot screen Next Takes the user to the First M oment Plot Changein Location of M ass O ver Time screen Help Provides information on the screen specific input requirements N ote The information displayed in this screen can also be viewed in report form First Moment Report from the M AROS Output Screen or by clicking on View Report see Appendix A 10 for an example report For further details on the Mann Kendall Analysis Method or Moment Analysis see Appendix A 2 and A 5 respectively Version 2 1 40 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATIO
202. aunay method is also introduced M ethod Description The Delaunay method is developed based on Delaunay triangulation which is the triangulation of a point set with the property that no point in the point set falls in the interior of the circumcircle of any triangle in the triangulation As seen in Figure A 3 1 all nodes potential well locations are joined by the blue lines which form the edges of Delaunay triangles The yellow lines form many polygons called Thiessen polygons or Voronoi diagrams which are the dual parts of Delaunay triangles 18 00 16 00 14 00 Yellow 12 00 d lines ANLE RIS zi i 10 00 i i 8 00 6 00 Blue 4 00 lines 2 00 0 00 0 00 2 00 4 00 6 00 8 00 10 00 12 00 14 00 16 00 18 00 20 00 Version 2 1 A 4 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Figure A 3 1 Illustration of the Delaunay Triangulation Delaunay triangles and Voronoi diagrams have been widely used for centuries for solving spatial distribution problems Okabe et al 1992 Watson 1994 In MAROS Delaunay triangulation is first used to generate a grid for the studied site with potential sampling locations as its nodes Then based on the formation of Delaunay triangles and Voronoi diagrams spatial analyses are made to determine the relative importance of each sampling location Finally spatial redundant locations are eliminated from the monitoring network
203. aunay2k xls 4 Trend Visualization Excel File xlsLOEresults xls 5 Location Addition Excel File xlsLocation xls Version 2 1 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 6 MAROS Manual afcee MAROS Manual pdf 7 MAROS Tutorial File TutorialExampleData xl s To start the software after installation double click on the afcee MAROS mdb file or open the file from within Access 20006 j Note Although some users are likely to have the complete atte xi se of libraries turned on to run the program the f eees following procedure should be applied the first time the Smoot feces o0 cocti cel M Microsoft DAO 3 6 Object Library software is used M Microsoft Graph 9 0 Object Library Browse ENS Excel 9 0 gu gis 4 u n LI licrosoft Visual Basic For lications Extensibili K 1 Start up main software afcee MAROS mdb The zmessoesoouei ey ty ues Start up screen will appear Press F11 on the keyboard H arabe COH concret DO Tyre Lbrary gt 1 VideoSoft vSFlexGrid 7 0 DAO RDO L 1 VideoSoft 5FlexGrid 7 0 Light 1i VideoSoft VSFlexGrid 7 0 OLEDB 4 2 TheM ain Access Program will appear Click on the tab P M odules Open the Module A MAROS Initial Start oue automation U p Referen ces A Location C WINNT System32 stdole2 tlb Language Standard 3 Go to the
204. b two methods for assessing the attainment of cleanup standards were given a fixed sample size test based on a confidence limit and a sequential t test using a likelihood ratio Both can be used to determine whether 1 the mean concentration is below the cleanup standard and 2 a selected percentile of all samples is below the cleanup standard The sequential t test will be presented here as it has the following advantages e Thenumber of samples required to reach a decision need not be known at the beginning of the sampling period In afixed sample size test the number of samples required to reach a decision should be determined in advance based on specified false positive and false negative rates e g 0 0 05 B 0 20 This is to ensure for a known or presumed degree of uncertainty in the sample population that the statistical test with the number of samples that will be collected will provide enough power 1 8 to detect the expected difference between the cleanup goal and the cleanup standard e On average and under the same levels of false positive and false negative rates the sequential t test will require fewer samples and therefore a shorter time to make the attainment decision than the fixed sample size test This method can be used to test wells individually or in a group and requires at least three years of data Yearly averages of samples are used in the sequential t test in order to reduce the effects of any serial corr
205. ber of data yearly Individual Well Cleanup Status Optional Power Analysis averages or ori gi nal d ata that is used in The screen shows the power analysis results of whether the mean concentration of a well is significantly lower 1 than the cleanup goal based on the Students t test on mean difference Sample size is the number of the eval uati on a concentration data in the time period selected by the user Power of Test and Expected Sample Size associated with the t test are also given The data are assumed to be either normally or lognomall distributed Feauke shown am based on oath averaces YES Significantly Cleanup Goal Indicates BENZENE ETHYLBENZENE TOLUENE xyLENES TOTAL whether the mean concentration of a Welwame Sample Sinicaniy Power Expected Disbuion wel is significantly lower than the NE a NENCAN 550500 deanup goal Results could be YES Ln g C E significantly lower than the deanup z Mia 7 NO 8138 500 MERO goal NO not significantly lower or ino x d z higher than the deanup goal or N C x Ma g ves mo s View not conducted due to insufficient data e Md 3 YES 0 996 4 i Log 9 ns Z N C not conducted due to insufficient data S E sample mean significantly exceeds cleanup goal Power of Test The probability that the a EE A MM Se P correct conclusion can be made when the seme ps average concentration from a well is truly lower than the cleanup goal
206. both Modeling and Empirical results click Next to continue to the next screen Results for modeling studies are entered on the External Plume Information M odeling Results screen Next results of any empirical evidence are entered on the External Plume Information Empirical Results screen To proceed dick Next The External Plume Information portion of the software is complete 13 Plume Analysis Menu From the Plume Analysis M enu select MAROS Analysis by dicking the button next to the label This action will take the user to the Lines of Evidence Summary by Well screen 14 MAROS Analysis In each screen select to weight the Lines of Evidence or individual wells as pertains to your site click Next to continue to the next screen Results for all lines of evidence are summarized on the Lines of Evidence Summary by Wall screen Next the choice to weight the Lines of Evidence by All Chemicals or Individual Chemicals is made on the Trend Summary Weighting screen Continue to the Results of Trend Weighting screen to view the results in table form Finally the option to weight individual wells is available on the Lines of Evidence by Well Weighting screen The M onitoring System Category screen shows a summary of the source and tail well results for the COCs chosen the Monitoring System Category is displayed for these results To proceed dick Next The Trend Analysis portion of the software is complete 15 Main Menu From the M
207. by increasing monitoring which can be expressed as cost The additional cost of lowering false positive rates comes from taking additional samples and using more precise analytical Version 2 1 October 2004 A 7 6 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE protocols Lowering false negative rates and requiring a simultaneous reduction of false positive rate usually can only be achieved by increasing sample size Therefore if a sampling strategy is sufficiently sensitive to detecting changes in contaminant concentrations at regulatory levels its false positive and false negative rates should be acceptable and need not be further reduced Table A 7 4 Two types of error in detection monitoring True condition in Decision based on a statistical sample the well Ho Site Not Contaminated H Site Contaminated Not Correct Conclusion False Positive Rate Contaminated Probability 1 a Probability False Negative Rate Correct Conclusion power Probability D Probability 1 D Contaminated The type of error that may cause facility wide problems Table A 7 5 Two types of error in corrective action monitoring Decision based on a statistical sample True condition in the well Ho Contaminated Does not H Clean Attains the attain the cleanup standard cleanup standards False Negative Rate Correct Conclusion power Probability
208. cal and tabular format This report can also be printed Close the report by dicking on the red button in thetop right hand corner of the screen The Reduced D ata Plot screen will return Select the Next button to proceed The Data Reduction Complete screen will appear 7 The Data Reduction Complete screen indicates that the data has been reduced according to the parameters entered The user may now proceed to the Statistical Plume Analysis and analyze the trends in the groundwater data E Mantloring and Rewediaties Optimization System MAROS Select the Trends Analysis button to return to the M ain M enu Data Consolidation Complete jur rb het Deer mood scooting noite pen en pou eese thas wat vov procured ta Step J b Statin Pham Aripa is anaia thee tedio s goateder dais Click hereto proceed Continun fo Seg gt gt TREND ANAL Y 5 5 Version 2 1 A 11 22 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Version 2 1 A 11 23 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 3 STATISTICAL PLUME ANALYSIS The Statistical Plume Analysis option allows the user to perform Mann Kendall Analysis and Linear Regression Analysis 1 From the Plume Analysis M enu select the Step 3b Statistical Plume Analysis option
209. ch COC Whereis the approximate center of mass and is it moving over time Arethere redundant wells in the current monitoring network What is the suggested future sampling frequency Do new wells need to be added to the monitoring network to adequately characterize the plume TheMAROS software can be utilized in a step by step fashion with each progressive step along the way yielding information that can be applied to answering site specific compliance monitoring questions At each phase in the software results that are presented are based on increasingly more consolidated data These data consolidation steps will lead to more stringent assumptions being used in order to reach a result or site specific results Figure 1 The assumptions you make along the way will affect the outcome of the software tool results However because the assumptions are arranged in a logical explicit fashion they can be reviewed and altered should more site data become available Also the validity of the results or recommendation will rely on the extent and quality of input data The data imported into the software must meet minimum data requirements as to the frequency of sampling duration of the sampling intervals for trend analysis and sampling density for the site as well as the quality of the measurements decreased amount of false positives negatives Version 2 1 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIA
210. coefficient is positive the user should go back to check if the flow angle or selected wells are correct Confidence in Coefficient The statistical confidence that the estimated coefficient is different from 0 Refer to Confidence in Trend in Linear Regression Analysis for details Back Returns the user to the Parameters for Risk Based Power Analysis screen View Report Generates a report with selected parameters and regression results for each COC The user can go back to re run the regression by selecting a different set of parameters Next Proceeds to the Centerline R egression Projected Concentrations screen Help Provides information on the screen specific input requirements Version 2 1 81 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Centerline Regression Projected Concentrations This screen accessed from the Plume Centerline Regression Results screen by clicking N ext is used to display the projected concentrations calculated using regression coefficients obtained in the previous screen Refer to Appendix A 6 for details Note projected concentrations are NOT calculated for sampling events with lt 3 wells user p rov i d ed d etecti on l i mi t for th e COC If true a check mark is shown in lt lt Back Select Wells View Report Analysis gt gt Help the ch eckbox Use in Analysis I
211. concentrations projected to the compliance boundary delineated based on the nearest downgradient receptor Concentrations from wells in a sampling event are used as a group in this analysis Refer to Appendix A 6 for details Back Returns the user to the Sampling Optimization screen O ptions Shows the Data Sufficiency Analysis Options screen where the parameters for the two types of analyses are defined Help Provides information on the screen specific input requirements Steps for use 1 Check data sufficiency analysis parameters by dicking the Options button The user can choose to use the default values or specify new values for the parameters Missing or invalidated values of certain parameters may prevent the analysis from proceeding 2 Sincethetwo analyses are independent from each other the user can choose to perform any analysis first Version 2 1 73 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D ata Sufficiency Analysis Options This screen accessed from the D ata Sufficiency Analysis M enu screen by clicking Options is used for setting the parameters required in the statistical power analysis Monitoring and Remediation Optimization System MAROS E These parameters include Cleanup Goal the PRG Data Sufficiency Analysis Options mg L Target Lea mg L Alpha Leva the Define the Target Level used in the individual well celan
212. ct Sampling Optimization from the Main Menu and the Sampling Optimization M enu screen will appear The Sampling Optimization Menu screen serves at the center of the sampling optimization user interface The user can choose to perform either sampling location analysis or sampling frequency analysis first Data sufficiency analysis will become available after sampling frequency analysis is completed tos INCEMONITORING OPTION 1 SAMPLING LOCATION ANALYSIS Vulg vid Berpeshel bow C pieno eth Vf Aas ee eaea Proses emt Suy Oto Managemare Mer tn pm nol ami PIRE es hem munt rt fury ert mor d om tte Ste Detales e apt de Qr Aet a o ees m c nees Phime Anateis Fetua bes Coniabenstos Coase at Pues erwies Taba orent Select Sampling Location Analysis from 3 Monitoring and Remediation Optimization System MAROS xi the Sampling Optimization M enu Sampling Optimization Menu The Sampling Optimization Menuisthemsnme Click hereto _ fucina samping The Wal Redundancy Analysis D daunay eraen tomtene opea A proceed M ethod screen will appear Select One Option Option 3 can only be selected after running Option 2 1 Select the sampling events for analysis a l i Ru dem from the From and To dropdown E Delaunay method andor addon of new sine lists The From sampling event must Eis Opion2 Sampling Frequency Analysis not be later than the To sampling E LN LEM event a Opti
213. ctober 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE For example if your plume is much shorter than predicted plume length then there may be secondary evidence that your plume has a higher potential to expand You should select ncreasing or Probably Increasing and enter in software On the other hand if your plume is much longer than the predicted plume length there may be secondary evidence that your plume has a lower potential to expand You should select Decreasing or Probably Decreasing and enter in software Correlation Equations for BTEX Plumes Some correlation equations for BTEX plumes are provided in Wiedemeier et al 1999 see page 229 230 A more detailed correlation analysis was performed for the American Petroleum Study by Nevin e al analyzed plume length data UST and petroleum release sites taken from the four sources the HGDB Air Force plumes the Texas BEG study and Florida RBCA study The database includes sites ranging from small retail gas stations to large distribution sites covering thousands of square feet This wide range of site sizes makes the study database different from the databases used in the Lawrence Livermore LLNL see Rice et al 1995 and Texas BEG see Mace et al 1997 studies which were almost entirely retail sites Using this database correlations were performed on a number of hydrogeologic and source parameters The correlation resu
214. d As individual steps ofthe process are e Step 1 Data Management leted options to select b n lab completed options to select become successively available Step 2 Site Details Proceed Through Steps 1 5 Step 3 PI ume A nal ysi S Step 1 Data Management e Step 4 Sampling Optimization ERPIMS archivi fae aoa ae Sa IMS files archiving current site e Step 5 M A ROS O utput o Step2 Site Details zt letails of the sit luding hydt l ters tail H g weldesinalinendconsitusiscteancen Select the desired option by dicking the Step 3 Plume Analysis A 5 Perform Data Consolidation Statistical Trend Analysis Spatial Moment ap pl l cabl e bu tton s P roceed th rou gh e Analysis and Enter External Plume Information E Optional Step4 L Sampling Optimization Steps 1 5 lu D ata M anagement Allows data import of Excel and ERPIMS files archiving current site data and manual data addition Site D etails Initial definition of site specific data including choosing the Source and Tail wells sample events and providing site specific Constituents of Concern COC s PlumeAnalysis Allowsthe user to perform data reduction as well astrend analysis through both Statistical Plume Analysis Spatial Moment Analysis and External Plume Information Also allows the user to apply final Analysis Consolidation to the trend results Sampling O ptimization Allows the user to perform sampling optimization through various
215. d Stable Probably Decreasing Decreasing or Not Applicable Insufficient Data Other statistics displayed indude the Mann Kendall Statistic S the Confidence in Trend and the Coefficient of Variation COV Refer to Appendix A 1 and A 5 for further details Second M oment analysis results showing Lesser kepet lieira ter ant p tw rataa o the spread of the plume over time are displayed for benzene The scale shown is logarithmic Log y o Version 2 1 A11 33 e October 2004 Z amp AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Change graph type to Linear by clicking on the open circle next to Linear Click on Graph to display The graph displays covariance type Sxx Second Moment Plot Clan Grim Visi pot Ti representing spreading of the plume in the STE Sms e punt Me canet meray hne mit pin Pon nee ne direction of groundwater flow The Second Moment Trend of the spread of the plume over time is shown to be probably increasing PI This indicates that although the concentrations are decreasing the plume Hume eet is spreading over time Frenne J To view the results for covariance type Syy dick on the cirde labeled Syy under RETA oUm Covariance Type Sys meno View Fart Then dick on the Graph button The graph now displays covariance type B mm Syy representing spreading of the Sieond Moment Plot plume in the direction perpendicular t
216. d The Excel M odule button will be activated only if the sampling events in both From and To dropdown lists are the same 3 Click either A ccess M oduleor Excel M odule if activated to proceed Version 2 1 52 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Access M odule Potential Locations Setup This screen accessed from the Well Redundancy Analysis D daunay M ethod screen by clicking A ccess M odule is used to set up the properties of potential locations and the options used in the Delaunay method amp Monitoring and Remediation Optimization System MAROS E nl xl Sel ected Deci des Ww hether or not a Access Module Potential Locations Setup location is induded in the analysi S Sampling locations will be determined from the following potential sampling locations These potential C heck the button to incl ude or uncheck peek ao aet some ICSF 1o aire eiecit dde P UPOS the button to remove this location from BENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL the list of potential locations LociD ESCoord NSCoord Selected Removable zl Removable Decides Ww hether or not a MW 1 130 200 location is allowed to be eliminated by the optimizing process if it is considered MW 12 100 0 8 0 MN 13 550 230 as potential locations The Selected status will be set to True for all locations for the selected COC
217. d Eliminated jo we 130 200 m MW 12 100 0 80 x The Acces Module All in one wis so a J MW 14 1020 200 Ll Results screen shows the MWS 190 0 125 0 m optimization results for each s NN 2 a 550 370 4 E MWS 40 70 0 g Eo MN 6 770 50 Ll zl oO Eliminated whether or not the well is abandoned from the monitoring zZz network as a redundant well a 2 2 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE well after considering all COCs In this step a well is eliminated only if it is elimnated for all COCs If a well can be eliminated for all COCs then this well can either be abandoned or terminated for sampling Since in this example only benzene is analyzed the results at this step is the same as the results from the previous step A summary report is available for review by clicking the View Report button Click Next gt gt to proceed and the Wal Sufficiency Analysis New Locations screen will appear Note The decision to terminate sampling for one COC or all COCs at a well may require further considerations more than just recommendations from the above described optimization The above recommendations are based solely on the statistical point of view In practice decisions are always made out of the scope of technical considerations Regulatory considerations for example need to be incoprated into the decision process Also if some of the parameters were changed in the above
218. d Linear Regression Analysis should have Medium weighting This means that weighting will not be applied bi Pod tad eel tien Dre Om Feed rhon aue Thea fend tant quee won emt lit feee Den We Worried tios wan sf ten boum Fe wes ww fj i j en ww Wtteet neue ej Mr temen r To proceed click N ext to see results of weighting Click hereto proceed Note If more than one COC was being used the user could choose to weight the trend methods applied to each COC individually select Individual Chemicals or to weight all chemicals select All Chemicals 4 Results of Information Weighting allows the user to view the weighted statistical modeling and empirical lines of evidencefor each COC Trend results for benzene are displayed for each well E Men Puede aed Bonet Npe oen Lysis Results of Information Weighting Mec primm te omi iare oso wma iem cia eim n aari feit babes Select Next to proceed HOSE Weetteem t Targ Tei mere Taw d E rbd emm ma Jaano EY Ho T peel BIT Aa guia DY Coad ee ANALYSIS Click here to proceed Version 2 1 A 11 39 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Click hereto proceed Ee Note If more than one COC was being used the user would navigate the results for individual constituents by clicking on the tabs at the top o
219. d Newell C J Groundwater Services Inc This appendix details the moment analysis procedures employed by the Monitoring and Remediation Optimization System MAROS Software The procedures outlined below were developed to assess plume stability for groundwater plumes based on scientifically sound quantitative analyses of current and historical site groundwater conditions The moment analysis results can also be used to further assess possible information loss due to eliminating sample locations in the long term monitoring network Plume Stability Analysis Confirmation of the effective performance of monitored natural attenuation as a stand alone remedial measure requires the demonstration of actual measurement of stable or shrinking plume conditions based on evaluation of historical groundwater monitoring data For this analysis an overall plume condition was determined for each COC based on a statistical trend analysis of moments for each sample event as described below The function that describes residence time of mass in a field is difficult to characterize exactly An infinite set of parameters are needed to fully characterize the distribution and the mean residence time and variance are often inadequate as well It is more convenient to characterize the approximate distribution rather than the exact distribution in terms of the moments Rasmuson 1985 The moment calculations can predict how the plume will change in the future if further
220. d at the end of the Site D etails section of the software The archive file will contain the site details such as seepage velocity and source and tail well designations Archive files are in Access format mdb and should be named to distinguish them from MAROS Output files Version 2 1 18 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE M anual Data A ddition M anual Record A ddition accessed from the D ata M anagement M enu Screen can be used to add individual Records to the database E Monitoring and Remediation Optimization System MAROS x Steps for use Do you want to replace existing data in the database or enter additional data to the dataset already in the software 1 Choose to a Replace Data or A ppend Data d aa MEET to the groundwater data already in the software 2 Fill in the appropriate information within each field Fields such as Constituent Type and Constituent have dropdown boxes to assist in data entry Choose Constituent Type before choosing the Constituent Note If the result is ND then fill in the D etection Limit 3 Review information before adding the record When all the data is entered click on the Add Record button Add Record To add a new record choose the iprsemm L entries from the selection boxes or type in the MANUALDATA ADDITION record information welNm
221. d sampling events The average SF value is used to determine the overall redundancy of a wel The smaller it is the more redundant the well is Access Module Slope Factor Values events are shown below for each potential sampling lol tions with smaller SF values may be eliminated in a later step BENZENE ESCoord NSCoord Avg SF Min SF 13 0 20 0 0 259 0 039 MW 12 1000 80 0165 0 000 0 262 MW 13 650 230 0 254 0 000 0 395 MW 14 1020 200 0 064 0 000 0278 MW 15 190 0 125 0 0 421 0197 0538 M2 20 300 MNS 35 0 100 Click to proceed MNA 550 37 0 Avg SF Slope Factor value averaged across sample events chosgf earlier MWS 4 0 70 0 Min SF and Max SF minimum and maximum Slope Factor valfies Note that there are several wells with Optimize by CoC gt gt em average SF values less than 0 2 Click Optimize by COC gt gt to proceed The Access M odule Results by COC screen will appear SAMPLING OPTIMIZATION The Access M odule Results by COC i xl screen shows the optimization results for each COC Wells that are Sacra clone foreach COC we determines as shown the flog lta These diriden P P A d E samling locations marked as Eliminated are eliminated from the monitoring network Eliminated el imin ated are i d enti fi ed Ww ith a ch eck statifs can be interpreted here as stopping sampling for a certain COC at a certain sampling
222. d the data first the method described in M ETHOD 8 beforeusing the following procedures Procedures Seasonal Effects Only Step 1 Consider a time series of N observations that exhibit m seasonal patterns The jth j 1 2 m seasonal average is _ 1 28 x x wheren is the number of non missing observations for season j n k l Step 2 Calculate the sample residuals after correcting for the seasonal means as eg Re ey Step 3 Compute the mean square error as m j 2 e 2 j l k l N m Step 4 The standard error of the mean is 2 Se and the Df associated with itis N m s x Procedures Serial Correlation Only Step 1 Consider a time series of N observations that exhibit serial correlation but no seasonal effects The observed serial correlation coefficient is p which has been proved as statistically significant using the Durbin Watson test presented in METHOD 8 Step 2 Assume the variance estimated from this time series is s2 The standard error of the mean when N is large is approximately 2 ge un and the Df associated with it is approximately yal rounded to the i 3 nearest smaller integer Procedures Seasonal Effects and Serial Correlation Step 1 Consider a time series of N observations that exhibit m seasonal patterns and serial correlation The mean square error calculated from this set of data is s which is estimated by using Procedures Seasonal Eff
223. d the location of potential receptors the software suggests an optimal plan along with an analysis of individual monitoring wells for the current monitoring system The software uses both statistical plume analyses parametric and nonparametric trend analysis developed by Groundwater Services Inc as well as allowing users to enter External Plume Information empirical or modeling results for the site These analyses allow recommendations as to future sampling frequency location and density in order to optimize the current site monitoring network while maintaining while maintaining adequate delineation of the plume as well as knowledge of the plume state over time in order to meet future compliance monitoring goals for their specific site This User s Guide will walk the user through several typical uses of the software as well as provide screen by screen detailed instructions INTENDED USES FOR THE MAROS SOFTWARE The MAROS software tool is designed to analyze data from a mature site investigation specifically a groundwater plume that has been delineated and monitored for more than four sample events Along with the guidance found in the Long Term Monitoring Optimization Guide AFCEE 1997 you can use the software to answer important compliance monitoring data questions What COCs are identified at the site Is the temporal trend in the groundwater site analytical data significant What is the spatial distribution of the temporal trends for ea
224. d to other analyses f Herr cong atte Scnmikatian Dyrtiesta abun Vete PII a Back Returns to the Risk Based Power Analysis Results Risk Based Power Analysis Complete screen The user can go back to re run the analysis by JA E selecting a different set of parameters N Faia ied Pinas Aurei aaah COD for Pe An procent b fs rhe ndy ot E j aa char niri Data Sufficiency Analysis Menu Returns the user MUT erre to the Data Sufficiency Analysis M enu screen g meee a z lt Back Dala Sethcicncy Analysts Meca Q a Version 2 1 86 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE M AROS Output Reports Graphs MAROS Output Reports Graphs accessed from the Main Menu screen allows the user to view print reports and graphs from the site trend analyses as well as a preliminary Site Recommendation Report Sample Reports are located in Appendix A 10 topi ee prin pemen 9 MAROS Output Reports Graphs View Print Report To view print reports choose Vei sti u ial maa ae the report of interest and dick View Print Report Wee Mapon View Print G raph To view print a graph choose the graph of interest and click View Print Graph Export MAROS Analysis Results Results erre e can be exported to a Microsoft Access database The user can then use the results rU d displayed in tables to plot data in
225. dence e The application of the empirical data is subjective and controlled by the user i e MAROS does not take data compare to the empirical data and make a conclusion e Touseempirical data as a secondary line of evidence the user Version 2 1 A 4 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE i ii iii reviews the empirical data in this appendix based on the user s judgement assigns a plume stability class for each COC i e designates each COC plume in the source and tail as Increasing Probably Increasing No Trend Stable Probably Decreasing or Decreasing assigns a weighting where the importance of the empirical data a secondary line of evidence is compared to the importance of the other three lines of evidence i e Mann Kendall analysis a primary line of evidence Linear Regression a primary line of evidence and modeling results a secondary line of evidence see LTM Analysis section for a discussion of weighting the different lines of evidence Note that the default weighting system in the software is to weight the two Statistical Plume Analyses with a medium weight while the two External Plume Information including empirical rules is weighted low Again if the users does not want to use empirical rules as a secondary line of evidence then the user can select that option in the software or select Don t Use in the w
226. der M and Wolfe D A 1973 Nonparametric Statistical M ethods Wiley New York NY ASTM 1998 Standard Guide for Remediation of Groundwater by N atural Attenuation at Petroleum Release Sites E 1943 98 43 p U S Environmental Protection Agency 1999 Use of Monitored Natural Attenuation at Superfund RCRA Corrective Action and Underground Storage Tank Site Office of Solid Waste and Emergency Response OSWER Directive 9200 4 17P Final Draft Washington D C April 21 1999 U S Environmental Protection Agency 1998 Technical Protocol for Evaluating N atural Attenuation of Chlorinated Solvents in Groundwater EPA 600 R 128 Washington D C Sept 1998 Version 2 1 A 2 10 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 3 WELL REDUNDANCY SUFFICIENCY ANALYSIS DELAUNAY METHOD Authors Ling M and Rifai H S University of Houston This appendix introduces the approach used in MAROS for well sufficiency analysis the Delaunay method The Delaunay method is designed to select the minimum number of sampling locations based on the spatial analysis of the relative importance of each sampling location in the monitoring network The approach allows elimination of sampling locations that have little impact on the historical characterization of a contaminant plume A well sufficiency analysis method i e recommend new locations based on the Del
227. distributed Optional Power Analysis Shows the Individual Wal Cleanup Status Optional Power Analysis screen where power analyses results based on the Student s t test on mean difference are given Back Returns the user to the Individual Wal Cleanup Status screen View Report Generates a report with individual well cleanup status results for the type of data and time period selected by the user The user can go back to rerun the analysis by selecting a different type of data or by selecting a different time period Visualize Views the results in a map in which wells are shown spatially with different colors indicating their cleanup status This provides a way to visualize the individual well deanup status spatially on the site scale Next Proceeds to the Individual W dl Power Analysis Complete screen Help Provides information on the screen specific input requirements Version 2 1 76 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Individual Well Cleanup Status Optional Power Analysis This screen accessed from the Individual Well Cleanup Status Results screen by clicking O ptional Power Analysis is used to show power analysis results of whether the mean concentration of a well is significantly lower than the cleanup goal based on the Student s t test on mean difference 1 Monitoring and Remediation Optimization System MAROS j x Sample Size The num
228. distribution in a specific site This general spatial pattern is the underlying assumption for the spatial analysis In the Delaunay method we find the general pattern by Version 2 1 A 4 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE averaging across sampling events In addition since the spatial patterns of COCs may be different from each other the optimization is performed based on each COC Therefore results are given separately in terms of each COC Finally we provide the all in one results simply by considering the most conservative result from all COCs The major steps of this algorithm are as follows 1 Select a series of continuous sampling events for analysis They could be all sampling events in the monitoring history They could also be any segment of sampling events in the monitoring history eg sampling events in the past five years 2 Calculate SF values of potential locations for all sampling events selected by the users and for each COC 3 Average SF values of potential locations across the selected sampling events for each COC weighted by the number of locations contained in each sampling event The results are lumped SF values of potential locations for each COC 4 Eliminate one location at a step from each COC starting from the location with smallest lumped SF value Calculate CR and AR ratios for each sampling event and then average th
229. dvantage gained by this approach involves the cases where outliers in the data would produce biased estimates of the least squares estimated slope Parametric tests such as first order regression analysis make assumptions on the normality of the data distribution allowing results to be affected by outliers in the data in some cases However the advantage of parametric methods involve more accurate trend assessments result from data where there is a normal distribution of the residuals Therefore when the data is normally distributed the nonparametric method the Mann Kendall test is not as efficient Both tests are utilized in the MAROS software Primary Line of Evidence 1 M ann K endall Analysis GENERAL The Mann Kendall test is a non parametric statistical procedure that is well suited for analyzing trends in data over time Gilbert 1987 The Mann Kendall test can be viewed as a nonparametric test for zero slope of the first order regression of time ordered concentration data versus time The AFCEE MAROS Tool includes this test to assist in the analysis of groundwater plume stability The M ann Kendall test does not require any assumptions as to the statistical distribution of the data eg normal lognormal etc and can be used with data sets which include irregular sampling intervals and missing data The Mann Kendall test is designed for analyzing a single groundwater constituent multiple constituents are analyzed separately For this e
230. dy the BEG studies and the AFCEE chlorinated database For further Empirical result guidelines see A ppendix A 4 Also state rules are provided to guide the user to site specific guidelines for natural attenuation Results for the empirical trend that can be entered in the software indude Increasing I Probably Increasing PI No Trend NT Stable S Probably Decreasing PD Decreasing D or Not Applicable NA Insufficient Data External Plume Information Empirical Evidence accessed from the External Plume Information Empirical Results screen gives the user guidance for empirical evidence for trends by State Version 2 1 45 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE External Plume Information To view information pertaining to the state of interest choose the state name from the drop down box at the top left Information on general guidelines and regulations specific for Long Term Monitoring are shown E Monitoring and Remediation Optimization System MAROS E External Plume Information Empirical Evidence General guidelines and regulations specific to site location Massachusetts RNA Compounds Gasoline BTEX Diesel PAHs Fuel Cleanup Goals gen Tier l s s Tiers I II Requirements sc shsc fpr lt 0 5 inch RA mon Natural Attenuation ps Evidence Planning guidance Monitoring Location n Monitoring
231. e MW2 Xewdnde 2 Ycondnae 30 Delete Record To delete the record currently Sample Information shown on the screen Deleting a record is a Constituent Type ERGI Tl Sample Date 10747 permanent operation coco BHERIB z Resut 0002 mgl Fes Detection Limit 0 001 mg L Alls fields should be filled in to ensure minimum information for added records However if X and GS Add Record Delete Record Y coordinates are unknown these fields can beleft ipsus blank Back Takes the user back to the D ata M anagement screen Help Provides information on the screen specific input requirements Version 2 1 19 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Site Information accessed from the M ain M enu Screen is the first step in defining the site type as well as parameters unique to the site Provide information regarding the current site amp Monitoring and Remediation Optimization System MAROS Site Information General Project Air Force Base 1 Location Boston Maximum Plume Length Source Information Free Phase NAPL Present Yes M Current Plume Width 10 ft Current Plume Length 100 ft Hydrogeology and Plume Information Seepage Velociy 0 f 100 ft GW Fluctuations M Yes No State Massachusetts El Uy Main Co
232. e Thefollowing steps outline how this is implemented Under the heading Non Detect ND E click EN dick in the middle of the cirde next to here the option Uniform Detection Limit Under the heading Duplicates dick in the middle of the circle next to the option Average Under the heading Trace dick in the middle of the cirde next to the option Actual Value Select N ext to proceed The R edu ced Data Table will be displayed C Note Typically when applying statistics half the detection limit could be used However where therethe detection varies historically then setting a uniform detection limit will reduce the possibility of false trends Version 2 1 A 11 20 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 4 The Reduced Data Table allows the user to view the reduced data table with the COCs chosen as wel as the data consolidation performed This table is not available for editing Select Next to proceed to the R educed D ata Plot screen Click hereto proceed To display data for MW 1 Click the down arrow in the first text box Well to display the options and select MW 1 Click the down arrow in the second text box Chemical to display the options and select benzene LSU touj gt m upa iS Sa Under the heading Graph Type dick on the circle next to Linear TREND
233. e Sample Events screen will appear p Umm rend bem tm berre rm eno m lant wages EE Comas saa W Mis Milo STEP 3 SAMPLE EVENTS Sample Events allows the user to define sample events and dates to be used for graphing and data consolidation This grouping of individual sample days is important for the MAROS analysis to be performed Typically a sample event will last 1 2 weeks depending on how long it takes to sample all the wells at a site and be performed on a quarterly semi annual or annual basis The Effective D ate is selected by the user as representative of the sample event eg sample event start date The Auto Event option is used to automatically set up sample events as unique for each sample date This is appropriate only for a small site where all sampling can be completed on one day i e one date per sampling event nem C Sample Events To define sample events to the right of the heading Sample Events in Database click on Auto Event A list of all the sample events in the dataset will appear in the green boxes Mir Temei Everts m D odor Click the Next button to proceed to the mna Source Tail Selection screen weis Version 2 1 A 11 12 October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Click hereto proceed STEP 4 SOURCE TAIL ZONE SELECTION Source Tail Zone Selection allows the user to definethe well typefor the well
234. e action monitoring may be initiated The confidence limit should only be constructed from data collected during compliance monitoring and should be compared to the ACL computed from the average of background samples EPA 1992 It should not be compared to the maximum concentration limits MCLs The use of tolerance limits in compliance monitoring is questioned by Gibbons 1994 and should be avoided The method to construct a lower confidence limit for the mean concentration from EPA guidance EPA 1992 is presented below Statistical Assumptions Data values are independent and normally distributed If original measurements follow lognormal distribution their logarithms will follow normal distribution and should be used in the computation Procedures Step 1 Use pre determined ACL or estimate ACL from the average of background samples Step 2 Compute the mean x and standard deviation s from the n observations at least four at a compliance well for a constituent To reduce the false negative rate of the test i e to increase the power of the test a larger sample sizen should be used Step 3 Calculate its lower 9996 confidence limit as t 5 e n La x z Nn Version 2 1 A 7 16 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE where t n 1 a is the one sided 1 100 point of Student s t distribution with n 1 degrees of freedom Step 4 M
235. e appropriate information within each field Fields such as Constituent Type and Constituent have dropdown boxes to assist in data entry Choose Constituent Type before choosing the Constituent Review information before adding the record When all the data is entered click on the Add Record button Note If the result is ND non detect then fill in the Detection Limit in the Result cell How will MAROS help perform a trend analysis and give a Site Specific Recommendation based on groundwater data and site conditions TheMAROS Tool can generate a summary report for a selected set of data imported by the user To generate the summary report for the M ann Kendall or Linear Regression Trend A nalysis 1 Follow directions for Importing Entering Data above 2 Main Menu From the M ain M enu select Site Details by clicking the button next to the label This action will take the user to the Site Information screen 3 SiteDetails In each screen select the information that describes the site click on Next to continue to the next screen First enter the site details on the Site Information screen N ext define sample events on the Sample Events screen Then select the representative wells in the Source and Tail zones on the Source Tail Zone Selection screen Continue to the Constituents of Concern Decision screen to choose the representative COCs for the site The next screen Initial D ata Table will show the data to be evaluated To pr
236. e chemical of interest from the Change in Dissolved Mass Over Time dropdown boxes at the top of the screen eoi en ria a purus Choose the graph type i e Log or Linear Select Chemical Li ZEN c u n T nuo Click Graph on graph to proceed ea LLL ILE HL l pu Zeroth Moment Trend The Zero Moment o om _ trend over time is determined by using the 2 Er I Mann Kendall Trend Methodology The ix m Zeroth Moment Trend for each COC is H od m ae determined according to the rules outlined in 5 NN lt a M nes Appendix A 2 Results for the trend indude z Note E du TT Increasing Probably Increasi ng No Trend lt Desas DEN Ted MTI Not Applicable N A due to insufficient data Stable Probably Decreasi ng Decreasing or Not s ESI S a Applicable Insufficient Data MK S TheMann Kendall Statistic S measures the trend in the data Positive values indicate an increase in estimated mass over time whereas negative values indicate a decrease in estimated mass over time The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic i e large magnitudes indicate a strong trend Confidence in Trend The Confidence in Trend is the statistical confidence that the esti mated mass is increasing S20 or decreasing S 0 COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the s
237. e dimensional picture of the hydrologic and geochemical characteristics of the site High concentrations of dissolved contaminants can be the result of leachates rinse waters and rupture of water conveyance lines and are not necessarily associated with NAPLs This type of conceptual model differs from the more generic conceptual site models commonly used by risk assessors that qualitatively consider the location of contaminant sources release mechanisms transport pathways exposure points and receptors However the conceptual model of the ground water system facilitates identification of these risk assessment elements for the exposure pathways analysis After development the conceptual model can be used to help determine optimal placement of additional data collection points as necessary to aid in the natural attenuation investigation and to develop the solute fate and transport model Contracting and management controls must be flexible enough to allow for the potential for revisions to the conceptual model and thus the data collection effort Successful conceptual model development involves EPA 1998 e Definition of the problem to be solved generally the three dimensional nature magnitude and extent of existing and future contamination e Identification of the core or cores of the plume in three dimensions The core or cores contain the highest concentration of contaminants e Integration and presentation of available data including
238. e effect of small sample size 4 observations per well which is far from adequate to characterize the real spatial variability METHOD 8 METHODS FOR TESTING SERIAL CORRELATION Most statistical methods are based on the assumption of independence between observations This assumption is violated if serial correlation or autocorrelation exists in observations separated in time a time series This is common for groundwater quality data that are measured in high frequency such as weekly or monthly sampling To check if a time series dataset is significantly correlated the Durbin Watson test recommended in EPA guidance EPA 1992b can be used The Durbin Watson test is based on the first order or lag 1 autocorrelation model Box al 1994 or AR 1 model which states that the residual of an observation is dependent on the residual of its previous observation by a factor of p 1 p 1 or correlation coefficient The residuals are obtained from detrended and deseasonalized observations if any The AR 1 model can be expressed as e pe where a1 is the residual or error term at timet t 1 and amp is a random shock which is independent and normally distributed at time t Version 2 1 A 7 29 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE If there is no serial correlation between observations the expected value of p will be close to ze
239. e long term monitoring decisions As the monitoring program proceeds more recent sampling results can be added to historical data to assess the progress of the current monitoring strategy Then the optimization process can be reviewed and updated periodically using the MAROS guidance recommendations QUICK START Minimum System Requirements The AFCEE Monitoring and Remediation Optimization System Software runs with Microsoft Access 2000 database software and Microsoft Excel 2000 Operation requires an IBM amp compatible PC with Pentium or later processor To operate efficiently we recommend that the PC have a minimum of 64 MB RAM optimal 128 MB RAM Pentium III and EGA or VGA graphics display Microsoft Access 20008 Microsoft Excel 20006 plus Windows or later or Windows NT are required software Installation and Start Up Copy MAROS SETUP EXE to your hard drive then run MAROS SETUP EXE either by selecting Run from the File menu in Program Manager or by double dicking on the file MAROS SETUP EXE in File Manager or Windows 98 NT 2000 XP Explorer The installation process creates the CA AFCEE MAROS subdirectory on your hard drive unless you install it elsewhere and copies the MAROS files into the new directory This folder contains five files needed to use the software 1 AFCEE Monitoring and Remediation Optimization System Software afcee MAROS v2 mdb 2 Help file afcee MAROS hlp 3 Optimization Excel File xlsD el
240. e sequential t test NO indicates the mean concentration is 1 higher than the cleanup goal or 2 below the deanup goal but not statistically significant because the existence of large data variability prevents the test from resulting in significance The latter case corresponds to an inadequate power in the test In the case of NO sampling should be continued In the case of YES the result from the sequential t test should be consulted as to whether to continue sampling or stop sampling Power of Test Figure A 6 3 is the probability associated with the Student s t test that a well is confirmed to be clean when the mean contaminant concentration is truly below the cleanup goal A value dose to 1 0 may indicate that the data are distributed very close to the sample mean or the coefficient of variation is very small a small variability A value close to 0 indicates the opposite requiring collecting more samples for a future re evaluation A value greater than the expected power indicates data in the well provide sufficient information Expected Sample Size Figure A 6 3 is the number of samples associated with the Student s t test required to achieve the expected power with the variability shown in the data The smaller the value the smaller the data variability and the higher the statistical power If the expected sample size is smaller than the sample size the sampling frequency at this well may be reduced If the expected sample size is gre
241. e the coordinates of the source location for a particular COC First Moment Trend The First Moment trend of the distance to the center of mass over time is determined by using the M ann Kendall Trend M ethodology The First Moment trend for each COC is determined according to the rules outlined in Appendix A 1 Results for the trend include Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not A pplicable Insufficient Data MK S The Mann Kendall Statistic S measures the trend in the data Positive values indicate an increase in the distance from the source to the center of mass over time whereas negative values indicate a decrease in the distance from the source to the center of mass over time The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic i e large magnitudes indicate a strong trend Confidence in Trend The Confidence in Trend is the statistical confidence that the distance to the from the source to the center of mass is increasing S20 or decreasing S 0 COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average Values near 1 00 indicate that the data form a relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatte
242. e the user will have to utilize the Import New Data option where you can import raw electronic data from an Excel File or Access File The first time the data is entered the user can save the data as an archive file for future use The archive file can store analytical data and site details Refer to Appendix A 9 for import file formats Also see example import file M A ROS ExcellmportTemplate xls ver Woon MEROS Archive Fae inert geente ormme errr a j Export MAROS Archive Foe e PIC a monem dame ete tem md r DATA MANAGEMENT Version 2 1 A 11 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 3 IMPORT NEW DATA Thelmport N ew D ata screen is used to import electronic data files To import archived data into the full database MELLE dil d si 1 Enter the full file path and the filename of the archived file to import Folder CA AFCEE_MAROS Click here FileName TutorialExampleData xl s 9 fuos leew Daa A LI F ma wet fupra 73 Lag s RS DATA MANAGEMENT Note The Browse button can also be used to locate the import file 2 Click the Import button to proceed with importing the file to the existing database A screen will be displayed showing the total E Memel or ling eet Merived terion Optriizatios Byrja MANOS Sid dat liis boan yorsay feegurted number of wells and the dates range o
243. each a result or site specific recommendation The assumptions you make along the way will affect the outcome of the software tool results Also the validity of the results or recommendation will rely on the extent and quality of your data For instance more data doesn t necessarily mean better results The data must meet minimum data requirements as to the frequency of sampling duration of the sampling intervals for trend analysis and sampling density for the site as well as the quality of the measurements decreased amount of false positives negatives Version 2 1 12 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE M ain Menu The Main screen serves at the center of the user interface The user progressively steps through the Compliance Monitoring Trend Analysis and Optimization Evaluation process by navigating through the options displayed As individual steps of the process are completed options to select become successively available The M ain M enu screen allows the user to choose between performing Monitoring and Remediation Optimization System MAROS ZI x Perform sampling optimization through various statistical methods used to determine the sampling location and sampling frequency Step 5 MAROS Output View print site specific summary reports and graphs Main Menu the MAROS Analysis process by navigat through the options displaye
244. easing D ing INT NoTrend View Report To Print a summary report dick View Report coc Tail Stability Source Stability Category Result BENZENE PD D L IETHYLBENZENE PD PD L TOLUENE s NT M XYLENES TOTAL PD D L Back Returns the user to the Plume Information by W dl Weighting screen Worst Case 8 Next gt gt View Report Help TREND ANALYSIS Next Takes the user to the M ain M enu Screen Help Provides information on the screen specific input requirements Atthis point in the software your data has been analyzed and design category suggestions are complete You may now proceed to the Main M enu and choose to either perform Well by Well Sampling Optimization Analysis or choose M A ROS Output Print Standard Reports Graphs anane ta Eig d oF Stew 5 93 TREND ANALYSIS Standard Approach Detailed Approach OVERALL PLUME RESULTS WELL SPECIFIC RESULTS MAROS Output Choose to Print View Sampling Optimization Rigorous detailed Reports Standard Reports including statistical approach to sampling the one page heuristic approach to optimization with modules to optimize sampling optimization based on plume sampling location by Deaunay stability and site parameters with results Triagulation and Sampling Frequency for sampling frequency duration and by the Modified CES method or Power density GOTO PAGE 82 Analysis GOTO PAGE 47 Version 2 1 50 Air Force Center for October 2004 Environmental E
245. ecreasing or N ot Applicable Insufficient Data The information displayed in this screen can also be viewed in report form Mann Kendall Statistics Report from the MAROS Output Screen or by dicking on View Report see Appendix A 10 for an example report View Report To print the Mann Kendall Statistics Report or save the report in pdf format and consolidated data click View Report to proceed Back Returns the user to the PlumeA nalysis M enu Next Takes the user to the M ann Kendall Plot Screen Help Provides information on the screen specific input requirements Version 2 1 32 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Statistical Plume Analysis M ann Kendall Plot accessed from the M ann K endall Statistics screen allows the user to view the Mann Kendall Trend Analysis results by well and constituent Mann Kendall Plot inl xi Select a well and chemical below to graph The concentration trend result in the box below reflects the chemical and well chosen to be graphed wel MW 12 Chemical ETHYLBENZENE Graph Type Date E Log Select A Bo 5 S 9 Linear 3 ee Ss Graph View Data SF dw So gx 1 8E 01 Concentration mg L A s 6 0E 02 MK S 4 0E 02 2 0E 02 45 2 0500 Tewari tyssi STEER a Trend Ed 99 3 z MK Concentration Trend D cov a Note Increasing I
246. ection of screen setting or length e Inappropriate overall system design For example well screens not set to span an appropriate hydrostratigraphic zone can cause sample space problems The remedies for sample space problems include installing additional wells resampling the wells or deleting anomalous data collected from the suspect wells Details for the prevention recognition and correction of typical sample space problems are presented in Table A 7 1 System implementation problems refer to situations in which wells or other elements of the system do not perform as designed Typical problems include Well does not produce sufficient water Well silts up after installation Sand pack becomes clogged Well seals leak Well materials degrade Well is poorly constructed For example a well that dries or recharges too slowly to be sampled effectively is an indication of system implementation problems The remedies for system implementation problems include redeveloping the well redesigning a new well or abandoning and replacing the well Details for the prevention recognition and correction of typical system implementation problems are presented in Table A 7 2 Program implementation problems refer to situations in which field data collection or laboratory analysis procedures fail to produce high quality data Typical problems include Version 2 1 A 7 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITO
247. ects O nly The observed serial correlation Version 2 1 A 7 37 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE coefficient is 6 which has been proven to be statistically significant using the Durbin Watson test presented in M ETHOD 8 Step 2 The standard error of the mean when N islarge is approximately 2 goes a m and the D f associated with it is approximately rounded to the YN a 0 3 nearest smaller integer Example Step 1 A hypothetical dataset containing quarterly measurements of a contaminant from a monitoring well during a four year period N 16 is presented in Table A 7 16 The time series plot of this dataset is given in Figure A 7 4 which shows clear seasonal patterns m 4 To determine whether this well is in compliance a lower confidence limit on the sample mean needs to be constructed to compare to the background standard which is 6 5 mg L Step 2 The four seasonal means are 6 52 9 85 and 8 08 and 4 87 mg L respectively For example the first seasonal mean x 6 74 7 64 5 30 6 40 4 2 6 52 The deseasonalized residuals e are listed in the fourth column of Table A 7 16 Step 3 Based on the deseasonalized residuals the observed serial correlation coefficient is N Yee i i l _ 4 084 0 33 and the Durbin Watson statistic is i 2 E 12 261 b i l Y e EC p _ 16 027 ial
248. ed approach Each Delaunay triangle in the triangulated monitoring network is used as a potential area for new sampling locations Figure A 2 6 The SF value at a Delaunay triangle is estimated as follows Consider a Delaunay triangle with vertices Ni N2 and Ns Figure A 2 7 Assume Aj A2 and As are sub parts of the triangle divided based on the centroid of the triangle The average SF valuefor this triangle is estimated as _ SF A SF A SF A SF A A A where SF the sampling events averaged SF value at vertex N Version 2 1 A 4 8 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE SF the sampling events averaged SF value at vertex N 2 SF3 the sampling events averaged SF value at vertex N 3 The estimated SF values at these potential areas reflect the concentration estimation error at these regions for the time period specified by the sampling events For example a value of 0 9 indicates the ratio of the estimated to measured concentration is 1 10 or 10 1 a large discrepancy A value of 0 5 indicates the ratio of the estimated to measured concentration is only 1 2 or 2 1 arelatively small estimation error Potential areas for new sampling locations Delaunay triangles marked by blue lines 1200 4 200 0 Existing sampling locations Delaunay triangle Centroid FigureA 3 7 Division
249. ed to sites wells where concentration levels are consistently high and the ratio of ROC to concentration level is very small Conversely for sites wells where concentration levels are around the cleanup goal small ROC parameters need to be used to provide high sensitivity 10 Click the Analysis gt gt button to perform the analysis The Sampling Frequency Recommendation screen will appear The Recent Result and the Overall Result represent the frequency determined from the recent data and the overall data respectively The Sampling Frequency is the final recommendation after balancing the results obtained from both recent and overall data Click the View Report button to view a result report where the recommended sampling frequency and other details are listed for each well and each COC The user can print this report or export itin different formats Note The frequency recommendations given by the MAROS software should be reviewed in light of the number of samples considered number of non detects etc see Table A 11 2 for example For example if all measurements at a well are nondetects and the detection levels are consistently low a uniform value should be used to quantify the nondetects to avoid a false concentration trend which may leads to inappropriate sampling frequency Regulatory framework community issues and other site specific situations must also be considered in the final decision making see Table A
250. eed to the Step 3c Statistical Plume Analysis Compiete Spatial Moment Analysis Mantloring and Rewedietioe Oplintzalion System MAROS Taa fu Mone ef Tena Arties and Lear Fagin n os hans teen petewedt fou vr Vanwend nl Cain ty Stop Jc gt gt 0 n gt a a ul 4 Version 2 1 36 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Spatial Moment Analysis M oment Analysis Site D eails accessed from the Plume Analysis M enu screen allows the user to enter data by well and constituent to be used in the Moment Analysis Note All Data entry items are mandatory data required includes porosity groundwater flow direction approximate contaminant source location and aquifer saturated thickness The aurrent version of MAROS only allows for designation of one source location and one saturated thickness Back Returns the user to the Plume Analysis M enu amp Monitoring and Remediation Optimization System MAROS E D Io x a E ES zm wW z e E Es x a 2 igl x Moment Analysis Site Details In order to perform the moment analysis calculations there is additional site data that needs to be entered below Choose to either enter a representative saturated thickness of the aquifer at each Well by clicking on Variable Saturated Thickness difficult appoach and then entering the data for e
251. eighting selection Using Empirical D ata as Secondary Evidence APPROACH Step 1 Step 2 Step 3 Determine if you have a plume in one of the following general categories a BTEX Plumes Small Releases BTEX plume from a small fuel release such as a gas station release SEE PAGE A 4 4 b BTEX Plumes Larger Releases BTEX plume from a larger fuel release such as from a tank farm SEE PAGE A 4 8 c MTBE plumes from a small fuel release such as a gas station release SEE PAGE A 4 9 d Chlorinated solvent plumes SEE PAGE A 4 12 Compare the length of you plume to the statistical characteristics of the other plumes from its class by going to the appropriate section A B C or D below If your plume is much shorter than most of the other plumes in its class there may be secondary evidence that your plume has a higher potential to expand You should select Increasing or Probably Increasing and enter in software Of courseif you feel the evidence is not strong enough to be significant you have the option to not use empirical rules as a line of evidence If your plume is much longer than most of the other plumes in its class there may be secondary evidence that your plume has a lower potential to expand You should select Decreasing or Probably Decreasing and enter in software Of courseif you feel the evidence is not strong enough to be significant you have the option to not use empirical rules as a line o
252. elation in the measurements The test can only be performed after the termination of treatment remedial action and after the groundwater has returned back to steady state i e after the disappearance of the post effects of treatment Version 2 1 A 7 18 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Statistical Assumptions Yearly averages are independent and normally distributed If yearly averages follow lognormal distribution their logarithms will follow normal distribution and should be used in the computation When spatial variations between wells are significant joint testing of wells should be avoided Procedures testing wells individually Step 1 Determine the false positive rate a and false negative rate p for control Calculate parameters A and B as I A p Q pai oO Since the facility wide false negative rate FWFN R becomes important in corrective action monitoring the B of single test should be controlled at a low level For example if B 0 2 and thereare 10 wells the FWFNR could be as high as 0 89 So B 0 1 or 0 05 can be used when the number of comparisons is large By contrast of a single test can be moderately increased o 0 10 or 0 05 Step 2 Determine the cleanup standard Cs and the cleanup goal i ui Cs Step 3 Compute the yearly average x using nk samples at least four in year k for the m years of d
253. electing a different time period Help Provides information on the screen specific input requirements Version 2 1 77 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Individual Well Cleanup Status Visualization This screen accessed from the Individual Wal Cleanup Status Results screen by clicking Visualize allows the user to view the individual well cleanup status spatially on the site scale Results based on the period specified by the user are shown graphically for each COC A diamond indicates a well location The well s cleanup status is indicated by its color Well names are not shown for readability 83 Monitoring and Remediation Optimization System MAROS E lol xi C hoose the COC of i nterest from the Individual Well Cleanup Status Visualization d ropdow n list at the top of the screen The well cleanup status is indicated by the color of the well Select a COC to graph Distribution Assumption Then choose bu tton N or m al o r L og n or m al m a see explanations below or click button a el LL Graph to view The default graph type is K Lognormal Normal 160 100 50 50 wo 150 200 250 To facilitate the statistical power dc 0 Groundwater analysis concentration data are Mg so assumed to be ether normally or x lal gt lognormally distributed Results for ll both assumptions are calculated and M d a prov
254. ell Cleanup Status This screen accessed from the D ata Sufficiency Analysis M enu screen by clicking Analysis 1 is used for selecting thetype of data yearly averages or original data and time period defined by aseries of sampling events used in the cleanup status evaluation for individual wells z e 2 E a 9 o EN a 3 S Monitoring and Remediation Optimization System MAROS N Individual Well Cleanup Status This module will determine whether cleanup goals have been achieved at individual wells The user can choose to use either the yearly averages recommended or original data for analysis The user should also select the time period for which the cleanup status will be evaluated The statistical power and expected sample size will also be calculated Select the type of data for cleanup status evaluation r Options Use yearly averages annual mean concentrations from years specified below C Use original data concentrations from sampling events specified below Select time period for cleanup status evaluation Select the beginning and ending sampling events from below From 1888 zi To 1898 lt lt Back Analysis gt gt x 1 Select the type of data Two types of data can be used yearly averages or original data from each sampling event A yearly average is obtained by averaging data for that year and is treated as one sample The original dat
255. em across sampling events to provide sampling events averaged CR and AR values Compare sampling events averaged CR or AR values to thresholds and if thereis no significant information loss repeat this step with the next available location 5 Provide the COC categorized results after eliminating all redundant locations from each COC In this step elimination of a location in a COC means to stop sampling for that COC at that well in the next round of sampling 6 Provide the all in one results by eliminating only those locations that are eliminated from all COCs Here elimination of a location is equivalent to abandoning it i e to stop service of a well since no COC needs to be sampled at this well any more The user can also choose to analyze only one sampling event e g the latest sampling event In this case the step of averaging across sampling events is skipped Figure A 35 shows the detailed procedures of optimization in this simplified process In MAROS two modules are developed based on the Delaunay Method One is the Access M odule starting with screen Wal Redundancy Analysis Ddaunay M ethod which is introduced in chapter M AROS Detailed Screen Descriptions The other one is the Exce M odule xlsD elaunay2K which is a stand alone Microsoft Excel Worksheet discussed in chapter M AROS D amp ailed Screen Descriptions The Acces Module is designed to deal with multi sampling events analyses recognizing that a general spatial patt
256. en in fact Version 2 1 A 7 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE it is not The false negative is the decision that there is no contamination when in fact contamination is present However in corrective action monitoring where the site is undergoing active or passive remediation such as monitored natural attenuation the definition of the null hypothesis turns to the opposite The false positive is then the mistake of concluding that the groundwater is clean when contamination is still present The false negative becomes the condusion that the groundwater requires additional treatment when in fact it has attained the cleanup standards For compliance monitoring the definition of null hypothesis could take the form of the detection monitoring or the corrective action monitoring depending on the statistical methods used Table A 7 6 Three types of groundwater monitoring programs Type Purpose Intensity Implemented when Detection Detect a release to Sampling and analysis of No release to monitoring groundwater 15 inorganic and 47 groundwater has been organic compounds confirmed Compliance Determine if the groundwater Extended sampling of up Release to groundwater monitoring impact is significant to 17 inorganic and 213 has been confirmed by organic compounds detection monitoring Corrective Document the effectiveness of Extensive sampling for A statist
257. ence that the spread of the plume in the x or y direction is increasing S20 or decreasing S 0 COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average Values near 1 00 indicate that the data form a relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about the mean RESULTS AND INTERPRETATION OF RESULTS MOMENT TREND ANALYSIS The Moment Trend Analysis results are presented in the Spatial M oment Analysis Results screen accessed from the Moment Analysis Site Details screen The software uses the input data to calculate the Zeroth First and Second M oments for each sampling event see Figure A 5 1 ETHYLBENZENE TOLUENE XYLENES TOTAL Moment Analysis Results 0th Moment 1st Moment Center of Mass 2nd Moment Spread Pectieo Date eee ae Xe ft Ye ft Sux sq ft Syy sq ft 10 4 1988 1 1E 02 45 39 5 0 11471989 41E 02 7 17 1 299 11 517 3 1 1990 6 6E 03 54 4 1 695 11 576 5 31 1990 4 5E 02 18 19 797 2 568 9 13 1990 8 2E 03 25 17 1 692 5 020 4 3 1991 2 6E 02 17 22 410 706 gt Note xc and Yc are the Centers of Mass Sxx and Syy are the Second Moments which represent the plume spread the Estimated Mass is the Zero Moment Figure A 5 1 M oment A nalysis Re
258. endResults or move it to other folders Also the xIsLO ETrendResults worksheet will remain open until the user closes it All the results and graph output are kept if the user chooses to save the file before closing it The user should save the file under a different name by choosing Save as under the Excel menu option File Version 2 1 88 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Version 2 1 89 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE DATABASE COMPACTION To maintain performance the database must be routinely compacted to remove unused space whenever data is added or changed using thefollowing procedures e Return to theM ain M enu screen e Onthe Tool option of the upper toolbar select Database Utilities and then select Compact Database WARNING It is good practice to keep a backup copy of the database before compacting Should the compact process fail the original database software will still be available Initial Database configuration This software is an automated interface for an Access database containing groundwater data An experienced Access user can work directly with the database at any time by dicking on the command F11 or by choosing Unhide from the Windows Menu to reveal the Access database afcee MAROS The advanced user can use
259. ent 15 12 19 1998 MW 12 1 000E 03 1090 0 7 29E 08 3 541E 07 Yes Yes Sample Event 15 12 19 1998 MW 13 1 000E 03 1125 0 7 29E 03 2 743E 07 Yes Yes Sample Event 15 12 19 1998 MW 14 1 000E 03 1088 0 7 29E 03 3 593E 07 Yes Yes Sample Event 15 12 19 1998 MW 15 1 000E 03 1000 0 7 29E 03 6 824E 07 Yes Yes Sample Event 15 12 19 1998 MW 2 1 000E 03 1192 0 7 29E 03 1 683E 07 Yes Yes Sample Event 15 12 19 1998 MW 3 1 000E 03 1155 0 7 29E 03 2 204E 07 Yes Yes Sample Event 15 12 19 1998 MW 4 1 000E 03 1135 0 7 29E 03 2 550E 07 Yes Yes MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 5 of Project Example User Name Meng Location Service Station State Texas Observed Regression Projected Below Sampling Effective Well Concentration Distance Down Coefficient Concentration Detection Used in Event Date mg L Centerline ft 1 ft mg L Limit Analysis BENZENE Sample Event 15 12 19 1998 MW 5 1 520E 02 1194 0 7 29E 03 2 522E 06 Yes Yes Sample Event 15 12 19 1998 MW 6 1 000E 03 1267 0 7 29E 03 9 743E 08 Yes Yes Sample Event 15 12 19 1998 MW 7 1 000E 03 1277 0 7 29E 03 9 058E 08 Yes Yes Sample Event 15 12 19 1998 MW 8 1 000E 03 1245 0 7 29E 03 1 144E 07 Yes Yes Note Projected Concentrations that are below the user specified detection limit are indicated by a check mark to its right for sampling events with less than 3 selected plume centerline wells NO projected concentrations are calculated because no regression coefficient i
260. enter for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Summarizing Statistical Plume Analysis Trend Analysis Statistics Summary by W ell accessed from the Linear Regression Plot screen allows the user to view the Mann Kendall Trend Analysis and Linear Regression Analysis results by well and constituent 5x To navigate the results for individual Tree finepss Summary py WE constituents click on the tabs at the top of the The results from the Mann Kendall Analysis and Linear Regression Analysis for each COC are screen shown in the data tables sheets below To view the data from each well for individual COC s clicking on the tabs at the top The information displayed in this screen can Lo a e e also be viewed in report form Statistical m s ses oss B E Plume Analysis Summary Report from the S S HER I PS e p M AROS Output Screen or by dicking on View iwi T exor 5r s s Report IMAW 7 T 54E 04 1 68 s s 5 MA 6 T 5 0E 04 0 00 s s 3 pst o Mem or ois Back Returns the user to the Linear Regression E Desena No Tera NNi Nadeau NA Sea aA T Plot e z TT z Back nee venen Heir Next Takes the user to the M ain M enu Screen Help Provides information on the screen specific input requirements At this point the Mann Kendall Trend Analysis and Linear Regression Analysis have been performed You may now proc
261. enter for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE nonparametric test for linear zero slope of the time ordered concentration data versus time The Mann Kendall test does not require any assumptions as to the statistical distribution of the data e g normal lognormal etc and can be used with data sets which include irregular sampling intervals and missing data The Mann Kendall test is designed for analyzing a single groundwater constituent in a single monitoring well multiple constituents are analyzed separately The Mann Kendall statistic S measures trends in the data Positive values indicate an increase in constituent concentrations over time whereas negative values indicate a decrease The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic that is large magnitudes indicate a strong trend A variation of the Mann Kendall test developed by GSI Groundwater Services Inc 2000 is presented in this section for the characterization of both variability and the direction of the trend This modified Mann Kendall test evaluates the S statistic confidence level of the S statistic and the coefficient of variation COV of a time series in order to accurately characterize the concentration trend This trend is classified in six categories Decreasing Probably Decreasing Stable No Trend Probably Increasing and Increasing Statistical Assumptions
262. enter the well Well yields decrease over time Phthalates or inorganics increase over time Abandon and replace well 6 Well is poorly Hirea reliable driller Havean Evidence of poor Abandon and replace well constructed experienced hydrogeologist monitor well workmanship at surface installation Well is not vertical and aligned Water levels and quality appear anomalous Adapted from Table2 in Kufs 1994 Version 2 1 A 7 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 3 Prevention Recognition and Correction of program implementation problems Problem Prevention Recognition Correction 1 Well Require contractors to produce Construction details are missing Use downhole sensors and construction not boring logs and as built diagrams confusing or are not consistent adequately for each well installed Havean with measurements taken for the geophysical logs to documented experienced hydrogeologist monitor well approximate well installation construction details 2 Field data Usetrained field staff and detailed Data are missing or are If necessary resamplethe collection protocols A dapt the protocols to ambiguous well using improved procedures are the geologic conditions and protocols and or more inadequate contaminants expected experienced personnel 3 Sample Usetrained field staff and det
263. entration at this point This well can be the one that has the highest concentration or is screened in the representative aquifer interval with the geologic unit Data from dustered wells can also be averaged to form a single sample and then used in the Delaunay method References Mace R E et al 1997 Extent Mass and Duration of Hydrocarbon Plumes from Leaking Petroleum Storage Tank Sites in Texas University of Texas at Austin and TNRCC Okabe A Boots B and Sugihara K 1992 Spatial Tessellations Concepts and Applications of Voronoi Diagrams Wiley amp Sons New York Rice D W et al 1995 California Leaking Underground Fuel Tank LUFT Historical Analyses UCRL AR 122207 California State Water Resources Control Board Watson D 1994 Nngridr An Implementation of N atural N eighbor Interpolation D F Watson Claremont WA Australia Version 2 1 A 4 10 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 4 QUALITATIVE EVIDENCE EMPIRICAL DATA Authors Newall C J and Aziz J J Groundwater Services Inc Objective There is a growing body of empirical knowledge about the general behavior of groundwater plumes that in some cases might be a useful secondary line of evidence for evaluating plume behavior Webster s N ew Riverside Dictionary defines empirical as Relying on or gained from observation or experiment rather than
264. ere the user can choose which wells i e projected concentrations to use in the risk based power analysis View Report Generates a report with projected concentrations for the sampling events selected by the user for each COC The user can go back to re run the regression by selecting a different set of parameters Analysis Determines the risk based site cleanup status for the sampling events selected by the user The Risk Based Power Analysis Results screen will pop up Help Provides information on the screen specific input requirements Version 2 1 82 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well Selection Form This screen accessed from the Centerline Regression Projected Concentrations screen by clicking Select Wells is for selecting the wells i e the projected concentrations that the user wants to use in the risk based power analysis T x Usein Analysis Indicates whether the well will be used Well Selection Form in the risk based site cleanup evaluation If a well is amp Monitoring and Remediation Optimization System Clck onthe checkbox below to Select or Deselect the welsyouyantto Selected a check mark is displayed in the checkbox The rsen obec ee Pg Cocotte ocn user can select deselect a well by clicking on the checkbox Well Name Use in Analysis E Wn 5 Back Returns the user
265. ere wells representing very large areas both on the tip and the sides of the plume show decreasing concentrations This increasing trend in the spread of the plume shows that although the concentrations are decreasing over time the plume is moving down gradient Version 2 1 A 11 45 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE OVERVIEW STATISTICS PLUME ANALYSIS In evaluating overall plume stability the trend analysis results and all monitoring wells were assigned Medium weights within the MAROS software assuming equal importance for each well and each trend result in the overall analysis Overview Statistics Results e Overall trend for Source region Decreasing e Overall trend for Tail region Probably decreasing e Overall results from moment analysis indicate a decreasing dissolved mass of the plume e Overall monitoring intensity needed Limited These results matched with the judgment based on the visual comparison of benzene plumes over time as well as the Moment Analysis The benzene concentrations observed in 1991 are plotted in Figure A 11 3 The benzene plume concentrations observed in 1991 was very similar to that of 1994 Figure A 11 4 indicating that the benzene plume is relatively stable to decreasing over time z hiirega EBessassesd gt 8595s d m FIGURE A 11 3 SERVICE STATION BENZENE CONCENTRATIONS FOR SAMPLE E
266. ern may lie beneath what are revealed by each single sampling event It can also be used to analyze a single sampling event a special case of the multi sampling events analyses The Exca M odule is designed for one sampling event analyses which provides the user with graphical interface and convenient controls to the optimization process making the process of the Delaunay method better understood Version 2 1 A 4 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE For each well in order Removable wells with SF values less than the threshold Is CR less than its threshold Is AR less than its threshold Will this lead to significant information loss about the plume No No 4X The last well FigureA 3 5 Stepsin Sampling Location Optiomization for One Sampling Event Optimization Parameters Inside node Slope Factor The SF threshold for nodes locations located inside the triangulation domain When SF of an inside node is less than this threshold and if the node is Removable it will be eliminated from the monitoring network The current default value for this parameter is 0 1 Removable stands for the elimination property of a location If the Removable property of a location is False optimization cannot eliminate it no matter how small its SF value is This is important if you want to keep a location eg a
267. erned with the upper prediction limit the limit with a known confidence of not being exceeded by the next k measurements If any of the k measurements exceeds the limit it is probable that contamination occurs and compliance monitoring may be initiated The Simultaneous Normal Prediction Limit for the Next r of m Measurements at Each of k Monitoring Wells presented by Davis and McNichols 1987 is recommended for both inter well and intra well comparisons This method uses Bonferroni inequality to control the facility wide false positive rate FWFPR and verification resampling to minimize the false positive and false negative rates associated with a single comparison Furthermore the dependence in multiple comparisons against the same background inter well comparisons and the correlation due to repeated comparison of the resamples to the same prediction limit intra well comparisons are also handled Statistical Assumptions Data values are independent and normally distributed If original measurements follow lognormal distribution their logarithms will follow normal distribution and should be used in the computation When spatial variations between wells are significant the assumption of homogeneity of variances will be violated and the use of inter well data should be avoided In this case intra well prediction limits can be used instead Procedures background verses downgradient comparisons Step 1 Determine the facility wide false
268. ers However you can use the Excel menu option Save As and save the file under a different name It will open with the saved data in the future The data display can also be saved as a pdf file using the Adobe Acrobat application Version 2 1 60 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Sampling Location Analysis Complete A ccess M odule This screen accessed from the W ell Sufficiency Analysis New Locations screen by dicking N ext is a message screen telling that sampling location determination by the Access M odule has been completed and the user can go back to proceed to other analyses f Moret cong sid Sermmibartian Dyrtissiaabum Ayabe AAE Back Returns the user to the Well Sufficiency z Sampling Location Analysis Complete Analysis New Locations screen The user can go Access Mosul back to rerun the analysis by changing parameters E PIENO Pere Niet BIET e VA ree UNTRA or selecting a different series of sampling vents E pipigain Ere SONG ODORAS YOU ten Ue Q n best te caris avffad nemt t targu ound te ando H H Li z 4 Sampling Optimization Returns the user to the V gena patiba n vanita repent xpi no vau can prosi v ber H z S P eene nr o peat TEE T end 8 laa Sampling O ptimization screen a 4 Back Samples ptemc pees Version 2 1 61 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING
269. ers used Constituent Inside SF Hull SF Area Ratio Conc Ratio BENZENE 0 2 0 01 0 95 0 95 Average Minimum Maximum Well X feet Y feet Removable Slope Factor Slope Factor Slope Factor Eliminated BENZENE MW 1 13 00 20 00 0 259 0 039 0 386 L MW 12 100 00 8 00 0 165 0 000 0 262 MW 13 65 00 23 00 0 254 0 000 0 395 L MW 14 102 00 20 00 0 064 0 000 0 278 O MW 15 190 00 125 00 0 421 0 197 0 538 L MW 2 2 00 30 00 0 308 0 070 0 508 LJ MW 3 35 00 10 00 0 117 0 008 0 297 MW 4 55 00 37 00 0 165 0 035 0 266 MW 5 4 00 70 00 0 532 0 479 0 587 Li MW 6 77 00 5 00 0 526 0 304 0 625 LJ MW 7 87 00 75 00 0 417 0 269 0 490 L MW 8 55 00 95 00 0 645 0 490 0 715 O Note The Slope Factor indicates the relative importance of a well in the monitoring network at a given sampling event the larger the SF value of a well the more important the well is and vice versa the Average Slope Factor measures the overall well importance in the selected time period the state coordinates system i e X and Y refer to Easting and Northing respectively or local coordinates system may be used wells that are NOT selected for analysis are not shown above When the report is generated after running the Excel module SF values will NOT be shown above MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of MAROS Sampling Location Optimization Project Example Location Service Station Sampling Events Analyzed From Sample E
270. es whether or not a location is induded in the analysis of ETT 3 Cini Bere new sampling locations All wells are selected by default ESC M Cmyd fe SF Seer 1 xt D aM 1 Analysis Runs the Microsoft Excel module The xlsN ewLocation worksheet will pop up and becomes the current screen The analysis is performed for the currently selected COC is is s is is Hs s S LS Reset Selects all the sampling locations Nen Wet for the current COC The Sdected status of each location will be set to True SAMPLING OPTIMIZATION Back Returns the user to the A ccess M odule All in one Results screen Next Proceed to the Sampling Location A nalysis Complete Access M odule screen Help Provides additional information on software operation and screen specific input requirements Steps for use 1 Choose the COC for analysis by selecting from the COC dropdown list or typing in the name 2 Set the Selected check box of a location to decide whether this location is included in the analysis 3 Click the Analysis button and the screen will switch to Excel worksheet xIsN ewLocation The data will be transferred to xIsN ewLocation 4 Run xlsNewLocation following instructions given in screen xlsN ewLocation introduced below Version 2 1 58 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE xISN
271. ess or resume sampling If A LR lt B collect an additional year s worth of data and perform the test again If the groundwater from all wells or group of wells attain the cleanup standard conclude that groundwater at this site attains the cleanup standard When testing a group of wells data for the individual wells at each point in time should be used to produce a summary measure for the group as a whole This summary measure may be an average a maximum or a median These summary measures will be averaged over the yearly period Then the same steps for testing wells individually can be followed to make the hypothetical test Example Step 1 Hypothetical arsenic measurements presented in Table A 7 10 are used in this example Here we consider the false positive rate and false negative rate as equally important a B 20 10 Therefore A B 1 a 0 11 and B 1 B a 9 Step 2 The cleanup standard Cs is 5 ppb and the cleanup goal i is expected to be 4 5 ppb Step 3 Compute the yearly average x for each of the four years The results are listed in the fourth column of Table A 7 10 For example the yearly average of 1990 is x 5 67 4 65 2 62 4 07 4 4 25 Step 4 The mean and variance of yearly averages are x 4 58 and s 0 107 respectively Table A 7 10 Example arsenic measurements for sequential test Year Quarter Measurements Yearly average Mean Variance p jn 2 ppb Xi X s Version 2 1 A 7 20 Air
272. ession Analysis MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of Risk Based Power Analysis Projected Concentrations Project Example User Name Meng Location Service Station State Texas From Period 10 4 1988 to 12 19 1998 Distance from the most downgradient well to recep 1000 feet Observed Regression Projected Below Sampling Effective Well Concentration Distance Down Coefficient Concentration Detection Used in Event Date mg L Centerline ft 1 ft mg L Limit Analysis BENZENE Sample Event 1 10 4 1988 MW 1 2 500E 00 1177 0 2 88E 02 4 835E 15 Yes Yes Sample Event 1 10 4 1988 MW 12 2 000E 01 1090 0 2 88E 02 4 732E 15 Yes Yes Sample Event 1 10 4 1988 MW 13 3 500E 02 1125 0 2 88E 02 3 024E 16 Yes Yes Sample Event 1 10 4 1988 MW 14 4 000E 02 1088 0 2 88E 02 1 002E 15 Yes Yes Sample Event 1 10 4 1988 MW 15 1 000E 03 1000 0 2 88E 02 3 156E 16 Yes Yes Sample Event 1 10 4 1988 MW 2 2 000E 03 1192 0 2 88E 02 2 511E 18 Yes Yes Sample Event 1 10 4 1988 MW 3 2 000E 01 1155 0 2 88E 02 7 286E 16 Yes Yes Sample Event 1 10 4 1988 MW 4 2 900E 01 1135 0 2 88E 02 1 879E 15 Yes Yes Sample Event 1 10 4 1988 MW 5 1 500E 00 1194 0 2 88E 02 1 778E 15 Yes Yes Sample Event 1 10 4 1988 MW 6 1 000E 03 1267 0 2 88E 02 1 450E 19 Yes Yes Sample Event 1 10 4 1988 MW 7 1 000E 03 1277 0 2 88E 02 1 087E 19 Yes Yes Sample Event 1 10 4 1988 MW 8 1 000E 03 1245 0 2 88E 02 2 781E 19 Yes Yes Sample Event 2 11 17 1989 MW 1 1 900E
273. ethod with similar output as the M ann Kendall method Statistical analysis results displayed include e The Coefficient of Variation COV a statistical measure of how the individual data points vary about the mean value e Slope the slope of the least square fit through the given data indicates the trend in the data e The Confidence in Trend is the statistical confidence that the constituent concentration is increasing S20 or decreasing S lt 0 The Concentration Trend for each well Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not Applicable Insufficient Data Linear Regression Statistics SORS eate e EET T nem Statistical analysis for the benzene data is GA na displayed Select Next to proceed to the Linear Regression Plot screen Click hereto proceed Note If more than one COC was being used the user would navigate the results for individual constituents by clicking on the tabs at the top of the screen The information displayed can also be viewed in report form Linear Regression Statistics Report from the M ARO S Output Screen cg poet uit ey ir TREND ANALYS Vire Fg Help Ul Linear Regression Plot allows the user to view the linear regression data in graphical form A graph of benzene concentrations for well MW 4 is displayed The Linear Regression statistics are displayed for this well F
274. etur Pyar VANTS faye STEP 1 MAIN MENU Select MAROS Output from the Main Menu The MAROS Output Reports Graphs screen will be displayed Phime Anatyasis L Des Gaisnsta onka Nene erro Taba ent fecun Angga m termi Dom natn Ups t l sanding Optimiation ver ONG GMT Ay SHUM eee E eee Te ts vti eld CA ard BUNTY Visas wr yi MAROS Osnput bmw A Mmm a mcm a t zr A STEP 22 MAROS OUTPUT REPORTS GRAPHS M AROS Output Reports Graphs allows the user to view print reports and graphs from the site trend analyses as well as a preliminary Site Recommendation Report To select a report of graph dick on the title then select View Print Report or View Print Graph 1 Toview theMAROS trend summary results in tabular and graphical format Under Graph click on the arrow at MAROS Output Reports Graphs the bottom right hand corner of the text Heyer idi sat box The option Trend Summary Graphs should be visible Select Trend Summary Graphs Click on View Print Graph to display a table of data spen MARS Brahe Isone p gt Mole Mes Marrtestug ami Nave ditun ortavirattor 2mtam MARI Trend Summary Results Graphing allows Tid Siiweiy R cul Grakhi the user to view print graphical esse LS le summary results in Excel VAR st eto does Select Excel Graph s to spatially display the data This will open Excel on your computer to provide the trend result graphs
275. ewLocation xISN ewLocation accessed from the Well Sufficiency Analysis New Locations screen by clicking Analysis is a Microsoft Excel worksheet used to display the well sufficiency analysis results i e recommending potential areas for new sampling locations M ethod details can be found in the last section of Appendix A 3 The results are shown in the Well Locations chart sheet which is shown below A plot area is located in the center where the sampling locations are plotted in the state coordinate system or relative coordinate system Graph legends and command buttons are on the right side of the chart E xIsNewLocation m New Location Analysis for Existing Locations Potential areas for new locations are indicated by triangles with a high SF level Estimated SF Level S Small M Moderate L Large E Extremely large High SF gt high estimation error new locations needed Low SF gt iow estimation error gt ho need for new locations Back to Access I4 4 MN Well Locations Estimated SF Level The estimated Slope Factor SF value at a potential area indicated by a triangle formed by blue lines for new sampling locations The SF value is used to quantify the concentration estimation error at a potential area The larger the SF value the greater the estimation error Potential areas with high SF values could be regions in which new wells can be placed SF
276. exas Version 2 1 A 4 9 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MTBE PLUME LENGTH DATA USED FOR STEP 3 California Study H appel et al 1998 performed a study of 63 MTBE sites in California They concluded that MTBE plumes were typically equivalent in length or shorter than benzene plumes On a site by site basis this was also true in approximately 8196 of the cases Further at an individual LUFT site the length of a benzene plume was only moderately correlated with the length of the corresponding MTBE plume thus the length of a benzene plume cannot be used to predict the extent of MTBE impact TABLE A 4 2 CUMULATIVE DISTRIBUTIONS OF 1995 96 PLUME LENGTHS IN FT FOR BENZENE AND MTBE SOURCE FIGURE 4 1 HAPPEL ET AL 1998 IWOOT xs 50 120 25 Percentile 85 Minimum 0 Number of Sites 50 The median MTBE plume length was approximately 120 ft Mace and Choi studies 99 MBTE plumes in Texas and compiled the following distribution for MTBE plume lengths TABLE A 4 3 CUMULATIVE DISTRIBUTIONS OF 10 PPB MTBE PLUME LENGTHS IN FT FOR 99 SITES IN TEXAS SOURCE FIGURE 3 MACE AND CHOI 1998 120 EWPecn e L0 Minimum 0 Number of Sites 99 Mace and Choi found that MTBE plumes were on average only slightly longer than their companion benzene plumes Version 2 1 A 4 10 Air Force Center for October 2004
277. ext if deanup cannot be confirmed because of large data variability Results from these analyses provide hints that are helpful in answering these questions and suggestions for expansion or redundancy reduction of future sampling plans The two statistical power analysis methods are introduced in two different sections in this Appendix following a brief introduction of the technique itself The Basics of Statistical Power Analysis Statistical hypothesis tests are widely used in monitoring evaluations such as the statistical tests involved in the three tasks mentioned above In any statistical tests there are two types of error associated with the null hypothesis H o and the alternative hypothesis H 1 false positive type error and false negative type II error These concepts are illustrated in Table A 6 1 False positive refers to the decision that the null hypothesis is rejected when in fact it is true false negative is failing to reject the null hypothesis when it does not hold Correspondingly the false positive rate denoted by o is the probability of incorrectly declining the null hypothesis and the false negative rate denoted by p is the probability of incorrectly accepting the null hypothesis Statistical power is equal to 1 p the probability of correctly rejecting the null hypothesis when it is not true Table A 6 1 Two types of error in a statistical test True condition in Decision based on a statistical sample the well
278. f z jesus sample events Important Check the number Meu rou MM x of walls and date range to make sure all data was recognized Click OK to proceed Thelmport N ew D ata screen will be displayed again 3 From Import New Data screen click Back to return to the Data M anagement M enu Masiiacing arid Boren hater Ogtkzicaem Spi fS import New Data Cie mand a Bn itat Ne porai e mardini Caia Tal on Aceet Table at A mtem rpari mir Codes enm em fe prenom Len cathe set DATA MANAGEMEN Click hereto proceed Version 2 1 A 11 8 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 4 From the Data M anagement M enu screen click on Main Menu to return to the M ain M enu screen f patter pg ated Kewesd ertun pst peten fys io RARI Data Management Menu lbe Cms Meme rt mm doe side pmi cmm pcm cto neon nme ts miter md wm Tr mr ect rd to meet eoe tha lhe entem mite te pey atra De a bee mrret Ae omia rahan aa e are ome ante ttt tie ate er Kaiser Ore Opener Manusi Data Addition We net rsbeesai ieod Pn MANOT etry aere m import MAROS Archive File eee D i tt SCIT Dn Export MAROS Archyve File spot s MAPT mts 09 an moms ek n A Click hereto proceed The site data file will now have been imported from the Excel file The next stage is to define the site details Version 2 1 October 20
279. f evidence If your plume is about the same length than most of the other plumes in its class may be weak secondary evidence that your plume may neither increase or decrease in length You should select Stable or No Trend and enter in software Of course if you feel the evidence is not strong enough to be significant you have the option to not use empirical rules as a line of evidence Version 2 1 A 4 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 4 If available review the data about plume stability for your particular plume class of interest For example plume a thon studies of fuel plumes in California Rice et al 1995 and Texas M ace et al 1997 indicate that most BTEX plumes from small gasoline station releases are either stable shrinking or exhausted If your plume is a BTEX plume from a small release such as a gas station there may be additional secondary evidence that your plume is more likely Stable or Probably Decreasing or Decreasing as opposed to Increasing It is important that the user s experience about the site is used when applying the empirical rules For example a very recent release has a much higher potential for expanding than most of the plumes in the plumea thon databases In summary the empirical data are designed to be supporting External Plume Information that are used carefully based on the user s exper
280. f the MAROS software and itis by no means a complete site analysis MANN KENDALL LINEAR REGRESSION ANALYSIS All 12 monitoring wells had sufficient data within the time period of October 1988 to December 1998 greater than three years of semi annual data to assess the trends in the wells Trend results from the M ann K endall and Linear Regression temporal trend analysis for both Upper Aquifer monitoring wells are given in Table A 11 3 TABLE A 11 3 SERVICE STATION BENZENE WELL TREND RESULTS Numbe Well Well Well Mann Linear Overall Number r Comments Type 2 Category 3 Kendall Regression Trend 5 of of of Trend Trend Samples D etects Detects Consistent detect but MW 1 MW Source D D D 15 15 decreasing trend Inconsistent pattern on MW 2 MW Source NT PD S 15 7 detects and NDs Consistent detect but MW 3 MW Source D D D 15 12 decreasing trend Consistent detect but decreasing trend M ost MW 4 MW Tail D D D 15 14 recentND Consistent detect but MW 5 MW Source D D D 15 15 decreasing trend since 1994 MW 6 MW Source S ND S ND S ND 15 O All samples ND MW 7 MW Source S ND S ND S ND 15 1 Almost all samples ND MW 8 MW Source S ND S ND S ND 15 1 Almost all samples ND MW 12 MW Tail D D D 15 11 Consistent detect until 1994 MW 13 MW Tail D D D 15 10 Consistent detect until 1992 MW 14 MW Tail D D D 15 7 Consistent detect until 1991 MW 15 MW Tail S ND S ND S ND 15 O All
281. f the screen Plume Information by Well Weighting screen allows the user to weight individual wells by all chemicals or by constituent To weigh individual wells the option Weight Wells could be selected on the right of the screen Choices for weighting methods range from High to Low Mte The screen displays Do Not Weight TUTTI Plume information by Well weighting Wells as default This means that the LLL SSeS weighting applied to the trend methods Smee DnS em emen will be applied equally to all the wells This is the option required for this tutorial Select Next to see the results of the weighting The Monitoring System Category screen will be displayed Var veseus Fas Tecserg PI teat Fu a Immer a EL Cmm LIP me FEL ie emet Veri tas ell Option to en Hear Weight Wells Note If more than one COC was being used the user could choose to weight the trend methods applied to each COC individually select Individual Chemicals or to weight all chemicals select All Chemicals M onitoring System Category screen allows the user to view the suggested design category for each COC Trend results for both tail and source wells are given From these results a monitoring system category that characterizes the site for an individual constituent is shown Categories include Extensive E Moderate M and Limited L long term monitoring required for the site Poa Monitoring Sys Cafagor The
282. fined and effective dates E eee eel eg act Dreri ballar Deiis o ethan yn wrs MARIS initial Data Table This table is not available for editing Select N ext to continue daten a AA LM LM pul M paas GU EES arr Moa tapaa Dart vein vat deco IH pen ws gom mang sont thee T a 2 te eec LOT tam pune X Miihe i Benet Loud The Site Details Complete screen will appear be 1 Sewer tent be 1 bwwm Gronks coupe a er 2 terete dice ure ea L2 wewelowtik tithes Unite tate wm wee tw it unt bee serpet b en Pwti tater Mr I ieiet qM a yn ret dt wt atdi zu Baa d Jg a amp Click hereto proceed At this point your data has been imported the wells have been divided into source and tail zones and the constituents of concern have been selected Monitoring and Remediation Opiimtzaiion Systum MAROS Site Details Complete Click on Continue to Step 3 to proceed to Trend Analysis to analyze the plume behavior The Main Menu screen will be displayed de xad fus put onsen Fee ThAMAROL Ont MAROS chron Vo cxt arz ba paarveter and rota kom the Ste Deta Meter Cn rinde ii ETAILS Click hereto proceed nontinun 10 Saeg 15 SITE D Note There is also an option to create an archive file of the site details which have been entered Create MAROS Archive File The user can continue to the analysi
283. for normal porbability plot Quarter Original data Ordered Order ppb data x i 1 13 96 3 87 1 2 12 77 5 15 2 3 9 66 7 00 3 4 8 46 8 16 4 1 8 77 8 46 5 2 3 87 8 56 6 3 5 15 8 68 7 4 10 11 8 77 8 1 8 56 8 84 9 2 8 16 8 97 10 3 8 97 9 66 11 4 8 84 10 11 12 1 11 05 10 82 13 2 10 82 11 05 14 3 8 68 12 77 15 4 7 00 13 96 16 0 2 c S 3 o E o z 7 00 Figure A 7 2 Example normal probability plot Normal Probbility Plot 9 00 11 00 Benzene mg L A 7 26 Cumulative Normal probability 0 quantile y 13 00 0 06 0 12 0 18 0 24 0 29 0 35 0 41 0 47 0 53 0 59 0 65 0 71 0 76 0 82 0 88 0 94 1 56 1 19 0 93 0 72 0 54 0 38 0 22 0 07 0 07 0 22 0 38 0 54 0 72 0 93 1 19 1 56 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE METHOD 7 METHODS FOR TESTING HOMOGENEITY OF VARIANCE The assumption that variances of different groups of data are approximately equal is required for many statistical methods that make references from different groups of data such as the analysis of variance ANOVA parametric or non parametric presented in EPA guidance EPA 1989 EPA 1992 for detection monitoring Violation of this assumption when using these kinds of statistical methods may result in excessive false positive rate or false negative rate Davis and McNichols 1994 Natural spatial variability inherent in a site is
284. generated by monitoring each well Factors such as the location of the screened intervals of monitoring wells in relation to water bearing zones and the hydrogeologic position of each monitoring point in relation to the plume should be considered For example a well screened at an incorrect interval in relation to the water bearing zone will provide misleading information regarding contaminant concentrations The use of sample data from this well in the data evaluation process will lead to high false positive or false negative rates Therefore before proceeding with the details of this study we will briefly review major problems that affect the quality of sample data and ways to address them QUALITATIVE EVALUATION OF GROUNDWATER MONITORING PROBLEMS Kufs 1994 has provided a comprehensive analysis of problems that may affect groundwater monitoring and has provided response measures for these problems The identified problems include sample space system implementation program implementation geologic uniformity hydrologic uniformity and geochemical interaction The first three types of problems that are relatively important in terms of generating useful information are reviewed below Sample space problems occur when the wells in a system are inappropriately located for monitoring a specified volume of the aquifer Typical sample space problems include e Inadequate arrangement of wells for evaluating the extent of contamination e Improper sel
285. gle Conc 310 X ETE 100 00 5 00 0 01 TRUE TRUE MW 12 193832 00 4 EEE 55 00 23 00 0 00 TRUE TRUE MWw 13 14971750 5 040 13 102 00 20 00 0 01 TRUE TRUE MWw 14 1693 50 8 095 2 00 30 00 0 44 TRUE TRUE MWw 2 1802 50 0 009 asi 35 00 10 00 049 TRUE TRUE MWw 3 97745 50 25 100 15 55 00 37 00 0 01 TRUE TRUE Mw 4 619120 00 4 018 t 77 00 5 00 0 00 TRUE TRUE MWw 6 7835 59 42 0 99 18 87 00 75 00 0 00 TRUE TRUE Mw 7 9636 41 116 0 88 agl 55 00 95 00 0 00 TRUE TRUE MWw 8 732585 00 84 035 E i Back to 357899 00 is 79 cet Access 1175355 12 7 066 zaa ree 320828 78 156 031 2234 85027 78 85 0 03 S24 398622 44 90 046 ges 445490 50 85 029 a 307949 50 265 040 ze 663408 00 7 06 e8 2350086 16 1 00 29 1068666 53 7 042 0 45137 00 5 0 48 gst 5884 00 344 0 81 usa 4739 00 304 047 598 287830 50 64 037 Era 8465 00 50 2055 337649 50 67 6 214744 00 6 3 831550 00 181 538 10892 50 154 8 566595 00 158 zT An amp zi QARASI SN IRO ZIZI E WellLocations DataSheet lal puri sie Back to Access Sends results back to the Microsoft Access screen Wal Redundancy Analysis Excd M odule The user can also do this by clicking the button with the same name in the Well Locations chart sheet Source D ata Part Stores the data transferred from M icrosoft A ccess Output Part Outputs some of the intermediate results generated during the optimizing process including the
286. gnificant This test is a widely used method EPA Durbin Watson test Serial correlation in data 1992b Neter et al 1996 Version 2 1 A 7 42 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 19 Methods for dealing with nondetects Nondetects METHOD 9 Distributional Method Description Assumption Sub Method 1 Normal or The method replaces nondetects with half of Simple replacement Lognormal their MDLs for undetected measurements or PQLs for detected but not quantified measurements 15 50 Sub Method 2 Normal or The method makes Cohen s adjustments by Cohen s adjustment Lognormal including nondetects in the calculation 50 90 Sub Method 3 Not known The method uses a nonparametric version of Nonparametric statistical interval estimates e g prediction methods limit and confidence limit 90 100 Sub Method 4 Poisson The method uses the Poisson prediction limit or Poisson model tolerance limit constructed from counts of analytical hits only Sub Method 5 Not known The method uses laboratory specific QL or limits Specified limits required by applicable regulatory agency Table A 7 20 Methods for dealing with serial correlation and seasonal effects Condition METHOD 10 Method Description No seasonal patterns but Sub Method 1 The method adjusts for serial correlation may be serially correlated
287. han one COC was being used the user would navigate the results for individual constituents by dicking on the tabs at the top of the screen YTTYTYTYTIT The information displayed in this screen can also be viewed in report form Statistical Plume Analysis Summary Report from the M AROS Output Screen or by dicking on View Report In this particular example the Mann K endall results are the same as the Linear Regression results for all wells 7 The Statistical Plume Analysis Complete screen indicates that the M ann Kendall Trend Analysis and Linear Regression Analysis have been performed The next stage will be Spatial Moment Analysis Select Plume Analysis to return to the PlumeAnalysis M enu Mantloring and Rewediatioe Optimization System MAROS Statistical Plume Analysis Compiete Tous Te Mower terial Teen Arden and Lines Hap ener dn o howe teen par aoo You mag now pastos re Siy X Soda Mert Levin Click hereto proceed Version 2 1 A 11 28 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 4 SPATIAL MOMENT ANALYSIS The Spatial Moment Analysis option is used to perform Moment Analysis Zero First and Second M oments calculated 1 From the Plume Analysis M enu select the Step 3c Spatial Moment Analysis option The Moment Analysis Site Details screen will be displayed Plume Analysis Menu
288. he lowest sampling frequencies for these sampling locations It also provides data sufficiency analyses for the current monitoring program To access the Sampling Optimization module complete the following steps 1 Start Screen After starting the MAROS software the Start Screen is shown input user name and project name and click button Start You will enter the M ain M enu 2 Main Menu In the M ain M enu the Sampling Optimization module is the fourth option The Sampling Optimization label is red and the button next to it is deactivated Follow Version 2 1 9 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 3 instructions and complete the three modules above the Sampling O ptimization module in that order They are Data M anagement Site Details and Plume Analysis After running through the three modules go back to Main Menu the button next to label Sampling Optimization will be activated click this button the Sampling Optimization screen will appear Sampling O ptimization The sampling optimization screen is a main menu for three sub modules Sampling Location Analysis Sampling Frequency Analysis and Data Sufficiency Analysis Now you can follow the instructions and perform the three analyses To View Print Report 1 2 M odule end Results Reports At the Results screen s of each sub module e g screen Risk Based Power Analysis Results
289. he distance from each well to the compliance boundary The projected concentrations from each sampling event are then used in the risk based power analysis Since there may be more than one sampling event selected by the user the risk based power analysis results are given on an event by event basis To determine the site cleanup status a significance test based on the following statistic is used m ysin where c is the cleanup goal m and s are the mean and standard deviation estimated from the projected concentrations respectively n is the number of projected concentrations and t is the test statistic following t distribution with n 1 degrees of freedom When log transformed data are used i e under lognormal distribution assumption c is the logarithmic cleanup level and m and s arethe mean and standard deviation of the projected concentrations respectively e t Equation A 6 14 Concentrations Compliance boundary _ projected to this The nearest downgradient receptor Groundwater flow direction Figure A 6 5 Illustration of projected concentrations for risk based power analysis The significance of the site cleanup test is found by comparing the test statistic t with the critical t value under significance level a In calculating statistical power and the expected sample size Equation A 6 4 and Equation A 6 5 are used but with the statistics introduced in Equation A 6 14 Version 2 1 A 6 8
290. he following format for ERPIMS files in Microsoft Access Table A 1 2 5 or ERPIMS text files should be used for importing files into MAROS The Constituent Naming convention follows ERPIMS The Access template file MAROS AccessTemplate mdb should be followed to import an ERPIMS Access import file for the MAROS software Only the fields with an asterix below are mandatory fields for the ERPIM S Access import file Version 2 1 A 2 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 1 2 REQUIRED FIELD FORMAT FOR LDI IMPORT FILES LOCATION RESULTS Column Number Field Name Description 1 AFIID Air Force Installation 2 LOCID Location Identifier 3 LTCODE Location Classification Code 4 LPRCODE Location Proximity Code 5 NCOORD N orth State Plane Coordinate 6 ECOORD East State Plane Coordinate 7 CRDTYPE Coordinate System Type 8 CRDM ETH Coordinate System M ethod 9 CRDUHN Precision of the Coordinates 10 CRDUNITS Coordinates Units of Measure 1 ESTDATE Date Established 12 ESCCODE Establishing Company Code 13 DRLCODE Drilling Company Code 14 EXCCODE Excavating Company Code 15 CMCCODE Construction M ethod Code 16 ELEV Surface Elevation 17 ELEVMETH Elevation Determination Method 18 ELEVUN Precision of the Elevation 19 ELEVUNITS Elevati
291. his study were e Statistical methods such as general linear models and comparison of probability distributions of plume length indices are useful to quantify expected relationships between plume length and site and CVOC variables within a population of CVOC plumes In addition they provide population statistics that may be used to bound the uncertainty inherent in expected plume behaviors An important conclusion of this study is that the presence of a vinyl chloride plume indicates that reductive dehalogenation may be playing a role in reducing the extent of CVOC plumes at approximately one third of the sites examined In contrast the Version 2 1 A 4 13 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE presence of a cis 1 2 DCE plume in the absence of a vinyl chloride plume appears to indicate reductive dehalogenation rates that are insufficient to effectively reduce the extent of CVOC plumes at a site Little evidence was found in the data to suggest that plume lengths and plume growth rates are substantially affected by reductive dehalogenation in these circumstances There are no statistically significant differences between CVOC species with regard to their log transformed 10 ppb plume lengths induding likely transformation daughter products such as ds12DCE and vinyl chloride Plume lengths are positively correlated with maximum historical CVOC concentrations
292. hods for Evaluating theA ttainment of Cleanup Standards V olume 2 Ground Water Environmental Statistics and Information Division Office of Policy Planning and Evaluation U S Environmental Protection Agency Version 2 1 A 6 10 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 7 FALSE POSITIVE NEGATIVE MINIMIZATION METHODOLOGY Authors Ling M and Rifai H S University of Houston This appendix introduces the methods and strategies for minimizing false positive and false negative error rates in the statistical analysis of monitoring data Most of the methods introduced in this appendix have not been implemented in the MAROS software This appendix serves as a supplementary information source for those who have a deeper interest in this issue Introduction Data evaluation is an essential part of a long term monitoring program in that it aids in making decisions regarding plume conditions and the appropriate response measures Uncertainty in the sample data can cause false positives and false negatives in the data evaluation procedure resulting in misleading or incorrect conclusions False positive refers to falsely conduding the presence of a condition when it is in fact not present False negative refers to the failure of recognizing the presence of a condition when it is present In groundwater monitoring for example this condition could bethe conta
293. ically significant action remediation and the site characterization groundwater impact has monitoring attainment of cleanup combined with remedial been confirmed by standards actions compliance monitoring Adapted from Weber 1995 STATISTICAL ANALYSIS UNDERLYING ASSUMPTIONS An important statistical assumption that underlies all statistical methods used in groundwater monitoring is the assumption that observations are independently and identically distributed This can be darified using the following three assumptions e Independence Data values used in a statistical test are independent of each other This assumption forms the basis of both parametric and nonparametric tests used in groundwater monitoring Correlation between observations resulting from spatial or temporal correlation may violate this assumption e Homogeneity of Variances Data values used in a statistical test have equal variances for all values of the independent variables This assumption forms the basis of both parametric and nonparametric tests used in groundwater monitoring N atural spatial variation tends to violate this assumption when performing inter well analyses Inter well analysis refers to statistical tests performed using measurements from different wells e g upgradient versus downgradient comparisons Intra well analysis refers to statistical tests or analyses performed using measurements from the same well e g comparing new monitoring measuremen
294. iciency Analysis M enu amp Monitoring and Remediation Optimization System MAROS j lol xl D ata S uffi ci en cy A nal ysi S M enu accessed from the Sampling Optimization screen by dicking Data Sufficiency Analysis is is tao cer subline WRG SE fek basoa ow onan RE the main menu for data sufficiency Cleanup The two analyses are independent and the user can choose to perform any analysis first Data Data Sufficiency Analysis Menu sufficiency analysis parameters such as Cleanup Goal can be set in Options anal ysi S that i nd u des two ty pes of Select Any Analysis to Proceed stati sti cal power anal yses Analysis 1 Power Analysis at Individual Wells It allows the user to choose between Evaluation of cleanup status at individual wells based on erf i observed concentrations p orml ng e Power Analysis at Individual Analysis2 Risk based Power Analysis Wells Evaluation of risk based site cleanup based on virtual concentrations projected to the complaince boundary e Risk Based Power Analysis SEE Options Help The analyses accessed by each choice are as follows Power Analysis at Individual Wells Determines the cleanup status of individual wells using a sequential t test from EPA 1992 An optional power analysis based on the Student s t test on mean difference is also provided Refer to Appendix A 6 for details Risk based Power Analysis Determines the risk based site cleanup status using estimated
295. ided for comparison See Appendix 1 lt lt Back A 6for detailed explanations x E TEE E SNC Luwe Groundwater Flow Direction Indicates the general groundwater flow direction specified by the user in the Spatial Moment Analysis module If the flow direction is not previously specified a default direction is shown Normal Views results calculated under the assumption that data are normally distributed Lognormal Views results calculated under the assumption that data are lognormally distributed G raph Plots or refreshes the graph Back Closes this screen and returns to the Individual Well Cleanup Status R esults screen Help Provides information on the screen specific input requirements Note This graph can also be viewed and printed from the M AROS Output screen See Appendix A 10for an example graph Version 2 1 78 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Individual Well Power Analysis Complete This screen accessed from the Individual Well Cleanup Status Results screen by dicking Next isa message screen indicating that individual well power analysis has been completed and the user can proceed to other analyses E3 Monitoring and Remediation Optimization System MAROS x Back Returns to the Individual Wal Cleanup Status m individual Well Power Analysis Complete Results screen The user can go back to rerun the
296. ield coordinate system and then rotated counterdockwise using the standard Cartesian tensor rotational transformation with the following formulas S S cos0 25 sin0 cos0 S sin8 S S sin 25 sin0 cos0 S cos0 yy Version 2 1 A 5 5 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE where 6 is the representative groundwater direction measured anti clockwise from the X axis field coordinate system These are the actual values reported as second moments in MAROS Second M oment Trend The Second Moment trend of the Spread of the Plume in the X or Y direction over time is determined by using the M ann K endall Trend Methodology The Second Moment trend for each COC is determined according to the rules outlined in Appendix A 1 Results for the trend include Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or N ot Applicable Insufficient Data MK S TheMann Kendall Statistic S measures the trend in the data Positive values indicate an increase in the spread of the plume over time expanding plume whereas negative values indicate a decrease in the spread of the plume over time shrinking plume The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic i e large magnitudes indicate a strong trend Confidence in Trend The Confidence in Trend isthe statistical confid
297. ience and site knowledge Version 2 1 A 4 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE A Empirical Data BTEX Plumes Small Releases Recent studies of over 600 groundwater contamination sites throughout the U S provide important information regarding the fate and transport of petroleum hydrocarbons in the subsurface An API research summary Newell and Connor 1998 examined the findings of four independent research studies and addressed several key technical issues regarding the assessment and remediation of BTEX benzene toluene ethylbenzene xylene plumes Each study involved detailed analysis of data from a large number of sites primarily leaking underground storage tanks to identify the salient characteristics of groundwater contaminant plumes caused by petroleum hydrocarbon releases Two studies California and Texas evaluated the trends in dissolved petroleum hydrocarbon plumes PLUME LENGTH DATA USED FOR STEP 3 Title 2236 API pic A ZAPF eps Creator Canvas Preview This EPS picture was not saved with a preview included in it Comment This EPS picture will print to a PostScript printer but not to other types of printers FIGURE A 4 1 LOCATION OF BTEX PLUMES SMALL RELEASE STUDIES Version 2 1 A 4 4 October 2004 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SY
298. ify the period of interest below or leave blank if you would like to use all of the data the analysis For example if the import data ranges from 1979 to 2004 but the From 10 4 1988 To 12 19 1998 ET define the ti Choose the option to define the analyst is only interested in the ti me parod to concder whine dle representative statistical dataset period from 1999 to 2003 the user can Bea Pana Tin Cnain eee specify the date range at this location in a ced P the software The user should specify the E Ober Tinetriews sl Maium Highest period of interest in the boxes or leave E Data consolidation is recommended for datasets with greater than 40 sample events blank if all of the data is to be used g me Next gt gt m 2 Choose the option to consolidate the time period to consider within the dataset by clicking on the options on the bottom left of the screen If you do not wish to perform any data consolidation choose Do Not Perform Time Consolidation 3 Choose the option to define the representative statistical dataset within the consolidated time interval at the bottom right of the screen Note This option is not needed if you have chosen Do Not Perform Time Consolidation Back Returns the user to the PlumeA nalysis M enu screen Next Takes the user to the D ata Reduction Part 2 of 2 Screen Help Provides information on the screen specific input requirements Note Data cons
299. im Lach Woes ient Pririheriw we srties Use the scroll down arrow on the right of the screen to view results for wells not displayed uwaa Lap ta mapie ad dumpa mha Se ee e Lond oom a i ararat enr mi Click on View Report to print the Mann Kendall Statistics Report rel Click hereto view report Note If more than one COC was being used the user would navigate the results for individual constituents by clicking on the tabs at the top of the screen WR TT Fabi iw wi 5 p ION oben d Vrat IDIPDTPUIT IPS Rae igre Wt hg REND ANALY Ss Version 2 1 A 11 24 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Version 2 1 A 11 25 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The Mann Kendall Statistics Report displays consolidated data and results of the Mann Kendall analysis This report can also be printed MAROS Marmn Xendall Statistics Close the report by dicking on the red button in the top right hand corner of the screen The M ann K endall Statistics screen will return ee ee atte E 1o rr MOT l Pht tet meme orm de Lond bom ol m be ar mi VY CHR MONS Cebia m rent Cmm nem de x4 torte tm fim l ian Select the Next button on the Mann eae Kendall Statistics to proceed Click here to The M ann K endall
300. imported Help Provides information on the screen specific input requirements Note To import a MAROS 1 0 archive file the file must first be converted to Access 2000 To convert a an archive file to A ccess 2000 open the file within Access 2000 and choose the option Convert Database and save the file under a new name Once the archive file is converted to Access 2000 you will be able to import the file into the MAROS 2 0 software Version 2 1 17 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE ExportMAROS ArchiveFile Export ArchiveFile accessed from the D ata M anagement M enu screen is used to export a MAROS data file To export data into an archive database file BE Monitoring and Remediation Optinization System MAROS Export MAROS Archive Fife 1 Enter the full file path and filename of the archived file to export or click the browse button NICE uo TA AE to find the archive file to overwrite The Folder Sree a NES epu RS na rent T and File name you choose will appear in the top two boxes Folder p AGSI USERS 2236 LTMP archives Filename archivesiteB mdb 2 Click Create to proceed with exporting the data to thearchive file f Tum Back Takes the user back to the Data M anagement E screen 2 Z ees et Help Provides information on the screen specific input requirements A MAROS archive file can also be create
301. inated Solvent Plumes Two chlorinated solvent plume a thons are available for use as secondary evidence one performed for the Air Force Center for Environmental Excellence Tech Transfer Division by Groundwater Services Inc and one performed by the Lawrence Livermore National Laboratory CHLORINATED SOLVENT PLUME LENGTH DATA USED FOR STEP 3 AFCEE Study The AFCEE database Aziz et al in review used data from site investigation treatability and natural attenuation reports to compile the database Questionnaires were completed using mean hydrogeologic property values extracted from the site reports for the most contaminated unit Plume lengths were determined using isopleths for each chlorinated ethene or chlorinated ethane constituent included in the site report The project developed several correlations to plume length and estimated first order biodegradation rates for both parent compounds and daughter products using the BBOCHLOR model Aziz et al 1999 When comparing the chlorinated ethenes i e PCE TCE c DCE t DCE and vinyl chloride TCE and the DCE isomers have the longest median plume lengths all in the 1200 ft range as shown in TableA 5 4 Vinyl chloride has the shortest median plume length of 860 ft followed by PCE with a plume length of 970ft TABLE A 4 4 CUMULATIVE DISTRIBUTIONS OF CHLORINATED SOLVENT PLUME LENGTHS IN FT AND ASSOCIATED COMPOUNDS PLUME LENGTHS IN FT SOURCE TABLE 3 AZIZ ET AL IN REVIEW Plu
302. ination E Monitoring and Remediation Optimization System MAROS Source Tail Zone Selection im SS eee Source Zone Tail Zone Select representative wells in the Source S and Tail Ines or Not Used Choose either Tail or Source or Not Used by clicking on the box to the right of the well in the table below Well Coordinates accessed from the Source T ail Zone Selection Screen allows the user to define and or revise the well coordinates if they were not defined in the import file Well coordinates are mandatory and should be in feet e g State Plane coordinates or arbitrary site coordinates Next Takes the user to the COC Decision screen Back Returns the user back to the Source T ail Zone Selection screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 22 Select representative wells in the Source S and Tail T zones or Not Used Choose either Tail or Source or Not Used by clicking on the box to the right of the well in the table Select representative wells in the Source and Tail zones Next Takes the user to the Well Coordinates screen Back Returns the user back to the Sample Events screen Help Provides information on the screen specific input requirements rmm mh Pme teei Peor CRIT Well Coordinates v fr MATS anten m acm tomm Pe ody Pee we rere one Hee enn om B emnt ros he mieten ct er
303. inty at different spatial locations Temporal variation refers to systematic time effects in addition to random measurement errors Unlike the first two kinds of uncertainty which can be avoided or reduced by a well planned sampling strategy and analytical protocols uncertainty associated with natural variability can only be understood using appropriate statistical techniques The intent of this appendix is to develop data evaluation strategy using appropriate statistical techniques which will reduce the probability of making false positive and false negative decisions Therefore only the last type of uncertainty i e uncertainty due to natural variability is considered in this study assuming that the first two types have already been addressed Problems involving spatial correlation or temporal correlation between measurements will also be considered These correlations if not addressed cause violations of the statistical Version 2 1 A 7 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE assumptions that underlie most statistical methods and cause excessive false positive and false negative rates during the statistical tests A thorough evaluation for optimization of a long term monitoring program not only includes the development of an appropriate data evaluation strategy but also requires a qualitative review of the program to determine the value of the information
304. ion 2 2002 AFCEE 12 1 2003 Page 1 of 2 MAROS Second Moment Analysis Effective Date Constituent Sigma XX sq ft Sigma YY sq ft Number of Wells 12 10 1997 BENZENE 1 749 6 327 12 6 19 1998 BENZENE 1 711 5 988 12 12 19 1998 BENZENE 2 676 10 095 12 Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Due to insufficient Data 4 sampling events The Sigma XX and Sigma YY components are estimated using the given field coordinate system and then rotated to align with the estimated groundwater flow direction Moments are not calculated for sample events with less than 6 wells EE eee MAROS Version 2 2002 AFCEE 12 1 2003 Page 2 of 2 MAROS Site Results Project Location Service Station User Defined Site and Data Assumptions User Name State Texas Hydrogeology and Plume Information Groundwater Seepage Velocity 92 ft yr Current Plume Length 270 ft Down gradient Information Distance from Edge of Tail to Nearest Down gradient receptor 1000 ft Down gradient property 1000 ft Current Plume Widt 150 ft Distance from Source to Nearest Number of Tail Wells 5 Down gradient receptor 1000 ft Number of Source Wells 7 Down gradient property 1000 ft Source Information Source Treatment No Current Site Treatment NAPL is not observed at this site Data Consolidation Assumptions Plume Info
305. ion Summary Weighting Each trend method is shown in the tab sheets below Choose to weight the trend method applied to each COC individually by clicking Individual Chemicals hard approach or choose to weight all chemicals by selecting All Chemicals easy approach Choices for weighting methods range from High to Low If you choose not to weight trend methods leave the default of All Chemicals and Medium weight When finished click Next to see results of weighting All Chemicals Individual Chemicals 10 x AI COCs Line Of Evidence Source Weight Tail Weight Mann Kendall Statistics Medium v Medium Linear Regression Medium Medium Modeling Analysis Low zl Low Empirical Evidence Low zl Low Ea a lt lt Back Next gt gt Help REND Avatysis Each trend method is shown in the tab sheets Choose to weight the trend methods applied to each COC individually by clicking Individual Chemicals difficult approach or choose to weight all chemicals by selecting All Chemicals easy approach Choices for weighting methods range from High to Low If you choose not to weight trend methods leave the default of All Chemicals and Medium weight If you choose to not include the Empirical Evidence choose Not Used When finished click Next to see results of weighting Back Returns the user to the Statistical and Plume Information Summary by W dl screen Next T
306. ironmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Statistical Assumptions Data values are independent and normally distributed with mean u and standard deviation of c If original measurements follow lognormal distribution their logarithms will follow normal distribution and should be used in the computation Procedures Step 1 Estimate u and o by computing the mean x and standard deviation s of at least eight historical independent samples collected in a period of no less than one year Setp 2 Select the three Shewart CU SUM parameters h the value against which the cumulative sum will be compared c a parameter related to the displacement that should be quickly detected and SCL the upper Shewart limit that is the number of standard deviation units for an immediate release It is suggested that c 1 h 25 and SCL 4 5 are most appropriate for groundwater monitoring applications The false positive rate associated with these parameters is about 196 Step 3 Denote the new data value at time point t as xi and compute the standardized value zi Step 4 At each time period ti compute the cumulative sum Si as S max 0 z c S i where max A B is the maximum of A and B starting with So 0 Step 5 Plot the values of S y axis versus ti x axis on a time chart Step 6 Make decisions Declare an out of control situation for sampling period ti if for the first time S gt h
307. is expected to approximate a first order exponential decay for compliance monitoring groundwater data With actual site measurements apparent concentration trends may often be obscured by data scatter arising from non ideal hydrogeologic conditions sampling and analysis conditions However Version 2 1 A 2 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE even though the scatter may be of such magnitude as to yield a poor goodness of fit typically characterized by a low correlation coefficient e g R lt 1 for the first order relationship parametric and nonparametric methods can be utilized to obtain confidence intervals on the estimated first order coefficient i e the slope of the log transformed data Nonparametric tests such as the M ann K endall test for trend are suitable for analyzing data that do not follow a normal distribution Nonparametric methods focus on the location of the probability distribution of the sampled population rather than specific parameters of the population The outcome of the test is not determined by the overall magnitude of the data points but depends on the ranking of individual data points Assumptions on the distribution of the data are not necessary for nonparametric tests The Mann Kendall test for trend is a nonparametric test which has no distributional assumptions and irregularly spaced measurement periods are permitted The a
308. is runs by selecting lt lt Back Ef xlsNewLocation 10 x Ettg Locations Potential areas for new locaton t ar hdbaedby teiges wt aibi SF evel Ertnatd SF Leuet S Small M Moderate L Large E Extremely Bre Hp SF higi estatbi erpra poss bk need Tor rew boatos Low SF bw estimatbi eror YO need for iew beatbis Colored D Back to letter Access Delaunay triangle M NWell Locations amp Monitoring and Remediation Optimization System MAROS Click the Sampling Optimization button and the user will be brought back ON Sampling Location Analysis Complete cis m Access Module to the Sampling Optimization Menu KR You have finished the analysis of sampling locations by analyzing across screen T the sample events selected by you for each COC and across COC You may now proceed to other options of Sampling Optimization You can also O go back to choose another series of sample events for analysis If you would like to view the report right now you can proceed to Main Zz Menu from Sampling Optimization or go back to previous screens where a reports can be generated a z lt lt Back Version 2 1 A 11 52 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Note The redundancy reduction results based on the Delaunay method are provided in Table A 11 1 The u
309. ite soil samples lower than expected Install additional wells or find existing wells screened in the same aquifer In some cases packers can be used to test specific zones 4 Screen length Use background information and Water elevations appear to be Use packers to isolate zones not correctly geophysical surveys to project anomalous contaminant in open hole wells Install selected correct screen length to meet study concentrations lower than additional wells objectives Confirm length using expected soil samples collected from boreholes 5 System not Identify ultimate use of data and Groundwater flow or Resample wells and or adequatel y methods of data analysis to estimate contaminant migration appears install additional wells designed to minimum sample size to be ambiguous or illogical Augment direct data with accomplish study indirect data e g geophysics and soil gas Delete anomalous data collected from suspect wells Version 2 1 October 2004 Adapted from Table 1in Kufs 1994 A 7 4 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 2 Prevention Recognition and Correction of System implementation problems Problem Prevention Recognition Correction 1 Well does not If consistent with monitoring objectives Well is dry or rechargestoo Redevelop well Deepen produce sufficient screen
310. ites sites sites 0 01 996 4 496 2 8796 41 0 05 11 5 1396 6 71 36 0 1 13 6 15 7 72 34 0 2 21 10 19 9 60 28 0 3 21 10 26 12 53 25 0 5 23 11 28 13 49 23 TABLE A 4 7 TEMPORAL TRENDS IN PLUME LENGTH FOR CVOC PLUMES FROM THE NO REDUCTIVE DECHLORINATION AND WEAK REDUCTIVE DECHLORINATION GROUPS CHARACTERIZED BY MONITORING DATA FROM THREE OR MORE YEARS SOURCE MCNAB ET AL 1999 p value Plumes D ecreasing In Plumes Increasing In Plumes With No Length Length Significant Trend Sites Number Sites Number Sites Number sites sites sites 0 01 8 14 13 78 73 0 05 10 9 21 20 69 65 0 1 12 11 27 25 6296 58 02 1496 13 3496 32 5296 49 0 3 17 16 38 36 45 42 0 5 19 18 44 41 37 35 The authors concluded that Version 2 1 A 4 15 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Regardless of the confidence level the two populations of plumes do appear to differ from one another according to this analysis in that the plumes from the Strong RD group exhibit a diminished tendency toward increases in plume length than those plumes from the No RD and Weak RD groups Previous historical case analyses of fuel hydrocarbon plumes Rice et al 1995 Mace et al 1997 indicated that only a small minority of hydrocarbon plumes on the order of 1096 were experiencing discernable plume growth presumably as
311. itial Site Investigation Evaluation of groundwater plume conditions and appropriate response measures requires adequate site characterization including plume delineation Therefore for the compliance monitoring evaluation the minimum required site information includes e Constituents of Concern COCs Individual constituents must be identified along with their relevant source areas and transport mechanisms e SiteH ydrogeology Site stratigraphy and groundwater flow velocity and direction must be identified e Affected Groundwater Plume must be completely delineated for each COC to ensure that the results of the compliance monitoring assessment are reliable and not erroneously influenced by a migrating plume e TimeSeries Groundwater M onitoring D ata Historical record must be compiled for each COC and meet the minimum data requirements described below e Actual and Potential Groundwater Receptors Well locations groundwater to surface water discharge locations underground utilities or other points of exposure must be identified e Current or Near Term Impact Any current or near term receptor impact defined for this evaluation as occurring in zero to two years must be assessed Plumes posing current or near term impact on applicable receptors are referred for immediate evaluation of appropriate risk management measures Site Conceptual M odel The EPA recommends the use of conceptual site models to integrate data and guide bo
312. itive values indicate an increase in constituent concentrations over time whereas negative values indicate a decrease in constituent concentrations over time The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic i e large magnitudes indicate a strong trend e The Confidence in Trend is the statistical confidence that the constituent concentration is increasing S20 or decreasing S lt 0 e The Concentration Trend for each well is determined according to the following rules where COV is the coefficient of variation Version 2 1 A 2 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 2 1 MAROS MANN KENDALL ANALYSIS DECISION MATRIX Mann Kendall Confidence Concentration Statistic in Trend Trend 29596 Increasing 90 9596 Probably Increasing 9096 No Trend 9096 and COV 2 1 No Trend 9096 and COV 1 Stable 90 9596 Probably Decreasing 9596 Decreasing The MAROS Mann Kendall Analysis Decision Matrix was developed in house by Groundwater Services Inc The user can choose not to apply one of the two statistical plume analysis decision matrices Choose Not Used in the Trend Result weighting screen If the user would liketo use another decision matrix to determine stability of the plume they would need to do this outside the software Statistical Plume Analysis 2 Linear Regression Analysis GENERAL
313. its Cleanup Goal is in mg L all rate parameters are in mg L year Recommended Frequency Based Frequency Based Well Sampling Frequency on Recent Data on Overall Data BENZENE MW 1 Annual Annual Annual MW 12 Annual Annual Annual MW 13 Biennial Annual Annual MW 14 Biennial Annual Annual MW 15 Biennial Annual Annual MW 2 Biennial Annual Annual MW 3 Annual Annual Annual MW 4 Annual Annual Annual MW 5 Annual Annual Annual MW 6 Biennial Annual Annual MW 7 Biennial Annual Annual MW 8 Biennial Annual Annual Note Sampling frequency is determined considering both recent and overall concentration trends Sampling Frequency is the final recommendation Frequency Based on Recent Data is the frequency determined using recent short period of monitoring data Frequency Based on Overall Data is the frequency determined using overall long period of monitoring data If the recent period is defined using a different series of sampling events the results could be different MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page of Project Example Location Service Station From Period 4 3 1991 to 12 19 1998 User Name Meng State Texas Normal Distribution Assumption Lognormal Distribution Assumption Sample Sample Sample Significantly Expected Significantly Expected Well Szie Mean Stdev Cleanup Goal P9We sample Size Cleanup Goal POW sample Size BENZENE Cleanup Goal mg L 0 005 Alpha Level 0
314. l 1994 point out that Aitchison s adjustment is not appropriate for log transformed data and that there is a substantial amount of spatial variation involved This section only introduces Cohen s adjustment If the percent of nondetects is between 50 and 90 use nonparametric versions of statistical interval estimates For detection monitoring the nonparametric prediction limits developed by Davis e al 1992 can be used The nonparametric prediction limit is simply the largest or next to largest concentration found in the background or upgradient measurements Complete tabulations of confidence levels for these nonparametric prediction limits for different combinations of background sample size number of future comparisons and resample plans are available in Gibbons 1994 For compliance monitoring or corrective action monitoring the nonparametric confidence interval presented in the EPA document EPA 1989 or test of proportion from other sources EPA 1992b EPA 2000 can be used This section introduces the nonparametric prediction limit If the percent of nondetects is greater than 90 a situation that is not uncommon in detection monitoring use either the Poisson prediction limits ASTM 1998 or the nonparametric prediction limits discussed above Detailed discussion of Poisson prediction limits is provided in Gibbons 1994 Loftis et al 1999 doubted the validity of using the Poisson model for modeling concentration data because the v
315. l Locations Setup screen allows the user to select the sampling locations for analysis and set the optimization parameters amp 3 Monitoring and Remediation Optimization System MAROS iojxi Access Module Potential Locations Setup Sampling locati locations are cl You may also s a potential sampling locations These potential e locations from the analysis by deselecting them btimization parameters can be set in Options BENZENE In this case study all wells will be used in the analysis and all wells are assumed removable i e can be eliminated Therefore the Selected checkbox is checked for each well so is the Removable checkbox by default In practice if not all wells are suitable for analysis eg irrelevant a aa wells or wells with susceptible data lt Back Optlons the user can deselect them Similarly if not all wells are removable eg sentry wells deselect the checkboxes in the Removable column 13 0 100 0 Selected whether or fot to include the well in analysis Select all SAMPLING OPTIMIZATION Preliminary Analysis gt gt If the user deselects some of the wells and then wants to reselect them all click the Select All button to facilitate this process Click the Options button and the Wal Redundancy Analysis Options screen will appear Here the user can set the Slope Factor SF thresholds for we
316. le Sites with both decreasing Source and Tail Results are suggested to end the sampling TABLE A 8 3 DURATION DETERMINATION FOR SITES WITH MONITORED NATURAL ATTENUATION Sampling Record Source or Tail Trend Category I or PI Trends NT or N A S Trends PD or D Trends Small Consider reassessment of Insufficient 6 more years 3more years 2 yrs network if concentrations Data continue begin to decrease sampling Medium Consider reassessment of Insufficient 4 more years 2 more years 2 TTR 10 yrs network if concentrations Data continue begin to decrease sampling Large Consider reassessment of Insufficient 2 more years 1more year 7 10 yrs network if concentrations Data continue begin to decrease sampling SAMPLING DENSITY MAROS uses a simple rule of thumb to indicate how many wells at the site may be sufficient for groundwater monitoring Users can compare the number of wells at their site to the number of wells from the rule of thumb If their site has significantly more wells being sampled then some reduction in the number of wells is possible Note that users can use the sampling optimization Sample Location wing of the software to perform a more rigorous analysis of the number of wells required for monitoring Version 2 1 A 8 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The simple rule of thumb is b
317. le Event 9 5 2 1992 MW 2 1 000E 02 1192 0 8 53E 02 5 218E 21 Yes Yes Sample Event 9 5 2 1992 MW 3 6 400E 02 1155 0 3 53E 02 1 234E 19 Yes Yes Sample Event 9 5 2 1992 MW 4 6 000E 03 1135 0 8 53E 02 2 344E 20 Yes Yes MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 3 of Project Example User Name Meng Location Service Station State Texas Observed Regression Projected Below Sampling Effective Concentration Distance Down Coefficient Concentration Detection Used in Event Date Well mg L Centerline ft 1 ft mg L Limit Analysis BENZENE Sample Event 9 5 2 1992 MW 5 2 200E 00 1194 0 3 53E 02 1 070E 18 Yes Yes Sample Event 9 5 2 1992 MW 6 1 000E 03 1267 0 3 53E 02 3 692E 23 Yes Yes Sample Event 9 5 2 1992 MW 7 1 000E 03 1277 0 3 53E 02 2 593E 23 Yes Yes Sample Event 9 5 2 1992 MW 8 1 000E 03 1245 0 3 53E 02 8 029E 23 Yes Yes Sample Event 10 1 11 1994 MW 1 2 200E 01 1177 0 4 33E 02 1 585E 23 Yes Yes Sample Event 10 1 11 1994 MW 12 5 000E 03 1090 0 4 33E 02 1 560E 23 Yes Yes Sample Event 10 1 11 1994 MW 13 1 000E 03 1125 0 4 33E 02 6 851E 25 Yes Yes Sample Event 10 1 11 1994 MW 14 1 000E 03 1088 0 4 33E 02 3 403E 24 Yes Yes Sample Event 10 1 11 1994 MW 15 1 000E 03 1000 0 4 33E 02 1 539E 22 Yes Yes Sample Event 10 1 11 1994 MW 2 2 000E 03 1192 0 4 33E 02 7 522E 26 Yes Yes Sample Event 10 1 11 1994 MW 3 1 100E 02 1155 0 4 33E 02 2 055E 24 Yes Yes Sample Event 10 1 11 1994 MW 4 1 800E 02 1135 0 4 33E 02 7 996E 24
318. lect become successively available bue D pi er hr yere AETS Main Menu The Main Menu screen allows the user to choose between performing Step 1 Data Management Step 2 Site Details Step 3 Plume Analysis Step 4 Sampling Optimization Step 5 MAROS Output Dto Managemarc IER twv ird rr d bm ttem i Site Deisis Pre irb or Pe enne emne orm cm nm at den iA n ODER e Canes itp Phime Analysis Faure Dos Consatenstin tonsea Fund aroen pstan rwr narra wx Cri taari Po renter L Ups Saeed Sampling Optimization barwa sacia upuruos Click on the Data Management button to continue The Data M anagement Menu will appear Oa ore be Lay Ket ee at Rae Nes WOE MONITORING nasab MAROS Ovnput anm d e m camem a a tren 8 usi STEP 2 DATA MANAGEMENT MENU The Data Management M enu is used to perform database operations such as importing manual data addition and archiving These operations are used initially to import site data into the software in order to perform analysis For this tutorial analytical data will be imported from an _ Excel spreadsheet TutorialExampleData xls Site details will be entered manually in later screens Data Management Menu import New Data atone reumo Cr Select Import New Data from the Data M anagement M enu Click hereto proceed Note Typically the first time through the MAROS softwar
319. led exactly as in the MAROS_ConstituentList xls file under MAROS Constituent Names The sample date should be one date in short date format i e 3 12 12004 with no ranges or partial year designations For the input file the Result field should have one number no text corresponding to the laboratory result Non detect results should be blank Trace or J flag values can be included as a number The units should be included as indicated in Table A 1 1 MAROS analysis requires detection limits for analytical data Often detection limits are uncertain or unknown for historic data In the case of missing detection limits a reasonable guess or setting one consistent detection limit for all data points is a good alternative Data flags are limited to ND for non detect and TR for trace detections ND should always be in the flag field for an empty Result cell TR corresponds to J values The Excel template file MAROS ExcellmportTemplatexls or the Access template file MAROS Accessl mportTemplate mdb should be used to create an import file for the MAROS software Each row in the import file should contain one COC for one well and date Do not Version 2 1 A 2 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE enter spike matrices or blanks Use the Constituent list found in the MAROS ConstituentList xls file for naming con
320. lls inside the benzene plume and wells on the peripheral of the plume Wells that have SF values smaller than the thresholds will become potential candidates for elimination The Area Ratio and Concentration Ratio are thresholds Monitoring and Remediation Optimization System MAROS constraining the information loss after Well Redundancy Analysis Options elimination of wells For example 0 95 for That usd the pines roces ed Chon vales Concentration Ratio means the acceptable information loss in plume average Siope Factor SiopeFactor Ratio concentration estimation is 5 at most Es MA Version 2 1 A 11 49 October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Version 2 1 October 2004 In this example set the Inside node Slope Factor to 0 20 the Hull node Slope Factor to 0 01 and both the Area Ratio and Concentration Ratio to 0 95 Click the lt lt Back button to return to the Access M odule Potential Locations Setup screen Click Preliminary Analysis gt gt to proceed and the Access M odule Slope Factor Values screen will appear The Access M odule Slope Factor Values mume Ee zax screen shows a summary of the SF values for each well The Min SF column lists the smallest SF value and the Max SF column lists the largest SF value for a well respectively The Avg SF column lists the average SF value across selecte
321. location mark in the Eliminated column Itis seen from the table that wells M W 12 MW 3 and MW 4 can be eliminated for benzene If other COCs are also analyzed there will be other tabs available for selection Note Occasionally the MAROS method will indicate with one COC to eliminate sampling that well but another COC will indicate to retain the well in the monitoring network therefore keeping the wall in the network is the best option L is ts L L LY LY L MW 5 40 70 0 0 532 Eliminated whether or not the well is eliminated from the monitoring network as a redundant well View Report Compare Across COCs gt gt Help SAMPLING OPTIMIZATION The user can choose to view the report where results are categorized by COC by dicking the View Report button S Monitoring and Remediation Optimization System MAROS ni xi Access Module All in one Results Click the Compare Across COCs gt gt to proceed and the Access M odule All in one Results The final sampling locations after considering all COCs are determined as shown in the following table A sampling location is eliminated only if it is eliminated for all COCs Eliminated status can be interpreted here as stopping sampling a certain well in the Click to view View Re View Repo report screen Ww i ap pear LociD ESCoord NSCoor
322. lts agreed with results from the California and Texas plume a thon studies Rice e al 19995 M ace et al 1997 that showed that plume length is not correlated with groundwater velocity or other hydrogeologic characteristics of the site The correlation study also confirmed that the source size is a major determining factor for plume length Because transverse dispersion is a relatively weak process Pankow and Cherry 1996 the plume width was used as an approximation for the source width As shown below there is high degree of correlation R 0 67 was found between plume length and plume width Although this may appear to be self evident it is a key conclusion in that it supports the idea that BTEX plume length is largely driven by source factors and much less by hydrogeologic factors The resulting plume length prediction equation is Plume Length ft 22 0 Plume Width ft R 20 67 This results is supported by qualitative conclusions by the California and Texas plume a thon studies Rice et al 1995 concluded These hypothetical plume length controlling variables may be source mass and passive bioremediation rate M ace et al 1997 identified other factors such as the amount of spilled fuel and natural biodegradation rate as having a greater influence than hydrogeology or previous remediation activities C Empirical Data MTBE Plumes Two plume a thon studies have been conducted on MTBE plumes one if California and one in T
323. make more analysis runs dick the lt lt Back button Click the Data Sufficiency Analysis M enu button to return to the D ata Sufficiency Analysis M enu screen ON Risk Based Power Analysis Complete You have finished the Risk Based Power Analysis for each COC for the set of parameters you selected You can proceed to the other analysis of Data Sufficiency Analysis You may also go back to change certain parameters and re run this analysis If you would like to view the report right now you can proceed to Main Menu from Data Sufficiency Analysis Menu or go back to previous screens where reports can be generated lt lt Back E N m a 9 e zZ l a Ed o Note The above described data sufficiency analyses have some implicit assumptions For the correct use and a better understanding of the power analysis method refer to Appendix 6 of the user s guide To print report graphs after all analyses are finished click Back on the Data Sufficiency Analysis M enu screen and then click the Main M enu button on the Sampling Optimization M enu screen Version 2 1 A 11 59 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE From the M ain M enu screen select MAROS output Reports Graphs to view or print reports and graphs Version 2 1 A 11 60 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND
324. me Leng ths ft Minimum 25th Percentilel Media 75th Perontild Maximum Mem n 228 970 1335 13700 11 Key results from this study are Aziz et al in review e Atsites contaminated with chlorinated ethenes only TCE or c DCE was the most likely constituent to have the longest plumes at the site TCE and c DCE had median plume lengths of 1215 ft and 1205 ft respectively e VC had the shortest median plume length of 860 ft Because the daughter product plumes were coincident or almost coincident with the parent plumes these results indicate that vinyl chloride is unlikely to be the longest plume at a site This is an encouraging result given the relatively high associated carcinogenicity of vinyl chloride Version 2 1 A 4 12 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE e The plume width in the source area or source area width was used to represent the size of the NAPL affected source area The product of the source area width and the maximum dissolved phase solvent concentration was strongly correlated with plume length This finding indicates that source characteristics including the extent of DNAPL migration are the most important factors impacting the maximum dissolved chlorinated solvent plume length e Chlorinated ethene plume lengths were moderately correlated with seepage velodty and groundwater travel distance indicating that advec
325. me and the First Moment Decreases over time Version 2 1 A 5 8 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Second M oment T rend The Second M oment trend indicates the spread of the plume about the center of mass Analysis of the spread of the plume should be viewed as it relates to the direction of groundwater flow An increasing trend in the second moment indicates an expanding plume whereas a dedining trend in the plume indicates a shrinking plume No appreciable movement or a neutral trend in the center of mass would indicate plume stability The second moment provides a measure of the spread of the concentration distribution about the plume s center of mass However changes in the second moment over time do not necessarily completely characterize the changes in the concentration distribution and the mass over time Therefore in order to fully characterize the plume the Second Moment trend should be compared to the Zeroth moment trend mass change over time refer to Figures A 5 6 A 5 8 Dissolved Mass Dissolved Mase 100 kg 10kg No Change in either Syy or Syy Figure A 5 6 Moment Analysis Mann Kendall Second Moment Trend Results No Change in trend of either Sxx or Syy both parallel and perpendicular to the plume center line Mass Decreases over time Dissolved Mass Dissolved Mass 100 kg 100 ky Syx and Syy Both Decrease Figure A 5
326. me series data by assessing the Rate of Change ROC and Concentration Trend CT of each Constituent of Concern COC and considering both recent trends and overall long term trends of the data The analysis is performed according to each COC Major steps to be followed are a Sampling Frequency Analysis In this screen define the recent period by selecting the From and To sampling events and then click button Confirm Click button Option and change the Rate of Change parameters if necessary Click Analysis to proceed b Sampling Frequency Recommendation View results for all COCs and dick button N ext to complete Data Sufficiency Analysis This sub module uses statistical power analysis to determine the cleanup status and the significance of concentration trends at individual wells and the risk based site cleanup status Statistical power and the expected sample size associated with each evaluation are provided Results from this module can be used to assess the sufficiency of monitoring plans providing auxiliary information for optimizing sampling locations and sampling frequency Major steps to be followed are a Data Sufficiency Analysis Menu There are two types of analyses to choose at this screen Power Analysis at Individual Wells and Risk Based Power Analysis Before proceeding to either of the analyses dick button Options to enter screen D ata Sufficiency Analysis Options where the user should to check or specify the parameter
327. ments that are below the detection limit DL of the analytical procedure the traditional statistical methods based on all quantified values do not work Special procedures must be used to handle these nondetects Nondetects are usually reported with the appropriate limit of detection and refer to concentrations that lie somewhere between zero and the detection limit Data that include both detect and non detect results are called censored data in statistical literature General guidelines EPA 1992 ASTM 1998 that usually prove adequate in handling data with nondetects are introduced below If less than 1596 percent of all samples are nondetects replace nondetects by half their detection limit or DL or a fraction of DL or zero and proceed with a parametric analysis such as prediction limits or confidence limits It is shown that the results of parametric tests will not be substantially affected by this simple substitution EPA 1992 If the percent of nondetects is between 15 and 50 use Cohen s adjustment EPA 1989 or Aitchison s adjustment EPA 1992 to the sample mean and variance followed by a parametric analysis Aitchison s method imputes nondetects as zero concentration Cohen s method assumes nondetects are below detection limit but not necessarily zero Both methods require that data without nondetects be normally distributed A useful approach to selecting between the two methods is described in the EPA guidance EPA 2000 Davis et a
328. method is appropriate Other Considerations Onething to keep in mind is that if the coordinates of a sampling location are not available this location will be excluded and will not be shown in the analysis The potential locations for analysis are only those with coordinates from the raw set of locations in the raw database ERPIMS or others The minimum number of wells valid for analysis is 6 If there are less than 6 wells the Delaunay method is not applied and will give no recommendations Also before applying the Delaunay method for spatial redundancy analysis it is important to select the appropriate set of wells for analysis i e only the wells that contribute to the spatial delineation of the plume For example if wells are far from the plume and contribute little or nothing to the delineation of the plume eg some sentry wells or background wells far from the plume they should be exduded from the analysis One reason not to use these wells is that these wells usually are on the boundary of the triangulation and are hard to be eliminated since the Delaunay method protects boundary wells from being easily removed The elimination status of these wells in fact should be determined from the regulatory standpoint Another well type that could be excluded from analysis is one of a clustered well set because the Delaunay method is a two dimensional method Generally only one well is picked from the clustered well set to represent the conc
329. metric prediction limit described in METHOD 2 Example Nonparametric Prediction Limit Step 1 Consider developing a nonparametric prediction limit for a facility with r 2 15 monitoring wells In n 2 10 background measurements of benzene concentrations two were above the detection limit percentage of nondetects is greater than 50 These detects are 5 ppb and 8 ppb N ondetects are reported as lt 2 ppb Step 2 Plan use One of Two Samples in Bounds plan Plan II use One of Three Samples in Bounds plan Use the largest background measurement 8 ppb as the nonparametric prediction limit for both plans Step 3 For Plan I in Table 2a on page 164 of Davis and McNichols 1994 for n 10 and r 15 the Per Constituent significance level is 0 159 which is much higher than 5 For Plan II in Table 3a on page 165 of Davis and M cNichols 1994 for n 10 and r 16 the Per Constituent significance level is 0 0428 which is within 596 Therefore Plan II is an eligible plan if the facility wide false positive rate a is to be controlled at 596 and only one constituent benzene is considered Version 2 1 A 7 35 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Step 4 If Plan must be used to control the Per Constituent significance level a at 5 then in Table 2a on page 164 of Davis and McNichols 1994 for r 15 wefind a 0 388 if n 25 This mea
330. minant migration in groundwater MAROS is a decision support tool based on statistical methods applied to site specific data that account for hydrogeologic conditions groundwater plume stability and available monitoring data This process focuses on analyzing relevant current and historical site data and optimizing the current monitoring system in order to efficiently achieve the termination of the monitoring program For example plumes that appear to be decreasing in extent based on adequate monitoring data over a several year period can be analyzed statistically to determine the strength and reliability of the trend If it can be demonstrated statistically through statistical plume analyses i e M ann Kendall Trend Analysis and or Linear Regression Trend Analysis or Moment Analysis and or External Plume Information modeling or empirical that the plume is shrinking with a high degree of confidence then future monitoring can either be suspended or reduced in scope i e from annual monitoring to biennial monitoring MAROS has the option to either use simple rules based on trend analysis results and site information or more rigorous statistical methods to determine the minimum number of wells and the minimum sampling frequency and well density required for future compliance monitoring at the site These preliminary monitoring optimization recommendations will give the user a basis for which to make more cost effective scientifically based futur
331. mination of groundwater by petroleum hydrocarbons The uncertainty that causes false positives and false negatives comes from three primary sources 1 sampling uncertainty which originates from sampling procedures 2 analytical uncertainty which governs the ability to detect and quantify the level of a particular contaminant and 3 spatial and temporal variations which control the ability to determine the significance of changes within a population using the sample data Sampling uncertainty is the result of field sampling procedures where systematic errors or random errors may exist in the processes of purging the well collecting a sample performing field tests recording the test results and preserving and transporting the sample Designing appropriate sampling routines and employing an experienced sampling team can reduce sampling uncertainty Analytical uncertainty is caused by uncertainty associated with laboratory analysis of a sample Lab analysis is affected by the detection and quantitation methods of a particular contaminant and the stability of laboratory performance Using approved analytical methods and having samples analyzed by a laboratory with rigorous quality control protocols can reduce analytical uncertainty Spatial variation and temporal variation are caused by natural variability which is inherent in any subsurface system Spatial variation refers to the different level of contamination or different degree of uncerta
332. mining a compound s toxicity ID Insufficient Data lt lt Back Below is a list of COC recommendations from the available dataset based on the Repr Cone Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG Above PRG representative concentration for each compound over the entire site The compound representative concentrations are then compared with the chosen PRG for that compound with the percentage excedence from the PRG determining the Above PRG 252635 00 87170 38 34958 64 552 21 166 11 135 52 37 28 wi si COC Decision Toxicity shows a list of COC COC Decision Toxicity recommendations from the available dataset based on the Toxicity of the compounds Top COCs by toxicity were determined by examining a representative concentration for each compound over the entire site Note The representative concentration can be skewed by high variability in the detection limit for non detects The compound representative concentrations are then compared with the chosen PRG for that compound with the percentage excedence from the PRG determining the compound s toxicity Compounds listed first are those above the PRG and areshown on theCOC Decision screen Monitoring and Remediation Optimization System MAROS cp COC Decision Prevalence Below is a list of COC recommendations from the available dataset based on the Prevalence of the compounds Total
333. mit and if the verifying resamples confirm this exceedence then the groundwater is declared contaminated N ote that different plans have different requirements For application in intra well comparisons the prediction limit is computed separately in each monitoring well for each constituent and the procedures will vary slightly Procedures intra well comparisons Step 1 Determine the facility wide false positive rate o that needs to be controlled Usually a 0 05 Sincethe prediction limit is constructed separately for each well and each constituent o the significance level for each of the k comparisons i e monitoring wells amp constituents can be calculated using the Bonferroni inequality Step 2 Determine the resampling plan that will be used Step 3 Compute the mean x and standard deviation s using the first available n measurements at least 4 as background for each well for each constituent Step 5 Consult the tables for intra well comparisons in Davis and McNichols 1987 or Gibbons 1994 with n and a to locate the factor K Step 6 Calculatethe prediction limit as follows for each well for each constituent x t Ks Step 7 Make decisions For any downgradient well for any constituent if its initial sample exceeds the prediction limit and if the verifying resamples confirm this exceedence then the groundwater is dedared contaminated N ote that different plans have different requirements The above two
334. n Delaunay Method In this screen select the series of sample events intended for analysis by defining the From and To sampling events and dick Confirm Then choose between Access M odule and Excel M odule the latter one is available only when a single sampling event is chosen for analysis b Sampling Location Analysis Access Module In this screen set up the Selected and Removable properties of potential sampling locations and if needed change the optimization parameters by clicking button Options Then dick button Prdiminary Analysis to proceed All COCs will be analyzed and several steps are to be followed to complete this analysis Or c Sampling Location Analysis Excel Module In this screen set up the Sdected and Removable properties of potential sampling locations for a COC and then click Analysis Version 2 1 10 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 2 3 The xlsD elaunay2K worksheet will pop up and the user is required to finish optimization there After sending back the results for that COC from xlsD elaunay2K by clicking Back To Access in xlsD elaunay2K this screen will reappear Run through all COCs in the same way and dick N ext to proceed Sampling Frequency Analysis This sub module uses the M odified CES method to determine the lowest sampling frequency for each sampling location The method is based on the analysis of ti
335. n The results can be printed by selecting the View Report button L MOMENT ANALYSIS Zeroth M oment Plot allows the user to view the Zeroth Moment Analysis results by constituent over time The zero moment is a mass estimate for each sample event and COC The mass estimated indicate the change in total mass of the plume over time The Zero M oment trend over time is determined by using the Mann Kendall Trend Methodology The Zeroth Moment Trend for each COC is determined according to the rules outlined in Appendix A 1 Results for the trend indude Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not Applicable Insufficient Data Other statistics displayed include the Mann Kendall Statistic S the Confidence in Trend and the Coefficient of Variation COV Refer to Appendix A 1 and A 5 for further details Version 2 1 A 11 31 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Zeroth Moment Plo To display results for benzene dick misere on the drop down arrow next to the Chemical text box Click on benzene to select Click on graph to plot the data The Zero Moment Trend over time is shown to be stable D This indicates the dissolved plume mass is decreasing over time Deaths boc a ITAL MOMENT ANALYSIS Select Next to proceed to the First M oment cas being used plots of other
336. nagement screen is used to choose between importing ERPIMS files or an Excel or Access file in the standard MAROS format see A ppendix A 1 to the database as follows E Monitoring and Remediation Optimization System MAROS Choose the type of data import to be performed by clicking on the appropriate button import New Data ss Table or Access ERPIMS Next choose database by either typing in the correct To import a file select the type of file to import Excel Table or Acce xor option Finally select files to imp files to Ider contains the TES RES SAM and LDI data files Type or saved as text files ins the TES RES SAM and LDI tables To import data into the software the name of the Excel workbook e or select the name of the Access file 1 In the Step 1 dialog box enter the file MO n UIT type for the new data N ext enter the E Poen files la ERA ae File full file path and filename of the file to C ERPIMS Tent Fies import or dick the browse button to 9 olde EE RA find the import file The Folder and E n ees ax File name you choose will appear inthe B top two boxes See Notes below for E fum me Tm ERPIMS Access and _ Excel file format names 2 In the Step 2dialog box chooseif the data file will replace all data currently in the MAROS toolbox or replace the empty MAROS files or if the data should be appended to the current file 3 Click Imp
337. nasi T here Easca irl aun feum Wfrmas Cow In gt Umm armed bem tm berr rm hment m liges ter w bom t po Fd cH mmt langasta fF mane merge Fo t Cvmudus gam a V ET SOS N Mele Mets We y em DTE peter MM KES Site Information Pris adn erecting Pi aed ate 2C une a acm ann Quma O fax Enter details z 7 here Under Hydrogeology and Plume Information enter e Seepage Velocity 92 ft yr M m e Main Constituents BTEX mene e Current Plume Width 150 ft gm rues e Current Plume Length 270 ft ioe acess gee Version 2 1 f Deme my x October 2004 m weis Mets AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE e Maximum Plume Length 270 ft e GW Fluctuations Select Y es Version 2 1 A 11 11 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE PEE Site Information 3C plow a aom aen Under Source Information select e FreePhaseNAPL Present No e Current Source Treatment Click on down arrow to obtain list of choices Scroll down to co a m select No Current Site Treatment omit Witenes C i Ane i mior t liege aec T Competet gam sn S Mele Mete Site Information w4 ara 1C phew acm aen Under Down gradient Information enter e 1000 ft for all 4 boxes Select Next to continue Th
338. nd accuracy for individual wells and the risk based site cleanup status using statistical power analysis The theoretical basis of this analysis is given in Appendix A 6 M ain Menu Returns the user to the M ain M enu screen Reports on sampling optimization results are available by choosing M AROS Output in the M ain M enu screen H elp Provides additional information on software operation and screen specific input requirements Steps for use 1 Either Sampling Location Analysis or Sampling Frequency Analysis can be performed first D ata Sufficiency Analysis red label means it cannot be accessed will become available only after Sampling Frequency Analysis has been successfully finished 2 Result reports are available either during the analysis process or by choosing M AROS O utput in theM ain M enu screen Version 2 1 51 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well Redundancy Analysis D elaunay M ethod Well Redundancy Analysis Delaunay M ethod accessed from the Sampling Optimization screen by clicking Sampling Location Analysis is used to perform well redundancy analysis by the Delaunay method This is designed to eliminate redundant locations from the monitoring network based on analysis of spatial sampling data Details of the Delaunay method can be found in Appendix A 3 Tj Pisretarmy sev Thamedaliven Cghireasstion Apatan PARAT x Co
339. ndicates whether the projected concentration at this well will be used in the risk based site cleanup evaluation The user can make selections in screen Well Selection Form by clicking button Select W ells x Projected Concentrations Estimated Centerline Regression Projected Concentrations concentrations mg L projected to the Concentrations from each sampling location are projected to the compliance boundary at or upgraidient to compliance bou ndary delineated based the Prae fesal roaa rd the aen results hort ees step Ws panes ii th d d t t Th be replaced by ts Detection Limit DL ft is less than ts DL Use Select Wells to choose the set of wells you on e downgradient receptor e want to use in the risk based power analysis in the next step distance to the compliance boundary is BENZENE ETHYLBENZENE TOLUENE vuees TOTAL introduced into the exponential model to Samping vent WOH Proecied Below Uem 2 calculate the projected concentration Sample Evert T Mn THEO g X4 Data are not available for sampling CECI aid Se Y events with less than three centerline Sample Event 2 M12 1 28E 02 Ll 4 wel Is S Sample Event 2 MAN 6 3 53E 04 v ai Sample Event 2 MW 14 8 21E 03 Ll 4 E Sale Event 2 Wa mn a Below DL Indicates whether the 2 EE L U projected concentration is below the a Back Returns the user to the P arameters for Risk Based Power Analysis screen Select Wells Opens the Well Selection Form screen wh
340. ned as wx Weighted A verage where W 2 0 xv i l Wi is the weight of the value Xi in the MAROS software high medium and low weight correspond to values 3 2 and 1 respectively Version 2 1 A 8 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE No Current Site T reatment Or M onitored Natural Attenuation Sites not currently undergoing site treatment i e no current site remediation method other than monitored natural attenuation have separate decision matrices applied Tables A 8 1 to A 8 3 FREQUENCY MAROS uses a simple decision matrix to indicate how often wells at the site should be sampled to be sufficient for adequate groundwater monitoring Users can compare the frequency of the sampling at their site to the suggested frequency of monitoring evaluated based on the decision matrix below If their site has wells being sampled at a significantly higher interval then some reduction in the sampling frequency could be applied Note that user can apply the sampling optimization Sample Frequency wing of the software to perform a more rigorous analysis of the sampling frequency required for monitoring for individual well sampling frequency recommendations Another possibility for sites with slow moving groundwater higher TTR involves a comparison study of trends for a complete dataset and a censored dataset For example the u
341. new values in the corresponding fields directly Back Keeps the changes made by the user and returns the user to the Access M odule Potential Set to default Sets all parameters for all COCs to the system default Help Provides additional information on software operation and screen specific input requirements Version 2 1 October 2004 54 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Access M odule Slope Factor V alues This screen accessed from the Access M odule Potential Locations Setup screen by clicking Prdiminary Analysis is used to display the sampling events averaged SF values of sampling locations for each COC The lumped SF value of a location provides a measure of its overall importance to a monitoring network E Monitoring and Remediation Optimization System MAROS inl xi Av i SF Displays the lump ed SF value Access Module Slope Factor Values of a location that is calculated by anging icon aped COC Sank calore vai ena SE values may Do elerated in averaging the SF values obtained in each later stey H j sampling event across all sampling BENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL events sel ected by the user LociD ESCoord NSCoord Avg SF Min SF Max SF i ee ae ee Min SF Displays the minimum SF value E 300 eo ooo ooo 005 of a location obtained from one of the M 13 650 230 0 000 0 00
342. nfirm Confirms the series of Well Redundancy Analysis Delaunay Method continuous sampling events selected by NT GSP NURSERY USERS SA NTT the user The user can also choose to Rod Wop conet o ue cites lm Dic NAMA tt Pet NNO OMIRAN analyze one sampling event t Sait vni sit t iiim Cunt Access Module Applies the Delaunay Tet fe begining av ted vnd eser hon leke method built within Microsoft Access to wr l a a optimize sampling locations suitable for multiple events 2 ete soo qing ers d h E 2 Patan andes Excel Module Applies the Delaunay 3 method built within Microsoft Excel that e NIC Fn indudes a graphical interface and flexible 2 UENIRE Ff Ene operations Data are sent to Excel Module 2 ina a and results will be transferred back This s is applicable to the analysis of only one sampling event Back Returns the user to the Sampling O ptimization screen Help Provides additional information on software operation and screen specific input requirements Steps for use 1 Select the sampling events for analysis by choosing from the From and To dropdown lists or typing in the names of the sampling events The From sampling event should be no later than the To sampling event If one sampling event is to be analyzed simply select the same sampling event in both dropdown lists 2 Click button Confirm to confirm the selection After confirmation the A ccess M odule button will be activate
343. ng Results TREND ANALYSIS Noa nee Next Takes the user to the External Plume Information Summary Screen Note If Edit individual well trends based on separate empirical studies is chosen the next screen will allow this data entry H elp Provides information on the screen specific input requirements External Plume Information Empirical Results allows the user to enter empirical results by well and constituent Montoring and RemetionOptiniztion 5s AOS 0 Dei External Plume information Empirical Results Enter the results from empirical evidence e g audere pid steed nae DETALLE Increasing I Stable S etc in the blanks ree e aded dc ee ce provided next to the well name To navigate Al Chemical NT T the results for individual constituents dick on BENZENE ETHYLBENZENE TOLUENE senenes TOTAL the tabs at the top of the screen If you would peers SoureerTail empre x like to weight all chemicals the same choose we s x the button All Chemicals Otherwise enter em al the results for each COC and each well when IP zi a you choose Individual Chemicals Ata later E lies Hr ied i this ali id will be able to weight DICANT NM these lines of evidence S Ww ns lt lt Back Next gt gt Empirical results should be developed on the n mea em basis of data from previous similar site studies eg plume a thon studies such as the Lawrence Livermore stu
344. ng program For example if one were to choose a candidate or several well to remove from the monitoring program you could go back into the historic data and perform moment analysis on the data set minus the candidate well If similar zeroth first and second moments were generated then removing the wells would be not significantly effect the future delineation of the plume through a revised groundwater sampling network Validation of removing a well from a monitoring program can be especially helpful when the water analysis alternates between non detect and detection of very low concentrations References Chiang C Y P D Petkovsky et al 1995 Remediation and clean site dosure of a contaminated aquifer at the Wexford CPF Petroleum Hydrocarbons and Organic Chemicals in Ground Water Prevention Detection and Remediation Houston TX Freyberg D L 1986 A natural gradient experiment on solute transport in a sand aquifer 2 Spatial moments and the advection and dispersion of non reactive tracers Water Resources Research 22 13 2031 2046 Version 2 1 A 5 10 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Knox R C 1993 Spatial moment analysis for mass balance calculations and tracking movement of a subsurface hydrocarbon mound Groundwater Monitoring and Remediation Summer 1993 139 147 Rasmuson A 1985 Analysis of hydrodynamic disper
345. nge for a COC into three levels Low Medium and gh user provides a Site Specific value overtin tine period The unt for Cienup Goslis mpit Tha unis for rate of chanar According to the definition the default eme Medium rate is 0 005 mg L year and coc name Cleanup Goal LowRate MediumRate High Rate the default Low Medium rate is me z o Lt on L 2 0 002540 005 2 0 00375 amame etc For details on how to set these parameters refer to the corresponding parts in chapter M AROS D amp ailed Screen Descriptions Sampling Frequency Determination Options 0 7 035 0 7 14 In MAROS the determination of sampling frequencies by using the Modified CES method starts with screen Sampling Frequency Analysis which is introduced in chapter M AROS D etailed Screen D escriptions Version 2 1 A 9 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 1 DETERMINE FREQUENCY BY RECENT TRENDS Frequency can be determined by results from both recent trends and overall trends In this step we need to determine the frequency based on recent trends using the procedures shown below Y ROC L R345 ROC lt M Y CT is Incr ProbIncr with ROC gt LM Y E or NoTrend with ROC gt LM CT is Incr ProbIncr or NoTrend with ROC gt MH Then similar procedures are used to determine the sampling f
346. ns 25 background measurements are needed to meet the 5 requirement Example Poisson Prediction Limit Step 1 Consider developing a Poisson prediction limit for a facility with k 15 monitoring wells In n 2 12 background samples for which a 32 constituent VOC scan was conducted there were y 6 detections The percentage of nondetects in this case is far less than 10 6 12 32 0 016 1 6 Step 2 Use the One of Two Samples in Bounds plan and set the facility wide false positive rate o 0 05 The a for this resample plan is a l a 24 1 095 5 0 058 Since the minimum of 0 01 and 0 058 is 0 01 the significance level associated with each individual test is therefore o 0 01 The z value associated with o 0 01 is 2 236 Step 3 The Poisson prediction limitis 2 PoissonPL i t 2290 2290 je 12 12 2x12 12 This Poisson detection limit can also be translated into a total of 2 452 x 15 36 detections out of scans of 15 future samples For comparison the background samples have only 6 12 2 0 5 detections per scan 2 6 2 452 detections per scan METHOD 10 METHODS FOR DEALING WITH SEASONAL EFFECTS AND SERIAL CORRELATION When the data exhibit regular seasonal patterns or significant serial correlation the assumption of independence is violated and adjustments must be taken to remove these effects As is described in METHOD 8 adjusting for seasonal effects is usually achieved by
347. nson M 1998 and Groundwater Services Inc www gsi net com rbcapol igi x Monitoring and Remediation Optimization System MAROS External Plume information Screening Criteria State Texas Regulatory Authority Texas Natural Resource Conservation Commission TNRCC Citation TNACC Petroleum Storage Tank PST Division Interoffice Memorandum 2 10 97 Groundwater Remediation Criteria GW Exposure Pathway Screening Groundwater pathway may be excluded from further consideration if NAPL has been recovered to extent practicable there are no existing impacts to water supply wells or surface water in excess of applicable limits and the following conditions are met For GW Plumes Delineated to Drinking Water Limits If no future groundwater use anticipated in plume area and maximum plume concentration lt Class Ill ground water limits e g benzene lt 0 14 mg L NFA for groundwater Otherwise show plume stable and then NFA For GW Plumes Not Delineated to Drinking Water Limits If no existing water supply wells or surface water discharge within 1200 ft and no anticipated use within 1200 ft NFA for GW if maximum plume concentrations lt Class III limits e g benzene lt 0 14 mg L If Class Ill limits exceeded show plume stable and then NFA Notes DTW Depth to water GW Groundwater LUFT Leaking underground fuel tank LUST Leaking underground storage tank NAPL Non aqueous phase liquid N
348. nstituents Chlorinated Solvent Curent Source In situ Biodegradation z No Treatment No Current Site Treatment Down gradient Information o9 Distance from Source to Nearest Ed Downgradient receptor T00 Imi Downgradient property line 100 2 Ww 7 Main Menu Next gt gt Help Distance from Edge of Tail to Nearest ft Downgradient receptor 100 f ft Downgradient property line 100 ft Fill in the appropriate information within each field Fields such as State and Current Source Treatment have dropdown boxes to assist in data entry Note All fields on this form are mandatory entry The user will be prompted if the fields are not filled in Under the Downgradient Information section a non zero number is required in the distance to receptor cells The number can be small 1 or negative in the event the plume has extended beyond the possible point of exposure Next Takes the user to the Sample Events Main Menu Takes the user back to the M ain M enu screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 20 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Sample Events accessed from the Site Information screen allows the user to define sample events and dates to be used for graphing and data consolidation For this section a sample event is defined as
349. nter for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Data Consolidation Cie a ANALYZE INDIVIDUAL WELLS ANALYZE SAMPLING SAMPLING OVERALL PLUME FREQUENCY LOCATION Statistical Mament Modified CES Delaunay MM b Analysis Power Analysts Triangulatian Overall Fiume Summary FIGURE 1 MONITORING AND REMEDIATION OPTIMIZATION SYSTEM MAROS PROGRAM FLOW FUNDAMENTALS OF COMPLIANCE MONITORING Remediation monitoring of affected groundwater is a significant cost driver for future environmental restoration activities These monitoring systems whether applied for process control performance measurement or compliance purposes referred to as long term monitoring are dictated by RCRA CERCLA and UST programs Although an individual long term monitoring data point is relatively small the scale of the required data collection effort and the time commitment makes the cumulative costs very high Consequently improving the efficiency of these systems through improved methodology for developing future long term monitoring plans has the potential for substantial cost savings Version 2 1 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The features available in the MAROS software are designed to optimize a site specific monitoring program that is currently tracking the occurrence of conta
350. ntroduced above the sequential t test and the Student s t test Results from the two tests on a same dataset have the following relationships 1 Not Attained always corresponds to Not Significant 2 Attained always corresponds to Significant and 3 Cont Sampling may correspond to Not Significant or Significant because of the difference between the two tests Significance result from the Student s t test can be used as a secondary indication of cleanup status Power and the expected sample size from the optional analysis can be used to indicate data sufficiency RESULTS AND INTERPRETATION OF RESULTS The results of individual well cleanup status evaluation are presented in the Individual Well Cleanup Status Results and Individual Wal Cleanup Status Optional Power Analysis screens described in the MAROS Detailed Screens Description chapter Power analysis parameters involved in the evaluation see screen D ata Sufficiency A nalysis Options indude Cleanup Goal The cleanup standard for a COC also called the primary remediation goal PRG The default cleanup goal for a COC is its MCL if available in MAROS database Results shown are based on yearly averages NO BENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL Well Name Sample Cleanup E Distribution amp ssumption Size Achieved MW 15 4 Cont Sampling Normal MyN 2 8 Cont Sampling i MAG 5 Xibahed View Normal M4 9 Attained MW 5 6 Attained View Log MVY 6 8 C
351. nual Sample detection limits are required but can be estimated if the information is not available Laboratory Quality Assurance Quality Control data matrix spikes and field blanks are not required Water quality parameters such as pH and conductivity are not required 2 Aquifer and general plume characteristics should be identified before plume analysis begins The MAROS tool requires a general value for aquifer seepage velocity porosity saturated thickness and flow direction A MAROS file can be run multiple times using different aquifer parameters to examine sensitivity to varying hydraulic characteristics within an aquifer The plume length and width as well as an approximate source location and estimate of distance to potential receptors are also required Groundwater sample locations should be identified as being in the source or tail region of the plume 3 If you arerunning MAROS for the first time it is advisable to start with a limited data input set until you become familiar with the software MAROS can examine data for up to 5 constituents at once but a simple file with one to three constituents is easier to handle for a preliminary run How can I import enter groundwater data into M AROS TheMAROS Software allows manual data entry or importation of data into the software To import data within the software 1 Main Menu From the M ain M enu select Data M anagement by clicking on the button next to the label This will
352. o PR dod Smp M Statieticn Plume Analysis Mendimi smi rm prm ste Mere anm Spatial Moment Anatyors saai mem ti hom Auf Estoma F gne rtormatior fran Ple o6 Cnt m cem Pm erm Ce ts dac m s few MAROS Analysis Morsy er veg in committee ba sad wat Ox m 2 External Plume Information M odeling Results allows the user to enter modeling results by well and constituent or for all source or all tail wells e g Increasing I Stable S etc wa Modeling results should be taken from fate and transport models that take site specific data and predict the ultimate extent of constituent migration either for natural attenuation process or site undergoing remediation For this tutorial there are no additional modeling results The option No separate modeling studies have been performed should be already selected Select Next to proceed to the External Plume Information Empirical Results screen Click here to proceed TREND ANALYSIS tehut Ae o Version 2 1 A 11 36 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 3 External Plume Information Empirical Results allows the user to enter empirical trend information by well and constituent or for all source or all tail wells The rationale and limitations to this approach is outlined in Appendix A 4 This portion of the softwareis an optional utility which
353. o Qe ree emet 1an maneta ha un vini wm ve mitem groundwater flow viet tasa ETE 33 The Second Moment Trend of the spread of the plume over time is shown to be increasing I This indicates that although the concentrations are decreasing the plume is spreading over time AL MOMENT ANALYSIS Select Next to proceed to the Spatial Cli E ick hereto i M oment A nalysis Summary screen proceed 9 eer Sey Note If more than one COC was being used plots of other chemicals can be obtaind Click hereto plot Chemical drop down box at the top of the screen followed by selecting the Graph button The graph type can be specified as Log or Linear The graph can be printed by selecting the View Report button 8 Spatial Moment Analysis Summary allows the user to view the Moment Analysis M ann Kendall results by constituent Version 2 1 A 11 34 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE v SPATIAL MOMENT ANAL Y vn t Reve trian Dpieriosrer Eyton AEDS Sia Displayed are the spatial moment analysis Spatial Moment Analysis Summary results for benzene Tin Mae acd Nim Fa ips ind arde ai gar deti uota idle caie ars ananemed vacan Cant ans mn fom he mter be cee omnt Click Next to proceed to the M oment Analysis Complete screen Sane ras al id ot t Fon oaoa Fandom ean ur arat gue d Fon ee n mm Note
354. o Wilk test are available in Wilk and Shapiro 1968 and Gibbons 1994 The Shapiro Wilk test can be used for sample sizes up to 50 When the sample sizeis larger than 50 a slight modification of the procedure called the Shapiro Francia test can be used instead The Probability Plot Correlation Coefficient Filliben s statistic test is roughly equivalent to these two tests A brief evaluation is provided in EPA guidance EPA 2000 as to the scope of use and performance of each of these alternatives Since these test statistics i e W statistic and Filliben s statistic are difficult to compute manually this section only presents the construction Version 2 1 A 7 24 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE of a normal probability plot Readers can refer to EPA guidance EPA 1992 EPA 2000 for detailed procedures for these quantitative tests Procedures Normal Probability Plot Step 1 Arrange data in order from smallest to largest and denote them as xi i 1 2 n The data can be measurements from a single well or from a group of wells Step 2 The cumulative probability corresponding to each measurement is computed as i i where i is the order of the ith smallest measurement n Step 3 Compute the normal quantiles corresponding to the cumulative probabilities obtained from Step 2 as y 4 6 where amp denotes the inverse of
355. o analyses cleanup status evaluation for individual wells and risk based power analysis for site lata cleanup The two analyses are independent and the sufficiency analysis parameters such as Cleanup Gd Click to p erform Select Any Analysis to Proceed A n al ysi S 1 Analysis 1 D Power Analysis at Individual Wells Evaluation of cleanup status at individual wells based on observed concentrations 2 Analysis2 Risk based Power Analysis choose to perform the power analysis at ee oe individual wells and or risk based power analysis 2 l l l Ej Back Options Help 11 View or modify analysis parameters 12 Version 2 1 A 11 56 October 2004 by clicking the Options button The Data Sufficiency Analysis Options screen will appear The Cleanup Goal is the site specific or risk based cleanup goal for a COC as described earlier The Target Level is the concentration level the remediation is aimed to achieve which should be smaller than the Cleanup Goal indicating that the concentration level after remediation is below the cleanup level The Target Level IS by default set as 80 of the Define the Target Level used in the individual well celanup status evaluation Alpha Level Cleanup Goal The Alpha Level isthe ai eof staneta test Target poner eee and the typel error or significance level used in a Monitoring and Remediation Optimization System MAROS
356. ober 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Plume Analysis The Plume Analysis M enu screen serves at the center of the trend analysis user interface The user progressively steps through the Long Term Monitoring Plume Analysis process by navigating through the options displayed As individual steps of the process are completed options to select become successively available The Plume Analysis Menu screen allows the Plume Analysis Menu user to choose between performing The Plume Analysis Menu screen serves at the center of the Plume Analysis interface The user should progressively step through the Plume Analysis process by navigating through the options displayed As individual steps of the process are completed options to select become successively available Proceed Through Steps 3a e e Step 3a Data Consolidation Step 3a ii i H H H De ATE e Step 3b Statistical Plume Analysis for J and ND flags as well as consolidating duplicates Step3b Statistical Plume Analysis e Step 3c Spatial Moment Analysis Mann Kendall and Linear Regression Statistical Plume Analysis x S Saas are e Step 3d External Plumel nformation Spatial Moments Plume Analysis e Step 3e M A ROS A nalysi S Step 3d External Plume Information oee Selet the desired option by clicking the Siero eT E applicable button Proceed through Steps 3a statistical plume data 3e Main Menu
357. ocation ESCoord NSCoord Eliminated H Back Returns the user to the Access ZE 850 2 0 d M odule Results by COC screen MW 14 102 0 20 0 LI o T A dus View Report Generates a report with N RE ae z the all in one sampling location MN 4 55 0 37 0 H H gt z optimization results This report can be O MN wo 50 d x viewed or printed The user can go back iudi ee ep hae EE to rerun the analysis by changing a parameters or selecting a different series z lt lt Back View Report Next gt gt of sampl i ng events Next Proceeds to the Well Sufficiency Analysis screen Help Provides additional information on software operation and screen specific input requirements Version 2 1 57 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well Sufficiency Analysis New Locations This screen accessed from the A ccess M odule All in one Results screen by clicking Next is a control screen for applying a Microsoft Excel module that is used to perform well redundancy analysis that is recommending potential areas for new sampling locations x COC Seleds the COC you want to 2 Mentenns ancl Varssasdialiss Diglrresatien Syster MAALIIN Well Sufficiency Analysis New Locations analyze from the dropdown list S ugoex possible hen sa ics osocre Dave on esunebon gxcertornt ropemcented Or Gaps C A ERE PISA SM CH iar VERE Wr A Selected Decid
358. oceed click Next The site details portion of the software is complete 4 Main Menu From the M ain M enu select Trend Analysis by dicking the button next to the label This action will takethe user to the PlumeA nalysis M enu screen 5 Plume Analysis Menu From the Plume Analysis M enu select Data Reduction by clicking the button next to the label This action will take the user to the D ata Reduction Part 1 of 2 Screen 6 Data Reduction In each screen select the information that will define the data you would like to analyze click Next to continue to the next screen First enter the period of interest as well as data consolidation options on the D ata Reduction Part 1 of 2 screen Next define delimit the data on the D ata Reduction Part 2 of 2 screen Continueto the Reduced D ata Table screen to view the results of data consolidation To proceed dick Next The data reduction portion of the software is complete Version 2 1 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 7 Plume Analysis Menu From the Plume Analysis M enu select Statistical Plume Analysis by dicking the button next to the label This action will take the user to the M ann Kendall Statistics screen 8 Statistical Plume Analysis In each screen view the information from both the Mann Kendall and Linear Regression Statistical Analyses click Next to continue to the next sc
359. od of least squares is used to obtain estimates of the model parameters a b that minimize the sum of the squared residuals S or the measure of the distance between the estimate and the values we want to predict the y s i l The values for the intercept a and the slope b of the line that minimize the sum of the squared residuals S are given by Y 3v y psc and a y bx n Yu x i 1 where x and y arethemean x and y log COC concentration values in the dataset In order to test the confidence on the regression trend there is a need to place confidence limits on the slope of the regression line In this stage of the trend analysis it is assumed that for each x value the y distribution is normal A t test may be used to test that the true slope is different from zero This t test is preferentially used on data that is not serially correlated or seasonally cyclic or skewed The variance of y 6 is estimated by the quantity S KA where this quantity is defined as n Yi XY 2 _ ist ee n 2 where n is the number of samples The estimation of the standard deviation or standard error of the slope s e b is defined as To test significance of the slope calculated the following t test result can be used to find the confidence interval for the slope Version 2 1 A 2 8 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE b
360. of a Delaunay triangle for estimating its average SF value Version 2 1 October 2004 A 4 9 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE In MAROS a Microsoft Excel module xIsN ewLocation is developed to implement the method To help visualize the analysis results in xIsN ewLocation potential areas the triangles for new sampling locations are marked by blue dash lines A colored label is placed around the center of each triangle to indicate the estimated SF level at a potential area The estimated SF values are classified into four levels S Small 0 3 M M oderate 0 3 0 6 L Large 0 6 0 9 and E Extremely large 20 9 Those potential areas with the estimated SF value at the Extremdy large or Large level are candidate regions for new sampling locations New sampling locations can then be placed inside these regions eg at the centroid of a triangle region Refer to the MAROS D etailed Screens Description chapter for details on the usage of xIsN ewLocation It is emphasized that recommendations from the well sufficiency analysis are derived solely from the spatial configuration of the monitoring network and the spatial pattern of the contaminant plume No hydrogeologic conditions are considered in the analysis Therefore professional judgement and regulatory considerations must be used to decide whether an area for new sampling locations recommended using the above
361. ogic Water elevations do not Install additional wells or positioned for assumptions to estimate flow producea unique contour find existing wells screened in identifying directions Use ground penetrating pattern too few wells screened thesame water bearing zone groundwater flow radar GPR if possible to evaluate in the same zone wells installed directions the validity of the assumptions essentially along a line 2 Wells not Estimatethe distancethe Contaminant concentrations do Install additional wells or positioned for contaminant plume may have not produce a unique contour find existing wells screened in evaluating the migrated from the site based on site pattern the contamination the same aquifer In some extent of history hydrogeology and plume does not appear to be cases soil gas or EM surveys contamination contaminant geochemistry Use aerial images or electromagnetic conductivity EM and soil gas surveys to check estimation related to the suspected source or the contaminant pattern suggests undocumented sources can be used to augment monitoring well networks 3 Screen settings not correctly selected Use background geologic and geochemical information and geophysical surveys to project contaminant flow Compare information to on site soil samples collected from boreholes Water elevations appear to be anomalous apparent flow directions seem illogical or overly complex information for on s
362. ojected mean concentration level at the compliance boundary is truly below the cleanup goal A value close to 1 0 may indicate that the data are distributed very close to the sample mean or the coefficient of variation is very small a small variability A value close to 0 indicates the opposite requiring more sampling locations for the analysis to reach a higher power A value greater than the expected power means that data from the monitoring network provides sufficient information for the risk based site cleanup evaluation BENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL rH aS Sample Event 2 9 Attained 0 985 5 Lognormal Sample Event 3 9 Attained 1 000 lt 3 Sample Event 4 10 Not Attained S S Sample Event 5 g Attained 0 955 6 View Sample Event 5 8 Attained 0 991 Normal Sample Event 7 11 Attained 1 000 4 Sample Event 8 10 Not Attained 0 283 51 View Sample Event 9 10 Attained 1 000 3 Log N C not conducted due to insufficient data S E sample mean significantly exceeds cleanup goal Figure A 6 6 Individual well concentration trend results Expected Sample Size is the number of projected concentrations i e the number of wells required to achieve the expected power e g 0 80 with the variability shown in the projected concentrations The smaller the value the smaller the data variability and the higher the statistical power If the expected sample size is smaller than the sample size the monitoring netwo
363. olidation is recommended for datasets with greater than 40 sample events Version 2 1 29 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D ata Reduction D ata Reduction Part 2 of 2 accessed from the D ata Reduction Part 1 of 2 screen allows the user to consolidate the data based on concentration parameters chosen Select the factors by which you would liketo limit the data NM ed NOn PER ORE Duplicates Choose the option to consoli i date duplicates Note Duplicates are samples pistes binges pal vnik etc that have the same constituent date and well name If you have given the same effective M es date to two samples they will be consolidated Non Detect ND 1 2 Detection Limit Detection Limit Fraction of Detection Limit Specified Detection Limit C Maximum Average as duplicates Non D etect N D Choose the number value Toce TAN you would like to represent a non detect result ETHYLBENZENE Actual Value First Result Detection coc Limit mg L TOLUENE C 172 Detection Lini in the data If you would like to apply a XYLENES TOTAL oe EE specific detection limit for each chemical choose Specified Detection Limit The suggested detection limit is the minimum lt lt Back Next gt gt Help detection limit Note Changes in detection limit over time can create artifacts such as false
364. on This will work only when this worksheet is loaded through MAROS Stop the program by clicking the Terminate button in the Well Locations chart sheet Go to Step Lif you want to re analyze The xlsD elaunay2K worksheet will remain open until the user doses it All the results and graph output are kept if the user chooses to save the file before closing it The graph output in the plot area is similar to the screen shot shown below WARNING Version 2 1 October 2004 S et as n gura lt i i og POS i EOS ao eee Me 2 uM UN a RO rs x jum SON a a M x e deed E ON pis d 2 jg es aed ae j SM i Pe x T Lo A a ps v fiz2a A n 7 TEN x s hne m ite s S z SM i Fae NEAR y rao E ee cci eee MAE cm E io pu M m x cs 2 pog MT UD E et Chay Spe TEN MP aU QM Wee ee ae Te ie e NIS ae B ks i E EN quce un Han A pe ES p ee pouce t MN e e 2a oa eros Wee ae eo c c M NEEDS rs ae iE 2 QUE i Poe n e Ea Sea eer ES ee C I eee puces d T Cd ay NE a DON eke seh Se A ee DD n a Sanath repu cs Ea FTO y is J i Do not change the name of worksheet xIsD elaunay2K or move it to other folders Air Force Center for Environmental Excellence 67 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Excel M odule All in one Results This screen accessed from the W dl Redundancy Analysis Excel M odule screen by clicking N ext is used to display the all in one
365. on 3 Data Sufficiency Analysis 9 Murcia c ear If multiple sampling events are Es selected only the Access Module E Est E button will be activated If only one sampling event is selected i e select E Monitoring and Remediation Optimization System MAROS x the same event from both the From and To dropdown lists the Excel Module button will also be activated The Exca Module illustrates graphically how the Delaunay method works and gives the user more control Well Redundancy Analysis Delaunay Method Perform well redundancy analysis by using the Delaunay method to eliminate redundant locations that have Slope Factor values less than certain thresholds You may choose to use either the graphical method or the non graphical method Confirm Select sampling events for analysis Select the beginning and ending sampling events from below e Rom SmpetwniS of the optimization process 2 Usually multiple sampling events are used to detect the sampling locations that are redundant throughout a period rather than at a point in time In this case study multiple sampling SAMPLING OPTIMIZATION To analyze a single sampling event choose the same event in both dropdown lists N Perform analysis Non graphical method realized within Microsoft ule gt gt Access for multiple sampling events analysis yu Access M Graphical method realized within Mic
366. on Units of Measure 20 ELFLAG More Current Elevation Available in 21 WINTDEPTH Borehole Depth 22 BHDIAM Borehole Diameter 23 BHANGLE Angle of Borehole Drilling 24 BHAZIM Azimuth of Borehole Drilling 25 DATUM Geodetic Datum Identifier 26 STPZONE Coordinate Zone for Geodetic Datum attribute 27 STPPROJ Geographic Projection 28 UTMZONE Unit of Coordinate Zone for Geodetic Datum attribute 29 GEOLOG References Drilling Logs 30 MAPID Map Identifier 31 LOCDESC Location Description Version 2 1 A 2 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 1 3 REQUIRED FIELD FORMAT FOR TES IMPORT FILES TESTING RESULTS Column Number Field Name Description 1 SAMPLESEQ Sample Sequence N umber 2 ITESTSEQ Test Sequence N umber 3 LABCODE Laboratory Company Code 4 ANMCODE Analytical Method Code 5 EXMCODE Extraction M ethod Code 6 RUN NUMBERRun Number 7 LA BSA MPID Laboratory Sample Identification 8 LABRECDATE Date time of Reception by Lab 9 LABRECTEMP Sample Temperature at Reception 10 LABRECUNITS Celsius or Fahrenheit 11 EXTDATE Date time of Extraction 12 LCHDATE Date time of Leaching 13 LCHMETH Method of Leaching 14 LCHLOT Designator of a Group of Samples Leachated Together 15 ANADATE Date time of Analysis 16 ANALOT Designator of a Group of Samples Analyzed Together 17
367. on divided by sample mean and Confidence in Trend to determine the trend category For the details of this statistic refer to the corresponding part in Appendix A 1 The Rate of Change ROC parameters used for determining the linear trends of COC were generalized to include all possible ranges The ROC parameters fall into five categories Low L Low M edium LM Medium M Medium High MH and High H The ROC is simply the slope of the fitted line by linear regression The user is required to define three ROC parameters the Low rate Medium rate and High rate The other two rates Low Medium and Medium High will be automatically determined The term Cleanup Goal or PRG Primary Remediation Goal is used in MAROS to stand for MCL By default the Low rate is defined as 0 5PRG year the Medium rate is defined as 1 0PRG year and the High rate is defined as 2 0PRG year for all COCs The Low M edium rate is defined as the half way between the Low rate and the Medium rate as is the same for Medium High rate The user should provide more accurate values for these ROC values if accurate classification is available from the hydrogeologic setting in the studied site Theunit of the ROC parameters is mg L year For example in the right screen the Monitoring and Remediation Optimization System MANOS Cleanup Goal for Benzene is 0 005 mg L Then the default Low rate is 0 5 x 0 005 0 0025 mg L year unless the Classify the rate of cha
368. ont Sampling MW 8 Not Attained Optional Power Mw 8 8 Cont Sampling l Analysis N C not conducted due to insufficient data FigureA 6 2 Individual well deanup status results based on the Sequential t test Targ amp Levd The concentration level of COC in the well after attaining the cleanup goal The default value for this parameter is set to 0 8 times the cleanup goal This parameter is only used in the sequential t test The difference between the Cleanup Goal and the TargeLevd is the minimum detectable difference the sequential t test is supposed to detect AlphaLevd The significance level type error or false positive error rate used for all statistical tests in MAROS Data Sufficiency Analysis The default value for this parameter is 0 05 Version 2 1 A 6 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Targ amp Power The desired statistical power of all statistical tests in MAROS Data Sufficiency Analysis The default value is 0 80 Results from the sequential t test and the optional power analysis are illustrated in Figure A 6 2 and A 6 3 respectively Cleanup status power and expected sample size for each well with at least four samples yearly averages or original data are calculated for two distributional assumptions normal and lognormal When there are less than four data records NC is displayed in result fields indicating the analysis is
369. or example the Linear Regression Trend is shown to be decreasing D in the box in the left aiat frg id Kard lr e Dy nl in hee em AMARC Linear Regression Plot fave oec ctm d onmia eee mmm eme nip wc lo sasana syl nal daoia ie De parad 3 ases EOT D ret Top n m e 4 y PPE D Lite hand bottom corner Select the Next button to continue to i a S S ie the Trend Analysis Statistics Summary Graph Hew Date ree P n hem screen Click here to remos proceed wea so Vw Hepat Wem Version 2 1 A 11 27 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Note As discussed previously plots of other wells and chemicals can be obtained using the Well or Chemical drop down boxes in the top of the screen followed by selecting the Graph button The graph type can be specified as Log or Linear The graph can be printed by selecting the View Report button 6 Trend Analysis Statistics Summary by Wal allows the user to view the Mann Kendall Trend Analysis and Linear Regression Analysis results by well and constituent Statistical analysis for the benzene data is displayed Ke ed Rd Wes Dee Select Next to proceed to the Statistical hee m nm PlumeAnalysis Complete screen d ce tir Put 1 b 1 is tn 1 Click here to eie Ay AE eer Tre erri ase Se aunsS F proceed N ote If more t
370. orce Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE x a where a is the desired facility wide false positive rate Step 5 The use of this nonparametric prediction limit for future comparisons is the same as in the parametric prediction limit described in METHOD 2 Procedures Poisson Prediction Limit Step 1 Let y bethetotal number of detections analytical hits from multiple constituent scans e g for VOCs of n background samples Denote the number of future measurements i e number of monitoring wells or number of measurements in one well as k Step 2 Determine the resample plan and the facility wide false positive rate o Calculate the significance level o associated with each individual test as the minimum of 0 01 or one of the following a 1 105079 for One of Two Samples in Bounds plan a 1 10 0 for One of Three Samples in Bounds plan a 1 a 41 2 for First or Next Two Samples in Bounds plan Step 3 Compute the Poisson prediction limit as 2 2 PoissonPL E 0 n _ detections per scan n 2n n 4 wherez is the 1 100 upper percentage point of the standard normal distribution Step 4 If the average number of detections per scans of the k future samples is greater than this prediction limit it may indicate exceedence but must be verified by resamples The verification procedure is same as in the para
371. ordinates if they were not defined in the import file Well coordinates are mandatory and should be in feet e g State Plane coordinates or arbitrary site coordinates Well coordinates will have already been eI Well Coordinates specified Select the button Well Map to view the well location map Otr rm rorem mo rime inm mes ee Wit md M AA eerte ni Pm pom en mm ona Tm Perm Well Locations allows the user to review the well coordinates in their relative locations Wei Locations vogue nom Trace en eco aed Y vato mit ye e rd em hg ot wm Select Back to continue The Wal Coordinates screen will re appear SITE DETAILS Click here From the Well Coordinates screen select Next pm to continue The Constituents of Concern Decision Well Coordinates screen will appear Otr terret m enm mee inm m mces aste T Rien M Aene rim M ona Them Pee ie whine Tec HDi Y Oee m m Version 2 1 A 11 14 October 2004 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Click hereto proceed Version 2 1 A 11 15 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 6 CONSTITUENTS OF CONCERN Constituents of Concern Decision allows the user to define up to five constituents to be evaluated at the site Typically the only constituent which needs to be defined is the indicator for the site The site used
372. oring standards attainment decision than the fixed sample size test under the same levels of false positive and false negative rates Method 5 Where To determine whether This method does not require a test for Mann Kendall needed the trend of normality of data Test results of the GSI test for trends concentration data vs style Mann Kendall test are classified more time is increasing reasonably decreasing or stable etc 1 Distinguish the type of monitoring program in which the statistical tests will be used and set up the correct null hypothesis Refer to tables A 7 4 A 7 5 and A 7 6 2 Estimatethe percentage of nondetects in the observations and choose the correct category of statistical approach based on the percentage of nondetects Refer to Table A 7 19 This may need to be done on a well by well basis Version 2 1 October 2004 A 7 41 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 3 Adjust for seasonal patterns and serial correlation if needed before testing for the distributional assumption Refer to tables A 7 20 and A 7 18 This may need to be done on a well by well basis 4 Evaluate the possibility of using intra well analysis to avoid the influence of spatial variability based on the sufficiency of data from the historical database Tests for homogeneity of variance refer to Table A 7 18 Method 17 can be performed to determine the
373. ort to proceed with importing the fileto the existing database A dialog box will appear with the number of wells and the date range of the data check these data to make sure they are consistent with your import file Too few wells or too few dates means that some of your data is not importing properly and you may need to repair your input file Back Takes the user back to the D ata M anagement screen Help Provides information on the screen specific input requirements NOTES To import an Excel 2000 spreadsheet 1 Typeor select the name of the Excel workbook 2 Theimport option requires an Excel file format with fields identical in name and structure to those outlined in Appendix A 1 Each field must have the columns filled in Do not import files with missing data this will result in incorrect data evaluation within the software The columns must include the field names in the first line The template file MAROS ExcellmportTemplatexls is provided with the software with example data Also alist of permissible constituent names is found in the file MAROS ConstituentList xls To import an Access 2000 Table 1 Typeor select the name of the Access File 2 Theimport option requires an A ccess Table format with fields identical to those outlined in Appendix A 1 Only one import table should be in the Access file Each field must have the columns filled in Do not import files with missing data this will result in incorrect data evalua
374. ot Applicable N A Due to insufficient Data lt 4 sampling events COV Coefficient of Variation MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 1 MAROS Mann Kendall Statistics Summary Well MW 13 Time Period 10 4 1988 to 12 19 1998 Well Type T Consolidation Period No Time Consolidation COC BENZENE Consolidation Type Median Duplicate Consolidation Average ND Values Specified Detection Limit J Flag Values Actual Value Date PEE WR Na Wian egaat astal 64 Confidence in a Trend E 100 0 c 2 Coefficient of Variation 0 01 t 1 01 o 1 o o O 0 001 Mann Kendall Concentration Trend See Note 0 0001 5 Data Table Effective Number of Number of Well Well Type Date Constituent Result mg L Flag Samples Detects MW 13 if 10 4 1988 BENZENE 3 5E 02 1 1 MW 13 11 17 1989 BENZENE 2 6E 02 1 1 MW 13 i 3 1 1990 BENZENE 4 9E 02 1 1 MW 13 T 5 31 1990 BENZENE 5 2E 02 1 1 MW 13 T 9 13 1990 BENZENE 1 5E 02 1 1 MW 13 T 4 3 1991 BENZENE 1 9E 02 1 1 MW 13 T 7 10 1991 BENZENE 2 9E 02 1 1 MW 13 T 10 3 1991 BENZENE 3 5E 02 1 1 MW 13 T 5 2 1992 BENZENE 8 0E 03 1 1 MW 13 T 1 11 1994 BENZENE 1 0E 03 ND 1 0 MW 13 T 5 28 1996 BENZENE 1 0E 03 ND 1 0 MW 13 T 6 27 1997 BENZENE 1 0E 03 ND 1 0 MW 13 T 12 10 1997 BENZENE 5 2E 04 1 1 MW 13 T 6 19 1998 BENZENE 1 0E 03 ND 1 0 MW 13 T 12 19 1998 BENZENE 1 0E 03 ND 1 0 Note Increasing I Probably Increasing PI Stable S Prob
375. ould include data from at least four wells ASTM 1998 in which COCs have been detected May include up to two wells which have not exhibited COCs during more recent sampling events being analyzed but in which COCs were previously detected As many wells should be included in the evaluation as possible subject to the other minimum data requirements Minimum Data per Wal Data for each well should include at least four measured concentrations over six sampling events during the time period being analyzed For any well data may not be missing from more than two consecutive sampling events Guidelines given by ASTM 1998 notes that a minimum of more than one year of quarterly monitoring data of 4 or 5 wells is needed to establish a trend Number of Sampling Events Evaluation should include at least six most recent sampling events Sufficient Data At least four wells which satisfy the minimum groundwater data With four or more independent requirements specified above For this sampling events per wel are evaluation it is suggested that the user available consolidate multiple sampling dates within a single quarter to consider them to be a single Insufficient Data Fewer than four wells or fewer than 4 independent sampling event with multiple measurements of sampling events per wel are the same constituent subject to a user defined available consolidation eg average The sampling events do not need to bethe samefor each
376. ource ASTM 1998 Two types of data can be used for this regression 1 data from monitoring wells points located on or close to the centerline and 2 data estimated from hypothetical sampling points on the centerline through plume contouring The first type of data yields more accurate results than the second type and therefore is used in the risk based power analysis The user should select at least three plume centerline wells for the regression analysis see screen Version 2 1 A 6 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Parameters for Risk Based Power Analysis in chapter D amp ailed Screen D escription For convenience linear regression with log transformed concentrations is used in MAROS Data Sufficiency Analysis to estimate coefficients A and B Note B should be a negative value indicating declining concentrations away from the source The compliance boundary is assumed to be a line perpendicular to the preferential ground flow direction that is located at or upgradient of the nearest downgradient receptor Figure A 6 5 The user is asked to specify the whereabouts of the compliance boundary by providing the distance from the most downgradient well to the compliance boundary see screen P arameters for Risk Based Power Analysis in chapter M A ROS D amp ailed Screen D escription The projected concentrations are calculated by using Equation A 6 12 with t
377. pirical relationships to assist the user in optimizing a groundwater monitoring network system while maintaining adequate delineation of the plume as well as knowledge of the plume state over time Different users utilize the tool in different ways and interpret the results from a different viewpoint Therefore it is important to not only have a conceptual model see Appendix A 1 for the site before beginning the MAROS analysis but to also assess all of the MAROS results in conjunction with knowledge of site conditions regulatory framework community issues and other site specific situations Also the MAROS methodology assumes that the current sampling network adequately delineates the plume bounding wells have non detect values and that if a hydraulic containment system and or remediation system is currently in operation this will continue For a more detailed description of the structure of the software and further utilities refer to the Appendices 1 10 The goal of the tutorial is to show the user tips and pitfalls when applying MAROS ata typical site The tutorial example has been used only to illustrate the utilities of the MAROS software itis by no means a complete site analysis Version 2 1 A 11 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Note The MAROS software can be used to analyze sites more complex than this example with many more wells more s
378. ponding part in the user s guide Ei _ 4 APRA CUR for detai Is 2 fi Flow Direction l 8 nw Ay MAS p To visualize the cleanup status Es ous spatially click the Visualize button Ee li Eg and the Individual Wel Cleanup Status J int lt lt Back i i i i xi V ISU alization screen will appear z Attained Cont Sampling Not Attained N C He The cleanup status of each well is indicated with colored symbols on a scatter plot This plot allows the user to have a better understanding of the spatial distribution of individual well cleanup status over the site E3 Monitoring and Remediation Optimization System MAROS Clid the Back button to return and then dick Next gt gt on the Individual Wal Cleanup Status Results screen to proceed The Individual Wal Power Analysis Complete screen will appear Individual Well Power Analysis Complete ON You have finished the cleanup status evaluation for individual wells for the time periods selected by you You may now proceed to the other analysis of Data Sufficiency Analysis You may also go back to choose another time period for analysis If you would like to view the report right now you can proceed to Main Menu from Data Sufficiency Analysis Menu or go back to previous screens where reports can be generated lt lt Back SAMPLING OPTIMIZATI Version 2 1 A 11 57 October 2004 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM
379. pose For example if the SF threshold for all nodes is 0 10 those nodes with SF values less than 0 10 are potential nodes to be eliminated CR and AR thresholds are defined for the second purpose For example if CR Version 2 1 A 4 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE threshold is 0 95 elimination of locations is valid if the CR valueis greater than 0 95 In this case the acceptable level of information loss is 1 0 95 0 05 that is 5 If CR value is less than 0 95 the optimization should be terminated and the locations eliminated at this step should be re instated Details about these thresholds will be discussed shortly Circumcircle FigureA 3 3 Division of a Delaunay Triangle Sampling Location Elimination Status SF 0 SF 1 Interpretation Perfect estimation High estimation error CR or AR far from 1 significant information loss CR 1 and AR 1 less Information loss Figure A 3 4 Decision Process of the Elimination of a Location The Delaunay method performs the redundancy reduction by using an algorithm that considers all or a series of sampling events of which optimization based on a single sampling event is a special case Since each sampling event represents only one snapshot of the contaminant plume we need to examine all sampling events or parts of them to reveal the general spatial pattern of the contaminant
380. predict the ultimate extent of constituent migration either for natural attenuation process or site undergoing remediation Results for the modeling trend that can be entered in the software indude Increasing I Probably Increasing PI No Trend NT Stable S Probably Decreasing PD Decreasing D or Not Applicable NA Insufficient Data TREND ANALYSIS Version 2 1 44 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE External Plume Information External Plume Information Empirical Results accessed from the External Plume Information M odding Results screen allows the user to enter empirical trend information by well and constituent or for all source or all tail wells The rationale and limitations to this approach is outlined in Appendix A 4 Options include entering empirical trend External Plume Information Empirical Results results i based on separate empirical evidence for both source and tail wells ii individual ee See Empirical Evidence well trends based on separate empirical rules If there are no empirical results choose the e CD option No separate empirical evidence to be C Based on separate empirical evidence appl ied See Empirical Evidence Takes the user to the Empirical Evidence by State Edit individual well trends based on separate empirical evidence Back Returns the user to the M oddi
381. r 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Initial Data Table accessed from the COC D ecision screen allows the user to view the initial data table with the COCs chosen as well as the sample events defined and effective dates This table is not available for editing but should be used to check for proper importation and sorting of data for the rest of the analysis E id Remediation Opt tion System MAROS k h h COC ic ee lal Back Returns the user to the Decision Initial Data Table screen Below is the data table with all specified data reduction operations performed Dates shown are effective dates assigned for a given sample event Next Takes the user to the M ain M enu screen Effective Result Det 4 Well Name SIT SampleEvent pate coc mgl Limit IMA 1 S Sample Event 15 2 19 1 99 BENZENE 0 0192 0 001 i P MA 12 s Sample Event 2 1 17 198 BENZENE 0 046 0 001 p d es fi h SCI een om IM 15 s Sample Event 15 2 19 4 99 BENZENE ND 0 001 H e n P rovi l n o rmati o n o n t e r M 15 S Sample Event 15 2 19 1 9 ETHYLBENZENE ND 0 001 p ec fi p eq emen IMA 14 s Sample Event 15 2 19 1 99 BENZENE ND 0 001 S l C l n ut r u l r ts MVV 14 s Sample Event 15 2 19 1 99i ETHYLBENZENE 0 0077 0 001 MA 13 s Sample Event 15 2 9 99 BENZENE ND 0 001 IMW 13 S Sample Event 15 2 19 1 99 ETHYLBENZENE ND 0 001 MVV 14 s Sample Event 5 3 13 99C ETHYLBENZENE ND 0 001
382. r about the mean SECOND MOMENT SHOWS SPREAD OF THE PLUME OVER TIME The second moment indicates the spread of the contaminant about the center of mass ox and Oyy or equivalently Sx and Syy or the distance of contamination from the center of mass for a particular COC and sample event The Second M oment represents the spread of the plume over time in the x and y directions with x axis representing its major migration direction Freyberg 1986 describes the second moment about the center of mass as the spatial covariance tensor The components of the covariance tensor are indicative of the spreading of the contaminant Version 2 1 A 5 4 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE plume about the center of mass The components of the covariance tensor can be described in terms of an ellipse x the major axis and y the minor axis The values of oxx and oyy represent the axes of the covariance ellipse The 2 D covariance or second moment equations axial terms are as follows M M 2 0 0 2 2 0 0 2 6 m X O e M M 0 0 0 0 0 0 where oxx and oyy are the second moments for a particular COC i and sample event Xc Yc are the coordinates of the center of mass Similar to the other Moment calculations the data are spatially discontinuous therefore a numerical approximation to this equation is required To conduct the numerical integration the ho
383. r of mass over time is shown to be increasing I This means the center of mass is getting farther away from the original source area Select Next to proceed to the First M oment Plot Change in Location of M ass Over Time screen roe we rmm t PE macraiT Wa Twwi VIT in Mpc Raabe FE tm n eate 7 o o lt gt x z m zd O m ki v Note If more than one COC was being used plots of other chemicals can be obtained using the Chemical drop down box at the top of the screen followed by selecting the Graph button The Version 2 1 A 11 32 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE graph type can be specified as Log or Linear The graph can be printed by selecting the View Report button 2 First Moment Plot Changein Location of M ass Over Timeis used to view the First Moment Analysis results by constituent over time The first moment estimates the center of mass of the plume coordinates Xc and Yc for each sample event and COC The center of mass locations indicate the movement of the center of mass over time T First Moment analysis results showing the First Moment Plot change in location of mass over time are ois test s e Ta mantani naear aii bee displayed for benzene Each point represents a sample event The results can be compared to the groundwater flow direction displayed on
384. ration significantly exceeds the cleanup goal To facilitate the power analysis projected concentration data are assumed to be normally or lognormally distributed Results for both assumptions are calculated and provided for comparison In most cases they agree with each other See Appendix A 6 for detailed explanations View Normal Views results calculated under the assumption that data are normally distributed View Log Viewsresults calculated under the assumption that data are lognormally distributed Back Returns the user to the Centerline Regression Projected Concentrations screen View Report Generates a report with the risk based power analysis results for the sampling events selected by the user The user can go back to re run the analysis by selecting a different time set of parameters Next Proceeds to the Risk Based Power Analysis Complete screen Help Provides information on the screen specific input requirements Version 2 1 84 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Version 2 1 85 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Risk Based Power Analysis Complete This screen accessed from the Risk Based Power Analysis Results screen by clicking Next is a message screen indicating that risk based power analysis has been completed and the user can procee
385. ration trends can be set Back Returns the user to the Sampling O ptimization screen Analysis Determines sampling frequencies at all sampling locations for each COC by using the Modified CES method The Sampling Frequency Recommendation screen will pop up Help Provides additional information on software operation and screen specific input requirements Steps for use 1 Define the recent period first The ending sampling event should be later than the beginning sampling event A minimum of six sampling events is recommended for the analysis For analysis with less than six samples the results could be inaccurate 2O Use previously selected sampling events shown on the From and To dropdown lists Click the Confirm button to confirm the selection Click the Options button and enter the Sampling Frequency Analysis Options screen Define field specific Rate of Change parameters for COCs there Close that screen and return Default values will be used if parameters are not defined 4 Click theAnalysis button to perform the analysis iu M E Version 2 1 69 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Sampling Frequency Analysis Options This screen accessed from the Sampling Frequency Analysis screen by clicking O ptions is used for setting the Rate of Change ROC parameters that are required by the M odified CES method Sampling Frequency
386. re listed in the third column of Table A 7 15 For example e 5 51 4 97 0 54 Step 3 The observed serial correlation 0 is eit 0 151 0 318 1 035 2 03 Y 0 076 0 297 0 986 15 17 0 13 d i Step 4 The Durbin Watson statistic D is N 2 Le Ten 0 674 0 002 4 142 Ye 15 17 1 66 Step 5 Consulting the Durbin Watson table N eter et al 1996 page 1349 with 0 05 and n 16 in the column titled p 1 1 we find dy 1 37 Since D 1 66 gt d 1 37 condude that thereis no serial correlation in this set of manganese measurements Therefore in future use of this time series data independence between data can be assumed true The serial correlation between successive observations computed from the above procedures depends on the time interval between collecting groundwater samples For the AR 1 process the serial correlation between successive observations decays exponentially with the increase of separation interval px ptk where p is the serial correlation for time interval t and pe is the serial correlation for the time interval that is k times as long astime interval t Theinverse of this relation also holds As the time interval becomes longer eg from monthly to semiannual sampling the serial correlation between successive observations approximates zero This is the theoretical basis for achieving serially independent observations in a sampling design When
387. reate to Access File Damis V chui shitan d He senin ris cul mot Cet e Folder CA AFCEE MAROSV e Filename TutorialResults mdb Enter details here na paan rn Select Create l mene za Click Back to return to the MAROS O utput R eports Graphs screen Click hereto proceed DATA MANAGEMENT Version 2 1 A 11 68 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Tutorial Site Conclusions At this point in the software the user has gone through all of the optimization utilities and can consolidate the knowledge of the site with the MAROS analysis results to make a final determination of the site optimization The goal of the tutorial is to show the user tips and pitfalls when applying M AROS at a typical site The tutorial example has been used only to illustrate the utilities of the MAROS software and it is by no means a complete site analysis Results from the temporal trend analysis moment analysis sampling location determination sampling frequency determination and data frequency analysis indicate that e The 7 site monitoring source wells are located near the Tank Field These have historically elevated benzene concentrations There are 5 tail wells e 3 out of 7 source wells and 4 out of 5 tail wells have a Decreasing trend 4 out of 7 source wells and 1 out of 5 tail wells have a Stable trend
388. reen Results of the M ann Kendall Trend Analysis are shown on the M ann Kendall Statistics screen Next results of the Linear Regression Trend Analysis are shown on the Linear Regression Statistics screen Continue to the Linear Regression screen to view the results in graphical form Finally a summary of both the Mann Kendall and Linear Regression results are shown on the Trend Analysis Summary by Wal screen To proceed dick Next The Statistical Plume Analysis portion of the software is complete 9 Plume Analysis M enu From the Plume Analysis M enu select Spatial Moment Analysis by clicking the button next to the label This action will take the user to the M oment Site D etails Screen 10 Spatial Moment Analysis First enter the site details on the Moment Analysis Site D etails screen Then in each screen view the information from the Oh 1s and 27d Moment Analysis Results click Next to continue to the next screen Finally a summary of both the Moment Analysis results are shown on the Spatial M oment Analysis Summary screen To proceed dick Next The Spatial Moment Analysis portion of the software is complete 11 Plume Analysis M enu From the Plume Analysis M enu select External Plume Information by dicking the button next to the label This action will take the user to the External Plume Information M odding Results screen 12 External Plume Information In each screen select the information that pertains to the site for
389. rend include Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not Applicable Insufficient Data Mann Kendall Statistic S The Mann Kendall Statistic S measures the trend in the data Positive values indicate an increase in estimated mass over time whereas negative values indicate a decrease in estimated mass over time The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic i e large magnitudes indicate a strong trend However the zeroth moment calculation can show high variability over time largely due to the Version 2 1 A 5 2 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE fluctuating concentrations at the most contaminated wells as well as varying monitoring well network sampling Confidence in Trend The Confidence in Trend is the statistical confidence that the distance to the from the source to the center of mass is increasing S20 or decreasing S 0 COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average Values near 1 00 indicate that the data form a relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about the mean FIRST MOMENT S
390. requency based on overall trends In this step the determined sampling frequency can be one of three possible results Annual Semi Annual or Quarterly The adjustment based on recent overall ratio will be performed in the next step Figure A 3 1 gives a quick decision matrix that is similar in function to the above flowchart but is more illustrative of the results Version 2 1 A 9 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 2 ADJUSTMENT BASED ON RECENT OVERALL RATIO If the frequency determined from overall trend is greater than that from the recent trend e g the overall frequency is Quarterly while the current frequency is Annual we might need to adjust the recent frequency by one level When the recent trend is significantly lower than the long term trend reducing the sampling frequency gradually will ensure safety The steps to be followed are shown in thefollowing flow chart Recent frequency is less N than Overall frequency Overall CT is Incr ProbIncr or NoTrend Recent frequency is SemiAnnual Recent frequency is Annual Overall CT is Incr Probi ncr or NoTrend Overall frequency is SemiAnnual Overall CT is Incr Probi ncr or NoT rend E Version 2 1 A 9 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION
391. rizontal plane x y was divided into contiguous triangular regions with the apex of each triangle defined by a well sampling location with an associated COC concentration at each sample location A spatial interpolation method over these triangles allows the first moment calculations using Delaunay Triangulation see Appendix A 2 for methodology The Delaunay triangulation is a rough way to discretize the domain The following formulas represent the 2 D approximation of the spatial covariance tensors R Y X X Y V C in Y Y ZXYXo 7 Vs MEE Ca X X Ea X Xr Y MC LE VG Where Sy Syy and Sy the diagonal term are the spatial covariance tensors for a particular COC i and sample event where Ciayg is the geometric mean concentration of each triangle for a particular COC i X and Y are the spatial coordinates the easting northing coordinates of the center of each triangle Vi is the volume of the triangle calculated by d Aj where d is the averaged saturated thickness and A is the are of the triangle S n In order to analyze the behavior of the plume the values of the spatial covariance tensors need to be adjusted relativeto the orientation of the plume elliptical axes It is assumed that the major elliptical axis x is parallel to the estimated mean groundwater velocity vector and the minor elliptical axis y is perpendicular to the groundwater direction The components are estimated using the f
392. rk has more than enough wells to detect the risk based site cleanup status If the expected Version 2 1 A 6 9 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE sample size is greater than the sample size more sampling locations are needed to confirm the cleanup status Distribution Assumption shows the assumption of data distribution for the results currently shown Results for both normal and lognormal assumptions are given Because normality tests for small size sample e g lt 20 may not be accurate presenting results under both assumptions provides a chance for comparison so that the conservative results may be used In addition to AlphaLevel and Targ amp P ower power analysis parameters used in the risk based site cleanup evaluation include Detection Limit The uniform detection limit for a COC specified by the user It is only used in the risk based power analysis to indicate that the projected concentrations are below the detection limit The detection limit for a COC is by default set to 20 of the MCL of a COC if available in MAROS database References ASTM 1998 Standard Guide for Remediation of Groundwater by Natural Attenuation at Petroleum Release Sites American Society of Test and Material E1943 98 Cohen J 1988 Statistical Power Analysis for the Behavioral Sciences Lawrence Erlbaum Associates Hillsdale N ew Jersey U S EPA 1992 M et
393. rmation Weighting Assumptions Consolidation Step 1 Weight Plume Information by Chemical Summary Weighting Weighting Applied to All Chemicals Equally Consolidation Step 2 Weight Well Information by Chemical Time Period 10 4 1988 to 12 19 1998 Consolidation Period No Time Consolidation Consolidation Type Median Duplicate Consolidation Average ND Values 1 2 Detection Limit J Flag Values Actual Value Well Weighting No Weighting of Wells was Applied Chemical Weighting No Weighting of Chemicals was Applied Note These assumptions were made when consolidating the historical montoring data and lumping the Wells and COCs 1 Compliance Monitoring Remediation Optimization Results Preliminary Monitoring System Optimization Results Based on site classification source treatment and Monitoring System Category the following suggestions are made for site Sampling Frequency Duration of Sampling before reassessment and Well Density These criteria take into consideration Plume Stability Type of Plume and Groundwater Velocity Tail Source Levelof Sampling Sampling Sampling coc Stability Stability Effort Duration Frequency Density BENZENE D PD L Sample more year Biannually 6 months 15 Note Plume Status l Increasing Pl Probably Increasing S Stable NT No Trend PD Probably Decreasing D Decreasing Design Categories E Extensive M Moderate L Limited N A Not Applicable Insufficient Data Available
394. ro However the estimated value of p isunlikely to be zero even if the actual serial correlation is zero The Durbin Watson statistic can be used to test whether the observed value of p denoted as 6 is significantly different from zero The procedures below introduce how to calculate p the observed value of p followed by the Durbin Watson test Procedures Step 1 List the observations measured consecutively at a constant interval ordered by time denoting them as xi i 1 2 N where N is the number of total observations Estimate the trend and or seasonality from this set of data The trend is commonly estimated by the least square method and expressed as a linear trend y b b t where b and b are the intercept and slope of the regression line respectively is the estimate of the observation at time ti In the case of no obvious trend in the time series the model is b where by is the overall mean of the observations Seasonal variability is generally indicated by a regular pattern that is repeated every year The seasonal mean or median is usually used to characterize the average concentration level of a season This average level is simply the mean or median of the detrended observations in a certain season denoted as w j 1 2 k where represents a season and k is thetotal number of seasons in a year Step 2 Calculate the residuals 4 by subtracting the observations from their trend and or respective
395. roceed An Excel chart called xIsN ewLocation will pop up 7 The xIsN ewLocation Excel chart indicates the concentration estimation uncertainty at the center of each Delaunay triangle with a colored letter S represents small M represents moderate L represents large and E represents extremely large The interpretations of the results are also provided on the chart The areas with L or E code can be considered for new sampling locations Version 2 1 A 11 51 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE In this example there are no triangles having L or E letters Since the plume is decreasing over time see results from Plume Analysis therefore no new locations need to be recommended After viewing the results the user can print out this chart or save it with a different name for future use Press Back to Access to return to the W ell Sufficiency Analysis N ew Locations screen or simply switch back by selecting the M AROS application In the Wal Sufficiency Analysis New Locations screen click Next gt gt to proceed The Sampling Location Analysis Complete A ccess M odule screen will appear This screen indicates that the sampling location optimization is completed The user may now choose to perform other sampling optimization analyses by selecting Sampling Optimization or go back to modify certain parameters and make more analys
396. roethane 1 2 1 2DCA EDC DCA12 12 DICHLOROETHANE ORG 156 59 2 Dichloroethene cis 1 2 1 2 cis DCE DCE12C ds 12 DICHLOROETHYLENE ORG 156 60 5 Dichloroethene 1 2 trans 1 2 trans DCE DCE12T trans 1 2 DICHLOROETHENE ORG Dichlorometh 75 09 2 Methylene chloride ane MTLNCL METHYLENE CHLORIDE ORG 79 345 Tetrachloroethane 1 1 2 2 PCA 1 1 2 2 TETRACH LOROETHANE ORG 127 18 4 Tetrachloroethene PCE Perc PCE TETRACH LOROETH YLEN E PCE ORG 120 82 1 Trichlorobenzene 1 2 4 TCB124 1 2 4 TRICHLOROBEN ZENE ORG 71 55 6 Trichloroethane 1 1 1 TCA111 1 1 1 TRICHLOROETHANE ORG 79 00 5 Trichloroethane 1 1 2 TCA TCA112 1 1 2 TRICHLOROETHANE ORG 79 01 6 Trichloroethene TCE TCE TRICHLOROETH YLEN E TCE ORG 75 69 4 Trichlorofluoromethane FC11 TRICHLOROFLUOROMETHANE ORG 75 01 4 Vinyl chloride vc vc VINYL CHLORIDE ORG PAH COMPOUNDS 83 32 9 Acenaphthene ACNP ACENAPHTHENE ORG 208 96 8 Acenaphthylene ACN PY ACENAPHTHYLENE ORG 120 12 7 Anthracene ANTH ANTHRACENE ORG 205 99 2 Benzo b Fluoranthene BZBF BEN ZO b FLUORAN THENE ORG 191 24 2 Benzo g h i Perylene BZGHIP BENZO g h i PERY LENE ORG 207 08 9 Benzo k Fluoranthene BZKF BEN ZO k FLUORANTHENE ORG Version 2 1 A 2 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Abreviation MAROS CAS or ERPIMS Constituent Constituent Number Constituent Synonym Code Name Type 56 55 3 Benzo a Anthracene BZAA BENZO a ANTHRACENE ORG 50 32
397. rosoft Excel for the analysis of only one sampling event lt lt Back Excel Module gt gt A 11 48 Air Force Center for Environmental Excellence Version 2 1 October 2004 AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE events and the Access Module will be used to illustrate the sampling location optimization For optimization with the Excel Module please refer to Sampling Location Determination Excel Module in the MAROS Detailed Screen Descriptions section of the user s guide amp Monitoring and Remediation Optimization System MAROS Perform well redundancy d i to eliminate redundant locations thai CI ick to Pitain thresholds You may choose to use eil 1 sphical method confirm ri pe Select Sample Event 15 iethod realized within Microsoft Access Md ftle gt gt Access for multiple sampling events analysis Graphical method realized within Microsoft Excel E Pess for the analysis of only one sampling event xcel Module gt lt lt Back Now select Sampling Event 10 from the From dropdown list and Sampling Event 15 from the To dropdown list The latest five years of data will be used Click Confirm to confirm the selection and the Access Module button will be activated Click Access Module and the A ccess Module Potential Locations Setup screen will appear SAMPLING OP 2 TheAccess M odule Potentia
398. rvices Inc Ling M University of Houston Objective This tutorial has been set up to guide the user through the most commonly used features of the MAROS software The MAROS 2 1 software used to optimize the long term monitoring LTM network at the Service Station site is explained in general terms in this Appendix The software will be used to optimize the monitoring network and sampling plan at a hypothetical service station site where the groundwater is affected by BTEX compounds Figure A 11 1 The general objective of the tutorial is to optimize the Service Station long term monitoring network and sampling plan applying the MAROS 2 1 statistical and decision support methodology The key objectives of the tutorial include familiarizing the user with typical applications of the MAROS software to a hypothetical site To determine the overall plume stability through trend analysis and moment analysis e To evaluate individual well benzene concentration trends over time To reduce if possible redundant wells without information loss and addition of new wells for future sampling To provide future sampling frequency recommendations while maintaining sufficient plume stability information e To evaluate risk based site cleanup status using data sufficiency analysis MAROS is a collection of tools in one software package that is used in an explanatory non linear fashion The tool includes models statistics heuristic rules and em
399. s you can click button O ptions to change the optimization parameters that are used by the Delaunay method Each COC has its own parameters Note The Slope Factor in MAROS is a parameter indicating the relative importance of a location in the monitoring network and is not related to toxicological values for a particular COC i e carcinogenic risk Version 2 1 53 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well Redundancy Analysis Options This screen accessed from the Access Module Potential Locations Setup screen by clicking Options is used for setting the optimization parameters thresholds that are used by the Delaunay method Each COC has its own set of parameters Maret crarg end Leriertatae Cigiorrtossthun Watery MARIS Weti Redundancy Analysis Options the Da aes cam Pathe oobe soon prootes ore defe betes Tore wake thui ramat cot titer ty a iri thus Ladies awe COC name ande nade Melao Ape Cencosanan Slope fado Supe Facite Rae icmsek Set tw deraut Lue Locations Setup screen These parameters include Inside node Slope Factor SF Hull node Slope Factor Area Ratio AR and Concentration Ratio CR The default values for these parameters are 0 10 0 01 0 95 and 0 95 respectively for all COCs For detailed explanations of these parameters refer to Appendix A 3 The user can change parameters by entering
400. s available MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 6 of MAROS Power Analysis for Individual Well Cleanup Status Project Example User Name Meng Location Service Station State Texas From Period 4 3 1991 to 12 19 1998 Normal Distribution Lognormal Distribution Sample Sample Sample Assumption Assumption Alpha Expected Well Szie Mean Stdev Cleanup Status Cleanup Status Level Power BENZENE Cleanup Goal mg L 0 005 Target Level mg L 0 004 MW 1 6 3 90E 01 5 44E 01 Cont Sampling Cont Sampling 0 05 0 8 MW 12 6 7 61E 03 1 01E 02 Cont Sampling Cont Sampling 0 05 0 8 MW 13 6 6 57E 03 1 07E 02 Cont Sampling Cont Sampling 0 05 0 8 MW 14 6 1 00E 03 1 50E 11 Attained Attained 0 05 0 8 MW 15 6 1 00E 03 1 50E 11 Attained Attained 0 05 0 8 MW 2 6 2 89E 03 X 3 53E 03 Cont Sampling Cont Sampling 0 05 0 8 MW 3 6 3 53bE 02 4 91E 02 Cont Sampling Cont Sampling 0 05 0 8 MW 4 6 1 87E 02 1 20E 02 Cont Sampling Cont Sampling 0 05 0 8 MW 5 6 1 11E 00 8 68E 01 Cont Sampling Cont Sampling 0 05 0 8 MW 6 6 1 00E 03 1 50E 11 Attained Attained 0 05 0 8 MW 7 6 1 00E 03 1 50E 11 Attained Attained 0 05 0 8 MW 8 6 1 00E 03 1 50E 11 Attained Attained 0 05 0 8 Note N C refers to not conducted because of insufficient data N lt 4 S E indicates the sample mean significantly exceeds the cleanup level and thus no analysis is conducted Sample Size is the number of concentration data in a sampling location that are used in the
401. s have been occurring for an extended period or PRG s have been met Sites with Source or Tail results that indicate an increasing plume size are recommended for indefinite remediation or consider increasing performance or remediation mechanism Sites with Stable in the Source and Tail suggest to consider removing the treatment system if previously reducing concentration or PRG met SAMPLING DENSITY The sampling density determination for a site currently undergoing remediation is identical to that not currently undergoing site treatment H owever the results should be considered in the context of evaluating both regulatory compliance as well as remediation method performance evaluation References Mace R E R S Fisher D M Welch and S P Parra Extent M ass and Duration of Hydrocarbon Plumes from Leaking Petroleum Storage Tank Sites in Texas Bureau of Economic Geology University of Texas at Austin Austin Texas Geologic Circular 97 1 1997 McNab W W D W R J Bear R Ragaini C Tuckfield and C Oldenburg 1999 Historical Case Analysis of Chlorinated Volatile Organic Compound Plumes Lawrence Livermore Laboratory University of California Livermore Ca 1999 http searchpdf adobe com proxies 0 5 69 6 html Version 2 1 A 8 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 3 SAMPLING FREQUENCY ANALYSIS MODIFIED CES METHOD A
402. s in a table format and allows the user to visualizethe results spatially BENZENE dropdown list and 1998 from the To s idi Hame Sample Cleanup a Distribution Assumption dropdown list i a Tear z MW 12 6 Cont Sampling T e ET oni Samping View Normal Click the Analysis gt gt button to d ae proceed The Individual Wal Cleanup E 2 E Viewtog Status Results screen will appear a NO S a Ed Wy The Sample Size column contains the number of yearly averages used in the analysis The Cleanup Achieved column shows the cleanup status at each well There are four types of results Attained Not Attained Cont Sampling continue sampling and N C not conducted The detailed results of the analysis are given in a report that can be accessed by clicking the View Report button The View Normal and View Lognormal buttons allow the user to view results calculated assuming that the data are ae zix normally distributed and lognormally Individual Well Cleanup Status Visualization d i stri bu ted respectively The The well cleanup status is indicated by the color of the well Select a COC to graph Distribution Assumption u EAT BENZENE z Normal Optional Power Analysis button allows the user to enter another screen AMT len Lol where detailed power analysis results MM ms Lognormal are provided refer to the 160 100 50 l J n WOMW 12 150 200 250 corres
403. s in the database The MAROS software divides the wells for the site into two different zones eg Source zone and Tail zone The Source area indudes zones with free phase NAPLs residual NAPL contaminated vadose zone soils and or other source materials The source area is generally the location with the highest groundwater concentrations of constituents of concern The source zone wells for this site indude MW 1 MW 2 MW 3 MW 5 MW 6 MW 7 and MW 5 Figure A 11 1 The downgradient groundwater plume Tail zone is the area downgradient of the contaminant source zone The Tail only contains contaminants in the dissolved phase and the sorbed phase but contains no sources of contamination The tail wells for this site include MW 4 MW 12 MW 13 MW 14 and MW 15 Figure A 11 1 Sauce var Rane Sefecdon Assign well categories as being in Source or Tail zones Table A 11 2 To do this click on the appropriate Source or Tail box adjacent to each well Use the scroll bar to the right of the box to view all wells _ Select Next to continue The Wall Coordinates screen will appear Click hereto proceed enn SITE DETARS Version 2 1 A 11 13 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 5 WELL COORDINATES W ell Coordinates allows the user to define and or revisethe well co
404. s stage at a later time by re importing the archive created Version 2 1 A 11 17 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Plume Analysis Step 3 Plume Analysis allows the user to perform data reduction as well as trend analysis through both Statistical Plume Analysis Spatial Moment Analysis and External Plume Information It also allows the user to apply final Analysis Consolidation to the trend results STEP 1 PLUME ANALYSIS E Wen tt eg md Berri oe Dpi ero tur Vf MANETS Pe Fm yi i Parame zl M OX erano Th town troup ite emer ewe anm arro c ten Click hereto proceed Dato Managemarc dei cha pee n To am FPE es arem mnt rs S08 er ord m tton Ste Detale DE at den Qr ror yionihens s Phime Anatetie Ferara Den Gonratenaton Coa aon ad Pina arion Anara act Stee Chri Dom rmn Sampling Optimization barwa ancia ee ee ee eue XA MAROS Oxnput Mii ie m cam me tma n un e pasa rer Select Plume Analysis from the Main M enu The PlumeAnalysis M enu will be displayed The Plume Analysis M enu screen serves at the center of the trend analysis user interface The user progressively steps through the Long Term Monitoring Plume Analysis process by navigating through the options displayed As individual steps of the process are completed options to select become successively available The Plume Analysis M enu screen
405. s the program while remaining in the worksheet To restart an analysis after pressing this button press INIT A pply again Optimize Performs optimization i e eliminating redundant locations from the network Options Shows the Options Form that includes optimization parameters used in the Delaunay method and the options for graphic output Back to Access Sends results back to the Microsoft Access screen Wd Redundancy Analysis Excel M odule Version 2 1 65 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE The Options Form can only be used in the Well Locations chart sheet Optimization Drawing Control Slope Factor Setting Reduction Ratio Inside nodes 0 1 Area Ratio 0 95 Concentration Hull nodes 0 01 Ratio 0 95 Plot Voronoi Diagram xj The Option Form is accessed by clicking the Options button It has two pages Shown on the left is the Optimization page Parameters indude Inside node Slope Factor SF Hull node Slope Factor Area Ratio AR and Concentration Ratio CR The default values are the same as those in the A ccess M odule Set to default Sets the parameters to system default The button will be activated only if the parameter value is not equal to the default value Shown on the left is the Drawing Control page Plot Delaunay Triangulation By checking this box the blue triangulation lines will be plot
406. s two power curves for the detection of changes in dissolved oxygen with a sample size of 8 and a significance level of 0 05 If the minimum detectable difference is 0 4 mg L and the sample standard deviation is 0 5 mg L the power to detect this change is 0 7 If the sample standard deviation is 1 0 mg L the power to detect this change is dramatically reduced to less than 0 3 If the same level of power 0 7 is to be maintained the minimum detectable difference doubles 0 8 versus 0 4 for the sample with a higher variability o 1 0 Therefore the sufficiency or power of a sampling plan can be evaluated in terms of the goal established in the sampling plan Power Probability of Detection 0 0 2 0 4 0 6 0 Minimum Detectable Difference mg L Figure A 6 1 Power curves for different variability Statistical power analysis provides additional information for interpreting the results of statistical tests The additional information includes 1 the power of the statistical test e g tests for trend or mean difference for individual wells or a group of wells and 2 the expected sample size of a future sampling plan given the minimum detectable difference it is supposed to detect Such information can assist users in modifying sampling plans to effectively achieve monitoring goals Version 2 1 A 6 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Power An
407. s used in the statistical power analysis Then click the corresponding button to perform any of the two analyses b Power Analysis at Individual Wells Clicking this button will take the user to the individual well cleanup status evaluation The user needs to select the series of sample events intended for analysis and define some other parameters There are several steps to follow to finish this analysis and results reports can be viewed immediately after the evaluations are finished c Risk Based Power Analysis Clicking this button will take the user to the risk based site cleanup evaluation The user needs to specify four sets of parameters in screen Parameters for Risk Based Power Analysis before continuing the analysis Regression of plume centerline concentrations projection of concentrations and the risk based site cleanup evaluation are determined sequentially Results reports become available immediately after each step is finished The user can choose to run either Sampling Location Analysis or Sampling Frequency Analysis first Because Data Sufficiency Analysis uses qualitative concentration trend results from Sampling Frequency Analysis it cannot be selected before Sampling Frequency Analysis is successfully performed For detailed instructions on how to run these modules refer to the next chapter MAROS DETAILED SCREEN DESCRIPTIONS Version 2 1 11 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND
408. sampling location optimization results for the analysis of only one sampling event A location is marked for elimination only if this location is eliminated for all COCs Elimination of a location is interpreted as stopping sampling at this location If in the previous step some COCs were not analyzed the results given in this form may be incorrect due to incomplete analyses igx Eliminated Displays whether a location 3 Monitoring and Remediation Optimization System MAROS go back to rerun the analysis by lt lt Back View Report Next gt gt changing parameters or by selecting a different series of sampling events If not all COCs are analyzed in the previous step results shown in the report may be incorrect due to incomplete analysis Excel Module All in one Results is considered redundant and should be The final optimization results after considering all COCs for ONLY ONE sampling event are shown in eliminated A check mark in this field the following table A sampling location is eliminated only if it is eliminated from all COCs Eliminated Y i oe eese EEO nee stands for the elimination of a location LociD ESCoord NSCoord Eliminated Back Returns the user to the Well WA 150 205 d Redundancy Analysis Excd Module MW 12 100 0 8 0 al MW 13 65 0 230 d screen z MW 14 102 0 20 0 LI 9 ies on mE Next Proceeds to the Sampling Location Lr MW 2 20 30 0 d S WWS 350
409. sclition Complete At this point your data has been reduced according to the parameters you entered You may now eye eon rey no ed ey werner ed proceed to Step 3b Statistical Plume Analysis and analyze the trends in the groundwater data facris te gun nea Canfeun s Step gt gt u 5 gt 4 a 5 z u r Version 2 1 31 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Statistical Plume Analysis M ann Kendall Statistics accessed from the Plume Analysis M enu allows the user to view the Mann Kendall Trend Analysis results by well and constituent For further details on the M ann Kendall Analysis Method see A ppendix A 2 x To navigate the results for individual Mann Kendall Statistics constituents dick on the tabs at the top of the The Mann Kendall Analysis is used for analyzing a single groundwater constituent multiple screen constituents are analyzed separately Each tab below shows the statistics for one constituent See manual test or Help for description of trend determination method COV The Coefficient of Variation COV is a Statistical Analysis Results Last column is the result for the trend statistical measure of how the individual data c S CON MKG Confidence in Trend EM points vary about the mean value The fBENZENE eTHYLBENZENE TOLUENE XYLENES TOTAL ou icis 3E coefficient of variation defined as the
410. se the test is Not Significant indicating the mean concentration is not significantly below or is higher than the cleanup goal The critical t value or quantile is obtained using the Microsoft Excel function TIN V In calculating statistical power and the expected sample size associated with the Student s t test an approximate power equation from Cohen 1988 is adopted in MAROS Data Sufficiency Analysis The approximate power equation is pe d n 1 V2n Equation A 6 4 2 n D 1242 106 l a where a is the significance level B is the type II error n is the sample size d is the effect size and Z isthe percentile of the standard normal distribution The effect size d is calculated as demum Equation A 6 5 S where c is the cleanup level e g MCL m and s are sample mean and standard deviation respectively When log transformed data are used c is the logarithmic cleanup goal and m and s arethe mean and standard deviation of the log transformed data respectively Version 2 1 A 6 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Statistical power i e 1 8 is obtained by transforming Z to probability using Microsoft Excel function NORM SDIST Given a B i e 1 power and d of the sample the expected sample sizen can be solved from Equation A 6 4 using N ewton R aphson algorithm Two tests for the cleanup status evaluation are i
411. se negative rate Type II error rate B is the probability of incorrectly accepting the null hypothesis is also referred to as the significance level of a statistical test 1 6 is equivalent to the power or sensitivity of a statistical test and is the probability of correctly rejecting the null hypothesis when it is not true These concepts are illustrated in Table A 7 4 and Table A 7 5 for two types of groundwater monitoring programs respectively Two questions arise regarding the control of false positive and false negative rates in groundwater monitoring 1 is it possible to completely avoid false posititves and false negatives and 2 to what level can we reduce false positive and false negative rates For the first question it is important to recognize that false positives and false negatives in groundwater monitoring are inevitable because of natural variability or uncertainty due to spatial and temporal variations In addition analytical determinations associated with method detection limits MDL and practical quantitation limits PQL have false positive rates by design For example the false positive rate associated with MDLs for rarely detected constituents such as volatile organic chemicals VOCs is intended to be 1 or larger Clayton 1987 In practice limiting factors such as the monitoring budget control the levels to which the false positive and false negative rates can be reduced A lower error rate is generally achieved
412. ser can choose to analyze all existing monitoring data then censor the data consistently choose 1 quarter s worth of data eg the first sample event for each year and run the trend analysis again Run the MAROS trend results on both the sets of data and then compare the results If both trend results are the same then the trend results could have been obtained from using only annual sampling Similarly if you would like to be able to sample at a frequency greater than biennial this same type of analysis could be applied You could choose to monitor the well greater than every 2 years if the trend results are consistent with less data This type of analysis is only appropriate with adequately characterized plumes and long time period sample datasets gt 8 years The sampling frequency at the site is determined by the M onitoring System Category assigned by the results from the Source and Tail Stability as well as the Time to Receptor Sites with both decreasing Source and Tail Results are recommended for closure TABLE A 8 1 FREQUENCY DETERMINATION FOR SITES WITH NO GROUNDWATER FLUCTUATIONS AND MONITORED NATURAL ATTENUATION TTR Monitoring System Category E M L Close TTR 2 yrs Quarterly Biannually Annually 6 months Medium 2 TTR 5 yrs Biannually Annually Annually 6 months Far TTR gt 5 yrs Annually Annually Biennially 2 year interval TTR timeto receptor distance to receptor seepage
413. ser may notice that both M A ROS recommendation and qualitative evaluation were used in making the final recommendations The reasoning in the table is only used to illustrate the importance of further considerations In practice the user may need to do this for each decision Version 2 1 A 11 53 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE OPTION 2 SAMPLING FREQUENCY ANALYSIS Select Sampling Frequency Analysis from the Sampling O ptimization M enu The Sampling Frequency Analysis screen will appear 8 Define the recent period by selecting the beginning and ending sampling events Select the starting sampling event from the From dropdown list and the ending sampling event from the To dropdown list The From sampling event must not be later than the To sampling event The recent period is defined in order to calculate the recent concentration trend This period should be shorter than the overall sampling history For example the latest two or three years can be recent period as opposed to a total sampling time of six years Note The sampling frequency analysis requires that at least six monitoring records be used If less than six records are used in the analysis the accuracy of the results may be significantly affected Correspondingly at least six sampling events are to be selected For example a period of two years will contain eight s
414. sion in discrete fracture networks using the method of moments Water Resources Research 21 11 1677 1683 Version 2 1 A 5 11 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 6 DATA SUFFICIENCY ANALYSIS Authors Ling M and Rifai H S University of Houston The sufficiency of data in the statistical context refers to whether the observed data are adequate both in quantity and in quality for revealing changes in the variable of interest In long term groundwater monitoring there are at least two conditions that require sufficiency analysis 1 the need to further increase confidence in monitoring results or to detect more subtle changes in contaminant concentrations and 2 an overall adequate monitoring program that is not adequate at specific sampling points eg the sampling frequency in a well at the plume edge is too low to reflect a possible sudden change in concentrations Statistical power analysis can be used to evaluate the sufficiency of data for groundwater LTM plans This appendix details the two posterior statistical power analysis methods employed in the D ata Sufficiency Analysis module of the MAROS software These statistical power analysis methods are designed to assess 1 the cleanup status at individual wells and 2 the risk based cleanup status for the entire site An example question arising from these evaluations is what to do n
415. sis Complete You have finished the determination of sampling frequency You may now proceed to other options of Sampling Optimization You may also go back to define different recent period or rate of change parameters and re run this module The user may now choose to perform other sampling optimization analyses by selecting Sampling Optimization or go back to modify analysis parameters and make more analysis runs by selecting lt lt Back If you would like to view the report right now you can proceed to Main Menu from Sampling Optimization or go back to previous screens where reports can be generated lt lt Back SAMPLING OPTIMIZATION Click the Sampling Optimization button and the user will be brought back to the Sampling O ptimization M enu screen Version 2 1 A 11 55 Air Force Center for October 2004 Environmental Excellence OPTION 3 DATA SUFFICIEN CY ANALYSIS Select Data Sufficiency Analysis from the Sampling O ptimization M enu The Data Sufficiency Analysis Menu screen will appear The Data Sufficiency Analysis M enu screen serves at the center of the data sufficiency analysis user interface The user can AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE amp Monitoring and Remediation Optimization System MAROS nl xj Data Sufficiency Analysis Menu The Data Sufficiency Analysis Menu screen serves at the center of the Data Sufficiency Analysis that includes tw
416. sis Mann Kendall results by constituent lol xi Monitoring and Remediation Optimization System MAROS 4 Spatial Moment Analysis Summary The Mann Kendall Moment Analysis is used for analyzing a single groundwater constituent multiple constituents are analyzed separately Each tab below shows the statistics for one constituent See manual text or Help for description of trend determination method ETHYLBENZENE TOLUENE XYLENES TOTAL Moment Analysis Results Last column is the result for the trend Confidence Moment Moment COV MK S in Trend Trend Zeroth Moment Mass 03 25 88 0 s st Moment Distance to Source 05 67 100 0 2nd Moment Sigma XX 07 23 85 9 NT 2nd Moment Sigma YY 09 13 72 4 NT Note Increasing I Probably Increasing PI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Source Tail S T COV Coefficient of Variation MKIS Mann Kendall Statistic Back Next gt gt Help View Report SPATIAL MOMENT ANALYSIS At this point the Spatial Moment Analysis has been performed You may now proceed to the Step 3d External Plume Information Version 2 1 October 2004 To navigate the results for individual constituents dick on the tabs at the top of the screen The information displayed in this screen can also be viewed in report form Spatial Moment Analysis Summary Report from the M
417. sly archived MAROS files Export MAROS Archive File Export a MAROS data to an archive database file to be used in Help The Database Management menu allows you to perform database operations such as importing manuel data addition and archiving These operations are used intially to import site data into the software in order to perform analysis After you have performed the data analysis you can archive your data file for future use contain the site data as well as site details Choose the option of interest by clicking the applicable button Import New Data New data can be imported from Excel Access or ERPIM S data files Manual Data Addition This option allows the user to input data manually Manual addition is generally useful for a very small amount of supplemental data Import MAROS Archive File MAROS archive files can be created in the software in two locations after the initial data have been imported Archive files are in Access and Export MAROS Archive File MAROS creates an archive database file containing the sample data in a format that can be imported under the previous protocol Main Menu Returns the user to the M ain M enu Help Provides information on the screen specific input requirements Version 2 1 October 2004 14 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Import N ew D ata Import New Data accessed from the Data Ma
418. sources Program Information Management System LLNL Lawrence Livermore N ational Laboratory LOE Lines of Evidence LTM Long Term Monitoring MAROS Monitoring and Remediation Optimization System MCL Maximum Concentration Level NAPL N on A queous Phase Liquids ND Non Detect PRG Preliminary Remediation Goal RCRA Risk based Corrective Action ROC Rate of Change SF Slope Factor Version 2 1 i Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE UST Underground Storage Tank Version 2 1 October 2004 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE INTRODUCTION The AFCEE Monitoring and Remediation Optimization System MAROS Software is a Microsoft A ccesse database application developed to assist users with groundwater data trend analysis and long term monitoring optimization at contaminated groundwater sites This program was developed in accordance with the Long Term Monitoring Optimization Guide Version 11 developed by AFCEE The Monitoring and Remediation Optimization System MAROS methodology provides an optimal monitoring network solution given the parameters within a complicated groundwater system which will increase its effectiveness By applying statistical techniques to existing historical and current site analytical data as well as considering hydrogeologic factors an
419. statistical analysis is applied to the moments to identify a trend in this case Mann Kendall Trend Analysis is applied The role of moment analysis in MAROS is to provide a relative measure of plume stability and condition but can also assist the user in evaluating the impact on plume delineation in future sampling events by removing identified redundant wells from a long term monitoring program Plume stability may vary by constituent therefore the MAROS Moment analysis can be used to evaluate multiple COCs simultaneously which can be used to provide a quick way of comparing individual plume parameters to determine the size and movement of constituents relative to one another To estimate the mass center of mass and the spread of the plume at each sample event spatial moment analysis of the discrete groundwater monitoring data was performed The ijkth moment of the 2 D concentration distribution in space M ix t is defined as Freyburg 1986 Foo Fco 4 co My r nc y z th y z dxdydz where C x y z is the concentration at a monitoring point nis the porosity and x y z are the spatial coordinates The zeroth first and second moments i 4j H 0 1 or 2 respectively provide measures of the mass location of the center of mass and spread of the plume The moment trends over time can be assessed by the Mann Kendall test which is a non parametric statistical procedure that is well suited for analyzing trends
420. suggestions may make the process easier 1 Limit your data at first Perform a preliminary analysis with a small file of the most recent data in electronic format to check for data format issues Creating a small test file may highlight common problems with data such as misspellings of well names and COCs numbers entered as text and missing data Finding and correcting these issues early can save considerable time 2 Precise data input All constituents must be spelled exactly as in the MAROS_ConstituentList xls under the heading MAROS Constituent Names see table A 1 6 For example BENZENE is recognized by MAROS but not benzene or BZ Cutting and pasting names from the constituent list is a good strategy Results and detection limits should be entered as numbers and not as text Detailed data input formats are discussed below Excel and Access Formats The following format for Microsoft Excel and Microsoft Access Table Files Table A 1 1 should be used for importing files into M AROS from Excel and Access The Well Name is a text field and dashes and other symbols can be included in the name The Well Name should be spelled consistently throughout the file The X and Y Coordinates should be in feet The coordinates can be in a geographic coordinate system such as State Plane or in a custom system such as plant coordinates The Constituent Naming convention follows ERPIMS As described above all constituent names must be spel
421. sults RESULTS AND INTERPRETATION The role of moment analysis in MAROS is to provide a relative measure of plume stability and condition over time but can also assist the user in evaluating the impact on plume delineation in Version 2 1 A 5 6 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE future sampling events by removing identified redundant wells from a long term monitoring program Plume stability may vary by constituent therefore the MAROS M oment analysis can be used to evaluate multiple COCs simultaneously which can be used to provide a quick way of comparing individual plume parameters to determine the size and movement of constituents relative to one another Zeroth M oment Trend The Zeroth M oment trend over time will allow the user to understand how the plume mass has changed historically A Concentration Trend and Confidence in Trend are reported for each sample event see Figure A 5 2 Zeroth moment calculations can show high variability over time largely due to the fluctuating concentrations at the most contaminated wells Field data can be highly variable dueto changes in physical factors such as aquifer recharge and temperature Plume analysis and delineation based exclusively on concentration can exhibit a large degree of temporal and spatial variability When considering the results of the Zeroth moment trend take into consideration
422. t Analysis Summary Project User Name Location Service Station State Texas Oth Moment 1st Moment Center of Mass 2nd Moment Spread Estimated Source Sigma XX Sigma YY Number of Effective Date Mass Kg Xc ft Yc ft Distance ft sq ft sq ft Wells BENZENE 10 4 1988 1 7E 01 46 49 67 980 2 591 12 11 17 1989 1 2E 01 38 48 61 1 165 5 923 12 3 1 1990 1 0E 01 47 61 77 1 234 2 769 12 5 31 1990 6 4E 02 48 48 68 1 369 3 937 12 9 13 1990 6 2E 02 43 59 73 987 3 106 12 4 3 1991 4 8E 02 41 53 68 849 2 891 12 7 10 1991 5 5E 02 41 59 72 860 3 080 12 10 3 1991 7 4E 02 43 60 74 896 3 269 12 5 2 1992 2 6E 02 42 70 82 1 254 5 210 12 1 11 1994 2 5E 02 44 80 91 1 164 3 844 12 5 28 1996 2 8E 02 41 75 85 909 3 386 12 6 27 1997 1 6E 02 49 94 106 1 118 4 164 12 12 10 1997 6 8E 03 48 103 113 1 486 5 578 12 6 19 1998 6 1E 03 57 96 112 1 540 5 138 12 12 19 1998 1 4E 03 56 109 122 2 534 9 481 12 MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page 1 of 2 Project User Name Location Service Station State Texas Coefficient Mann Kendall Confidence Moment Moment Type Consituent of Variation S Statistic in Trend Trend Zeroth Moment Mass BENZENE 0 88 91 100 096 D 1st Moment Distance to Source BENZENE 0 23 83 100 096 l 2nd Moment Sigma XX BENZENE 0 35 35 95 4 l 2nd Moment Sigma YY BENZENE 0 42 53 99 6 l Note The following assumptions were applied for the calculation of the Zeroth Moment Porosity 0
423. t based on season variation in rainfall or other hydraulic considerations No appredable movement or a neutral trend in the center of mass would indicate plume stability However changes in the first moment over time do not necessarily completely characterize the changes in the concentration distribution and the mass over time Therefore in order to fully Version 2 1 A 5 7 Air Force Center for November 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE characterize the plume the First M oment trend should be compared to the Zeroth moment trend mass change over time refer to Figures A 5 3 A 5 5 Dissolved Mass Dissolved Mass 100 kg 200 kg Oth Moment Increases 1st Moment Increases EE x center of mass Figure A 5 3 Moment Analysis Mann Kendall First Moment Trend Results Zeroth M oment Dissolved M ass Increases over time and the First M oment Increases over time Dissolved Mass Dissolved Mass 100 kg 10kg Oth Moment Decreases 1st Moment Increases x center of mass Figure A 5 4 Moment Analysis Mann Kendall First Moment Trend Results Zeroth M oment Dissolved M ass Decreases over time and the First Moment Increases over time Dissolved Mass Dissolved Mass 100 kg 10kg Oth Moment Decreases 1st Moment Decreases x center of mass Figure A 5 5 Moment Analysis Mann Kendall First Moment Trend Results Zeroth Moment Dissolved Mass Decreases over ti
424. t s ite me or iter st IDCM View s Tus pi Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Site D etails Well Locations accessed from the Wal Ey i Coordinates Screen allows the user to review the Weil Locations well coordinates in their relative locations Well diri reet iro Ure Senet rmn i sers om commete can he sedi coordinates are mandatory and should be in feet eg State Plane coordinates or arbitrary site coordinates Back Returns the user back to the Well Coordinates screen Constituents of Concern Decision accessed from the Source Tail Zone Selection Screen allows the user to define up to five constituents to be evaluated at the site amp Monitoring and Remediation Optimization System MAROS icf x Constituents of Concern Decision Enter up to 5 COCs for the site in the boxes Enter up to 5 COCs for the site in the boxes to the right 5 is the maximum if you have more than 5 then run to the ri ght 5 l S the Max mu m if you have the fft ti If ild like t list of sted COCs click on the button Ri ided COCs This Wil result in a summarized Ist of COC recommendations from the avaiable dataset as welasa more than 5 then run the software more detailed view of the process used to make the COC recommendation times In general choosing 1 to 3 COCs with different chemical characteristics per analysis COCs for
425. t savings over the long term at this BTEX site An approximate cost savings of 60 per year is projected for the tutorial site while still maintaining adequate delineation of the plume as well as knowledge of the plume state over time At more complex sites with many more wells and more sampling data cost savings would be greatly increased Version 2 1 A 11 70 Air Force Center for October 2004 Environmental Excellence
426. t to weight the wells choose Do Not Weight Wells below AIICOCs WellHame Source Tail Weight a C Weight Wells DoNot Weight Wells Note Increasing 1 Probably Increasing PI Stable 5 Probably jecteasing PD Decreasing D No Trend NT Not Applicable N A Source Tail S T lt lt Back Next gt gt Help Wells on the right side of the screen When finished click Next to see results of weighting Back Returns the user to the Results of Information Weighting screen Next Takes the user to the M onitoring System Category screen Help Provides information on the screen specific input requirements Version 2 1 October 2004 49 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE MAROS Analysis Overall Analysis M onitoring System Category accessed from the Plume Information by Wal Weighting screen allows the user to view the suggested design category for each COC Trend results for both tail and source wells mme are given From these results a monitoring Monitoring System Catagory system category that characterizes the site Tail ESN for an individual constituent is shown Monitoring System Categories te M Moderate L Limited Plume Status ANS ES Categories include Extensive E Moderate M and Limited L long term monitoring required for the site L L E PD Decr
427. take you to the D ata M anagement M enu Screen 2 Data Management M enu From the D ata M anagement M enu select Import New Data by clicking on the button next to the label This will take you to the Import N ew D ata Screen 3 Import New Data Choose the type of data import to be performed by dicking on the appropriate button Excel or ERPIMS Enter the full file path and filename of the file to import or dick the browse button to find the import file The Folder and File name you choose will appear in the top two boxes See Notes below for ERPIMS and Excel file format names Choose the import option that corresponds to the import data format N ote that the Import New Data option will replace the existing data in the database Click Version 2 1 6 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Import to proceed with importing the file to the existing database See Appendix A 1 for more information To enter individual data records manually within the software 1 Main M enu From the M ain M enu select Data Management by clicking on the button next to the label This will take you to the D ata M anagement M enu Screen 2 Data Management M enu From the Data M anagement M enu select Manual Data Addition by dicking on the button next to the label This will take you to the M anual Data A ddition Screen 3 Manual Data Addition Fill in th
428. tamination from the center of mass The Second M oment represents the spread of the plume over time The information displayed in this screen can also be viewed in report form Spatial Moment Analysis Report from the MAROS Output Screen or by clicking on View Report see Appendix A 10 for an example report The next screens will go through each moment analysis result in detail as well as looking at trends in the data over time For further details on the Spatial Moment Analysis Method see A ppendix A 5 View Report To print the Spatial Moment Analysis Report and analysis results click View Report to proceed Back Returns the user to the M oment A nalysis Site D etails Next Takes the user to the Z eroth M oment Plot Screen Help Provides information on the screen specific input requirements Version 2 1 38 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Spatial MomentAnalysis Change in Dissolved M ass Over Time Zeroth M oment Plot accessed from the M oment Analysis Statistics screen allows the user to view the Zeroth Moment Analysis results by constituent over time The zero moment in MAROS calculates an estimate of the mass of a constituent in the plume for each sample event The estimated mass over time is then evaluated using the Mann Kendall method to determine the trend in total mass of the plume over time ee Choose th
429. tandard deviation divided by the average Values near 1 00 indicate that the data form a relatively close group about the mean value Values either larger or smaller than 1 00 indicate that the data show a greater degree of scatter about the mean View Report To print the Zeroth Moment Analysis Report and analysis results dick View Report to proceed Back Returns the user to the M oment A nalysis Site D etails Next Takes the user to the First M oment Plot Screen Help Provides information on the screen specific input requirements Note The information displayed in this screen can also be viewed in report form Zeroth Moment Report from the MAROS Output Screen or by dicking on View Report see Appendix A 10 for an example report For further details on the Mann Kendall Analysis Method or Moment Analysis see A ppendix A 2 and A 5 respectively Version 2 1 39 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Spatial Moment Analysis Distance from Source to Center of M ass First M oment Plot Distance from Source to Center of M ass accessed from the Zero M oment Plot screen allows the user to view the First Moment Analysis results by constituent over time The first moment estimates the center of mass coordinates Xc and Yc for each sample event and COC The distance from the original source location to the center of mass locations indicate the movement of
430. ted in the plot area of the chart sheet Plot Voronoi Diagram By checking this box the Voronoi diagram or Thiessen polygon will be plotted in the plot area of the chart sheet Ok Saves user changes to the parameters and doses this form The changes will be effective the next time the user performs an optimization The drawing options will be effective immediately Cancel Cancels user changes and quits the form The Shortcut M enu allows you to locate a node location on the graph and sets its Selected status and Removale status easily The shortcut menu is available only in the Well Locations chart sheet Remove from system b Addtosystem SK Make Removable f Make Irremovable program To access the Shortcut M enu click left mouse button on a node or the name of the node beside it Click again at the same place and the shortcut menu will pop up The first click ensures the data series has been selected The second click returns the node information to the Remove from system Excludes a node from the network by setting Seected status to False Add to system Includes or inserts a node into the network by setting Selected status to True Make Removable M akes a node removable by setting Removable status to True Make Irremovable M akes a node irremovable by setting Removable status to False Version 2 1 October 2004 66 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION
431. tential seasonal effects in the monitoring data se yearly averages annual mean concentrations from years specified below C Use original data concentrations from sampling events specified below S g amp El E Ei 3 Ej amp Es S Q g 3 5 2 8 5 E z Q a s ie z mr a Ei a AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE and may also remove autocorrelation If there are many years of data available using yearly averages is recommended In this example select Use yearly averages Select the beginning and ending sampling events from the From and To dropdown lists to define the period to be used in the analysis Select 1991 from the From 3 Monitoring and Remediation Optimization System MAROS E x Individual Well Cleanup Status Results The cleanup status for a monitoring well is evaluated with the sequential t test from EPA 1332 Sample size is the number of concentration data in the time period selected by the user The data are assumed to be either normally or lognormally distributed and results under both assumptions can be compared Also available is an optional analysis where power analysis based on Student s t test on mean difference is performed Results shown are based on yearly averages YES 13 The Individual Wal Cleanup Status Results screen shows the analysis result
432. ter for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE APPENDIX A 10 MAROS SAMPLE REPORTS 1 COC Assessment Summary 2 Linear Regression Statistics Graph 3 Linear Regression Statistics Summary 4 Mann Kendall Statistics Graph 5 Mann Kendall Statistics Summary 6 Spatial Moment Analysis Summary 7 Zeroth First and Second Moment Graphs 8 Plume Analysis Summary 9 Site Results Summary 10 Sampling Location O ptimization Results 11 Sampling Location O ptimization Results Summary 12 Sampling Location Optimization Graph 13 Sampling Frequency Optimization Results 14 Power Analysis Individual Well Cleanup Status 15 Power Analysis Individual Well Cleanup Status Graph 16 Power Analysis Individual Well Cleanup Status O ptional 17 Risk Based Power Analysis Regression Results 18 Risk Based Power Analysis Projected Concentrations 19 Risk Based Power Analysis Site Cleanup Status Version 2 0 A 11 1 Air Force Center for November 2003 Environmental Excellence MAROS COC Assessment Project User Name Location Service Station State Texas Toxicity Representative Percent Concentration PRG Above Contaminant of Concern mg L mg L PRG LEAD 1 0E 01 1 5E 02 67296 0 BENZENE 2 1E 01 5 0E 03 4073 5 1 1 1 2 TETRACHLOROETHANE 3 8E 01 1 1E 01 241 2 1 2 DICHLOROBENZENE 9 8E 01 6 0E 01 64 1 TOLUENE 1 5E 00 1 0E 00 50 4 BARIUM 3 2E 00 2 3E 00 37 4 C
433. th investigative and remedial actions e g see EPA 1999 A conceptual site model CSM is a three dimensional representation that conveys what is known or suspected about contamination Sources release mechanisms and the transport and fate of those contaminants The conceptual model provides the basis for assessing potential remedial technologies at the site In the context of the MAROS software conceptual model development prior to software use would allow the user to better utilize the information gained through the various software modules as well as provide guidance for assessing the data that would best typify historical site conditions It is recommended that available site characterization data should be used to develop a conceptual model for the site prior to the use of the MAROS software The conceptual model Version 2 1 A 2 1 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE should include a three dimensional representation of the source area as a NAPL or region of highly contaminated ground water of the surrounding uncontaminated area of ground water flow properties and of the solute transport system based on available geological biological geochemical hydrological climatological and analytical data for the site EPA 1998 Data on the contaminant levels and aquifer characteristics should be obtained from wells and boreholes which will provide a dear thre
434. that the data form a relatively close group about the mean value A COV either larger or smaller than 1 00 indicates a greater degree of scatter about the mean Step 5 Determine the concentration trend by checking the decision matrix presented in Table A 7 11 with Mann Kendall statistic S confidence in trend CT and coefficient of variation COV For example if S gt 0 and CT gt 95 the concentration trend is Increasing Table A 7 11 Mann Kendall analysis decision matrix Mann Kendall Confidence Concentration Statistic in the Trend Trend 20 9596 Increasing 20 90 9596 Probably Increasing 20 9096 No Trend S lt 0 lt 90 and COV 2 1 No Trend S lt 0 lt 90 and COV lt 1 Stable S lt 0 90 95 Probably Decreasing S lt 0 95 Decreasing Example Step 1 Benzene concentrations from a monitoring well are presented in Table A 7 12 The signs of the difference between consecutive measurements are presented in the third to ninth rows of Table A 7 12 For example the sign of the difference between the first and third measurements is sgn Xs X1 sgn 0 034 0 026 sgn 0 008 1 Step 2 TheMann Kendall statistic S is found to be 8 The calculations are shown in Table A 7 12 Step 3 Consulting a Kendall probability table with S 8 and n 8 finds the confidence in the trend to be 0 801 or 80 196 In fact the Kendall probability table provides the probability that the unsigned M ann Kendall statistic S equals or
435. the right of the screen vis By Select Next to proceed to the First M oment An kgac i Plot Change in Location of Mass Over Time Screen Click hereto proceed Note If more than one COC was being used plots of other chemicals can be obtained using the Chemical drop down box at the top of the screen followed by selecting the Graph button The graph type can be specified as Log or Linear The graph can be printed by selecting the View Report button Data values can be viewed by selecting View Data This shows a table with the Xc Yc and Source Distance for all sample events SPATIAL MOMENT ANALYS S 7 Second Moment Plot Change in Plume Spread Over Time allows the user to view the Second Moment Analysis results by constituent over time The second moment indicates the spread of the contaminant about the center of mass Sxx and Syy or the distance of contamination from the center of mass Analysis of the spread of the plume should be viewed as it relates to the direction of groundwater flow The Second Moment represents the spread of the plume over time in both the x and y directions The Second M oment trend of the Spread of the Plume in the X or Y direction over time is determined by using the Mann Kendall Trend Methodology The Second Moment trend for each COC is determined according to the rules outlined in Appendix A 1 Results for the trend indude Increasing Probably Increasing No Tren
436. the Source Tail Zone Selection screen Continue to the Constituents of Concern D ecision screen to choose the representative COCs for the site 4 Constituents of Concern From the Constituents of Concern screen click on Recommended COCs The next screen Risk Leva Assessment will show the data for COCs that are currently in the database to be evaluated Choose from the list of generic Preliminary Remediation Goal PRG recommendations Choose from the list of generic Preliminary Remediation Goal PRG recommendations Click on the appropriate standard to be used in database comparisons for COC recommendations Enter your own modifications to cleanup goals under custom goals in mg L The next screen COC Decision screen shows up to 10 of the recommended COCs based on Toxicity Prevalence and Mobility Enter up to 5 COCs for the site in the boxes to the left If you would like a detailed view of the process used to make the COC recommendation click on Toxicity Prevalence or Mobility at the left side of the screen The information displayed in this screen can also be viewed in report form COC Assessment Report from the M AROS Output Screen To proceed with the next step in the software click Back How can I access the Sampling Optimization module The Sampling Optimization module is an optional extension of the MAROS software It may optimize the sampling plan by eliminating redundant sampling locations and determining t
437. the cumulative normal distribution Step 4 Plot the normal quantiles verses the concentration of each measurement i e yi verses xi If these points approximate a straight line it is evidence that the data are normally distributed Significant bends or curves in the plot indicate departures from a normal distribution Example Step 1 Hypothetical arsenic data from four wells are presented in Table A 7 13 These measurements are ordered from smallest to largest in the fourth column of Table A 7 13 The order of each measurement is listed in the fifth column of Table A 7 13 Step 2 The cumulative probability corresponding to each measurement is given in the sixth column of Table A 7 13 For example for the third smallest measurement jee 3 208 wel 1621 Step 3 The normal quantile corresponding to each of the cumulative probabilities is listed in the last column of Table A 7 13 In this example they are calculated with the function NORMINV in Microsoft Excel Step 4 The normal probability plot is presented in Figure A 7 2 The points do not approximate astraight line very well but bends in the plot are not significant indicating the data are approximately normally distributed Version 2 1 A 7 25 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Well MW 1 MW 2 Mw 4 Version 2 1 October 2004 Table A 7 13 Example aresenic data
438. the following factors which could effect the calculation and interpretation of the plume mass over time 1 Change in the spatial distribution of the wells sampled historically 2 Different wells sampled within the well network over time addition and subtraction of well within the network 3 Adequate versus inadequate delineation of the plume over time BENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL Moment Analysis Results Last column is the result for the trend Confidence Moment Moment COV MK S in Trend Trend Zeroth Moment Mass 0 8 25 88 096 S 1st Moment Distance to Source 0 6 67 100 0 l 2nd Moment Sigma XX 0 7 23 85 996 NT 2nd Moment Sigma YY 0 9 13 721 NT Note Increasing I Probably Increasing FI Stable S Probably Decreasing PD Decreasing D No Trend NT Not Applicable N A Source T ail S T COV Coefficient of Variation MK S Mann Kendall Statistic Figure A 5 2 Moment Analysis M ann Kendall Trend Results First Moment Trend The First Moment trend of the distance to the center of mass over time is shows movement of the plume in relation to the original source location over time Analysis of the movement of mass should be viewed as it relates to 1 the original source location of contamination 2 the direction of groundwater flow and or 3 source removal or remediation Spatial and temporal trends in the center of mass can indicate spreading or shrinking or transient movemen
439. the representative statistical dataset within the consolidated time interval If the user decides to consolidate the data yearly for instance the statistic chosen e g average will be the representative result for the year Version 2 1 A 11 19 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE 3 Data Reduction Part 2 of 2 allows the user to consolidate the data based on concentration parameters chosen The N on D etect N D option allows the user to choose the number value to represent a non detect result in the data To apply a specific detection limit for each chemical choose Uniform Detection Limit The suggested detection limit is the minimum detection limit The D uplicates option is used to consolidate duplicates N ote that duplicates are samples that have the same constituent date and well name Samples with the same effective date will be consolidated as duplicates The Trace TR option is used to specify the number value which will be used to represent a Trace result in the data The TR flag is equivalentto the J flag used by most labs to indicate a result that is reported but is below the method detection limit This particular tutorial will use a uniform detection limit of 0 001 mg L to represent non detect results duplicates will be consolidated using the average value and trace results will be analyzed based on the actual valu
440. there is a button named View Report Click this button and follow instructions to view or print the results report MAROS Output Reports After running through the Sampling Optimization module the M AROS Output Reports screen can be accessed from screen M ain M enu From the Report listbox select the report you want to view eg Sampling Location Optimization Report by dicking on that item available only after that sub module has been successfully performed Then click button V iew Print Report and follow instructions to view or print the report How will the Sampling O ptimization module help me optimize a sampling plan The Sampling Optimization module is used to determine the minimal number of sampling locations and the lowest sampling frequencies that can still meet the requirements of spatial sampling and temporal sampling for the monitoring program A data sufficiency analysis is also provided in this module to examine the deanup status and the significance of concentration trend at individual wells and the risk based site cleanup status These analyses are based on each Constituent of Concern COC and the results are given on a COC by COC basis 1 Sampling Location Analysis This sub module uses the Delaunay method to eliminate redundant wells from the monitoring network based on spatial data analyses Monitoring data from multiple sampling events can be used in this analysis Major steps to be followed are a Sampling Locatio
441. tion is also an important factor impacting chlorinated solvent plumes Therefore the seepage velocity should be accurately determined to predict plume lengths e Environmental factors such as temperature pH dissolved oxygen and redox potential were not strongly correlated with chlorinated ethene plume length However there was a strong trend of increasing PCE plume length with increasing redox potential once the PCE plume length was normalized to remove the effects of advection These results suggest that source width and strength and seepage velocity are more important factors impacting overall plume length than environmental conditions that are conducive to reductive dechlorination Lawrence Livermore Study McNab amp al 1999 collected and analyzed data from 65 sites representing a variety of hydrogeologic settings and release scenarios eg large industrial facilities dry cleaners and landfills Data collection involved a variety of federal and state agencies and induded participation from the U S Department of Defense the Department of Energy and private industry The distribution of chlorinated solvent plume lengths from their database is shown in Table A 4 5 TABLE A 4 5 SUMMARY OF FREQUENCY DISTRIBUTIONS OF MAXIMUM CVOC PLUME LENGTHS FT TO THE 10 PPB DEFINED PLUME PER SITE BASED ON THE INDICATED CONCENTRATION CONTOUR DEFINITION 2596 Percentile 1096 Percentile 120 Number of Sites 99 Key results from t
442. tion within the software The columns must include the field names as they are in the Access Template file and the table name should be ImportData The template file MAROS Accessi mportTemplate mdb is provided with the software with example data Version 2 1 15 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Also a lit of permissible constituent names is found in the file MAROS ConstituentList xls To import ERPIMS files 1 Ensure that the source folder contains the SAM TES RES and LDI data files 2 Typeor select only the RES file to import all needed files Before importing ERPIMS files they must be saved in text format in Microsoft Word 2000 with fields identical to those already in the database system i e the format matching that used by ERPIMS system To save the ERPIMS files as text files open each file SAM TES RES and LDI files one at a time in Word You will be prompted to Choose the encoding used for loading this file check Plain Text When the file is opened in Word under the Menu option dick Save as You will be prompted to Save as type choose Text only txt Make sure you do not have the txt extension on the end of the file name only the original file name with the RES SAM TES or LDI file extension should appear All files should have the same name eg Hillgwdata RES Hillgwdata LDI Hillgwdata TES and
443. tions for the sampling events selected by the user The user can go back to re run the analysis by changing parameters or selecting a different series of sampling vents Next Proceed to the Sampling Frequency Analysis Complete screen Help Provides additional information on software operation and screen specific input requirements Version 2 1 71 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Sampling Frequency Analysis Complete This screen accessed from the Sampling Frequency Recommendation screen by clicking Next is a message screen indicating that sampling frequency determination has been completed and the user can proceed to other analyses ti Sex Frere P ustintan md Pm dante Corte Back Returns to the Sampling Frequency E Sampling Frequency Analysis Complete Recommendation screen The user can go back to re E run the analysis by changing parameters or selecting E NOS IE ENIAN SPORE ER a different series of sampling vents T pere praed ts ote ties of angie pinrsion Tournay dgs b back io debeo Gere conr pari oue or ona pores es gt wrt in nin Evi rented i z usi EM Sampling Optimization Returns the user to the a Sampling Optimization screen a 3 lt lt Back Version 2 1 72 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE D ata Suff
444. tline and is not intended to be a guidance style flowchart Table A 7 17 Appropriate statistical methods used in long term monitoring programs Procedures Method Scope of Use Objective Method Description Method 1 Detection To determine if The control chart method is sensitive to Combined monitoring groundwater is both immediate and gradual releases As Shewart CUSUM contaminated and if an intra well comparison method control chart compliance monitoring problems associated spatial variations can is required be completely avoided Method 2 Detection To determine if The method is capable of controlling the Inter well and monitoring groundwater is facility wide false positive rate FWFPR intra well contaminated and if and minimizing the false positive and false prediction limits compliance monitoring negative rates associated with a single is required comparison Method 3 Compliance To find if there is The method is easy to perform The Confidence limits monitoring statistically significant requirement that the confidence limit be evidence of compared to the ACL from background contamination and if samples instead of the MCL is protective of corrective action human health or the environment EPA monitoring is required 1992a Method 4 Corrective To test if groundwater The method requires fewer samples and a Sequential test action has attained the cleanup corresponding shorter time to make the method monit
445. to Choose the graph type i e Log or Linear EIER v Choose the Covariance Type i e Sxx or Syy HO JT P Linear Click Graph on graph to proceed LEGG oes ys SF Covaiance Type 10000 4 t Sxx Second M oment Trend The Second Moment trend of the Spread of the Plume in the X or Y E direction over time is determined by using the Mann Kendall Trend Methodology The Confidence in Second Moment trend for each COC is see E determined according to the rules outlined in Oi a ap IURI errem nnn enini miae mas v Appendix A 2 Results for the trend indude pee e VewRepon Heb Increasing Probably Increasing No Trend Stable Probably Decreasing Decreasing or Not LA ee eem 1000 Ser o 2 gt E 2 W Q z x T Q 2 A pplicable Insufficient Data MK S The Mann Kendall Statistic S measures the trend in the data Positive values indicate an increase in the spread of the plume over time expanding plume whereas negative values indicate a decrease in the spread of the plume over time shrinking plume The strength of the trend is proportional to the magnitude of the Mann Kendall Statistic i e large magnitudes indicate a strong trend Confidence in Trend The Confidence in Trend isthe statistical confidence that the spread of the plume in the x or y direction is increasing S20 or decreasing S 0 COV
446. to stop sampling for that COC at that location since other COCs may still need to be sampled at this location Monitoring and Remediation Optimization System MAROS d loj x SF val ue D i splays the u mped SF val ue Access Module Results by COC of a location that is calculated by Samping cations for each COC sre determined as shown in the folowing table Those tedden averaging the SF values obtained in each sampling locations marked as Eliminated are eliminated from the monitoring network Eliminated a ss be sse here as stopping Esa fora piu COC ata piss sampling location sampl l ng event across al l samp l l ng BENZENE ETHYLBENZENE TOLUENE xYLENES TOTAL events selected by the user LociD ESCoord NSCoord SFvalue Eliminated Bl El i mi nated D i splays Ww hether Or not a MW 1 130 200 0 066 4 EN vane w0 80 025 j location is considered redundant and z M should be eliminated A check mark in e MW 14 1020 200 0433 d t i z Mis N00 259 03i8 u this field stands for the elimination of a IE location E MW 4 550 37 0 0 13 m e Ws zm Ste EN Back Returns the user to the Access dmn sepe sequela e cr M odule Slope Factor V alues screen a z lt lt Back View Report Compare Across COCs gt gt View Report Generates a report with sampling location optimization results for each COC This report can be viewed or printed The user can go back to re run the
447. ts for correlated data skewed data or correlated and skewed data either normally distributed or lognormally distributed data the log likelihood ratio method performs best among the other methods tested U S EPA 1992 This sequential t test has several advantages First for assessing attainment the objective is to test a hypothesis rather than to obtain an unbiased estimate of the mean or construct a confidence interval Second if the concentrations at the site are indeed below the cleanup Version 2 1 A 6 3 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE standards maintaining the expected power at the alternative hypothesis can protect against incorrectly concluding that additional deanup or monitoring is required Third reducing the sampling size results in cost savings for the monitoring program Fourth a good estimate of the measurement variance for calculating the sample size for the fixed size test may not be available In cases where there are not enough yearly averages available for analysis the original data from each sampling event without being yearly averaged are also allowed for the sequential t test An option is provided in MAROS for the user to choose between the two types of data An optional power analysis to the cleanup status evaluation is also provided This analysis uses the Student s t test on mean difference to determine 1 whether the
448. ts from the plume centerine regression analysis are given below for sampling analysis in the sampling event If this events selected by the user The regression coefficient is equal to the slope of the regression line of log E 2 n Ree re co eae ape eis Se ere caine cried number is less than three regression will not be performed BENZENE ETHYLBENZENE TOLUENE XYLENES TOTAL Sampling Event effectweDate rS Renression Confidence Regression Coefficient The first order Sample Event 12 5 27 1997 4 0 02605 67 0 coeffi ci ent 1 ft of the exponential Sample Event 13 12401997 4 0 024076 67 0 H Event 14 69998 4 0 023664 73 3 mod el Ww here p u me centerl l ne Sample Event 15 12941998 4 0 022742 67 096 concentrati ons are exp ressed as a Sample Event 2 115174989 3 0 006948 53 7 T S a Event 3 3171980 3 0 060229 358 fu nction of the d istance d own the pl ume Sores Everts ESE EL e centerline This regression coefficient is Sample Event 5 94343990 3 0 043764 95 8 gt 7 h a equivalent to the slope of the regression Z Note when the number of wells is less than 3 no regression is performed and all values are set to 0 i n e of O g tr an sf orm ed C enterl i n e z a ae eee ene See i concentrations against the distance down z Debak Wow Report mex Hem the plume centerline A negative coefficient indicates that the centerline concentrations drop with an increase in distance from the source If the
449. ts to statistics computed from historical measurements from the same well The use of intra well analysis can eliminate the problem caused by spatial variability between wells in different locations and should be used whenever possible e identically distributed Samples used in a statistical test have the same population distributions This assumption forms the basis of parametric and most nonparametric tests used in groundwater monitoring M ost parametric statistical methods assume data Version 2 1 A 7 8 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE or their transformations are normally distributed For nonparametric tests the distribution of data does not have to be normal but need to be identical in most cases Sometimes the normality of data can be achieved by transforming the original observations to make them normally distributed e g lognormal An algorithm for choosing the best transformation power is available in N eter et al 1996 Usually parametric methods need a smaller sample size and have a higher power than their nonparametric counterparts It is recommended that parametric methods be used for data evaluation whenever possible The use of any statistical method that fails to take the above assumptions into consideration may result in excessive false positive and false negative rates FACILITY WIDE FALSE POSITIVE NEGATIVE RATE Another issue
450. tted verses ti From Figure A 7 1 and Table A 7 7 we can see the process is out of control both in terms of absolute value and trend on thethird quarter of 1991 This result is confirmed in the fourth quarter of 1991 Table A 7 7 Example dataset for constructing the Shewart CUSUM charts Quarter Year Period Concentration Standardized CUSUM li Xi Zi Zi C Si 1 90 1 50 0 1 0 2 90 2 40 si 2 0 3 90 3 60 1 0 0 4 90 4 50 0 1 0 1 91 5 70 1 1 2 91 6 80 3 2 3 3 91 7 100 5 4 T 4 91 8 120 T 6 13 Shewart out of control limit exceeded z gt SCL 4 5 CUSUM out of control limit exceeded S gt h 5 Combined Shewart CUSUM Charts e Standardized Mean m CUSUM c 3 pe N o c E c iv o c o G x c o o c o o O NURON 1 N Jan 90 Apr 90 Jul 90 Oct 90 Jan 91 Apr 91 Jul 91 Oct 91 Jan 92 Sampling period Figure A 7 1 Example Combined Shewart cusum charts Version 2 1 A 7 12 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE METHOD 2 PREDICTION LIMITS IN DETECTION MONITORING Prediction limits are statistical estimates of the minimum or maximum concentration or both that will contain the next series of k measurements with a specified level of confidence e g 9996 confidence based on a sample of n background measurements In groundwater detection monitoring we are conc
451. ty table found in many statistical textbooks eg Hollander M and Wolfe D A 1973 The resulting confidence in the trend is applied in the Mann Kendall trend analysis as outlined in TableA 1 1 The Mann Kendall test is limited to 40 sample events AVERAGE The arithmetic mean of a sample of n values of a variable is the average of all the sample values written as STANDARD DEVIATION The standard deviation is the square root of the average of the square of the deviations from the sample mean written as The standard deviation is a measure of how the value fluctuates about the arithmetic mean of the data COEFFICIENT OF VARIATION COV The Coefficient of Variation COV is a statistical measure of how the individual data points vary about the mean value The coefficient of variation defined as the standard deviation divided by the average or COy x Values less than or near 1 00 indicate that the data form a relatively dose group about the mean value Values larger than 1 00 indicate that the data show a greater degree of scatter about the mean Version 2 1 A 2 5 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE RESULTS AND INTERPRETATION OF RESULTS MANN KENDALL ANALYSIS The Constituent Trend Analysis results are presented in the M ann Kendall Analysis Screen accessed from the Plume Analysis M enu The software uses the input data to calculate the
452. ual Annual the right screen the Sampling MW 14 Biennial Annual Annual Frequency for MW 15 is Biennial M 15 Biennial Annual Annual Both the Current and Overall Mer Senma Annual Annual results for MW 15 are Annual eios Shed E Myy 4 Annual Annual Annual Its recommended frequency can Myy 5 Annual Annual Annual zi be used in the future round of sampling Both parts of the sampling optimization sampling location determination based on the Delaunay Method and sampling frequency determination based on Modified CES method should be performed periodically to ensure regular optimization of the groundwater monitoring program Version 2 1 A 9 7 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE References AFCEE 1997 Long Term Monitoring Optimization Guide Version 1 1 HQ Air Force Center for Environmental Excellence Consultant Operations Division Brooks Air Force Base TX Barcelona M J et al 1989 Sampling Frequency for Ground water Quality Monitoring EPA 600 S4 89 032 Environmental M onitoring Systems Laboratory U S EPA NFESC 2000 Guide to Optimal Groundwater Monitoring Interim Final Naval Facilities Engineering Service Center Port H ueneme California Ridley M N et al 1995 Cost Effective Sampling of Groundwater Monitoring Wells the Regents of UC LLNL Lawrence Livermore National Laboratory Version 2 1 A 9 8 Air Force Cen
453. uating factors over time data can be consolidated to annual sampling and the zeroth moment trend re evaluated Another factor to consider when interpreting the mass increase over time is the change in the spatial distribution of the wells sampled historically If the service station site network had changes in the well distribution over time due to addition and subtraction of wells from the well network this could cause moment trends to be incorrect Also an observed mass increase could also stem from more mass being dissolved from the NAPL while a remediation system is operating The spatial and temporal trends in the center of mass distance from the source location first moment results can indicate transient movement based on season variation in rainfall or other hydraulic considerations The Service Station results that the source area concentration is decreasing faster than the tail area of the plume therefore the increasing trend in the first moment Even though the center of mass is moving the plume itself is still decreasing in concentration over time and the direction of movement is in the groundwater flow direction The second moment provides a measure of the spread of the concentration distribution about the plume s center of mass The second moment or spread of the plume over time in the x direction for each sample event shows an increasing trend over time Analysis of the spread of the plume indicates a shrinking to stable plume wh
454. uld be adequately delineated for the mass estimates to be considered The 3 D Zeroth Moment or M ass estimate was calculated using the following formula 00 00 00 Mass fnCidedvat where C is the concentration of the COC nis the porosity and x y z are the spatial coordinates Because the data are spatially discontinuous a numerical approximation to this equation is required To conduct the numerical integration the horizontal plane xy was divided into contiguous triangular regions with the apex of each triangle defined by a well sampling location with an associated COC concentration and saturated thickness at each sample location A spatial interpolation method over these triangles allows the zeroth moment calculations using Delaunay Triangulation see Appendix A 2 for methodology An approximation of the mass is obtained from calculating MASS gsinated YNV Cias where Ciavg is the geometric mean concentration of each triangle for a particular COC i Vi is the volume of the triangle calculated by d A where d is the averaged saturated thickness and A is the are of the triangle Zeroth Moment Trend The Zeroth Moment trend over time is determined by using the Mann Kendall Trend Methodology The Zeroth Moment Trend for each COC is determined according to the rules outlined in Appendix A 1 The Zeroth Moment trend test will allow the user to understand how the plume mass has changed over time Results for the t
455. up status evaluation Alpha Level signifi cance leva Target Power and Detection Limit ior Detection Lm usd nthe rl based power any fr mdi CO ES mg L The Cleanup Goal is by default set to the EFTA cleanup Target pha Target pesas Maximum Contaminant Level MCL of a COC If nd ili Arne Mr M e eL there is no available Cleanup Goal for a COC in the WIESE database the user is asked to define it and the TOLUENE 01 0 08 005 080 0 02 Targ amp Levd Otherwise the analysis for that COC SYUBES TOTAL al Ee CT nina will be canceled By default the Target Leva is set to 0 8PRG the Alpha Levd the significance level of a statistical test is set to 0 05 and the Target Power false negative rate is set to 0 80 In the risk based lt lt Back Set to Default power analysis the D etection Limit specified here is used to indicate whether the projected concentration is less than it If the user has already specified uniform D etection Limits in the D ata Reduction Part 2 of 2 screen they will show up in this screen as default values Refer to Appendix A 6 for details Back Returns to the D ata Sufficiency Analysis M enu screen Set to default Sets all parameters to the system default Help Provides information on the screen specific input requirements Version 2 1 74 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Individual W
456. urious data or outliers Here are some examples of questions to be asked e Which wells have trends for benzene concentration and which do not see Graph Trend Summary Results Graphing and Report M AROS PlumeA nalysis summary e From the trend analysis results is the plume increasing or decreasing see Report M AROS Spatial M oment Analysis Summary Zero M oment e Is the plume moving see Report MAROS Spatial Moment Analysis Summary First Moment e What are the trends in benzene concentrations over time see Graphs Linear Regression Graphs and M ann Kendall Graphs e Review the Mann Kendall and Linear Regression Trends Are there any differences for different wells See Report M AROS Plume Analysis summary and Mann Kendall and Linear Regression plots for individual wells e Are there wells on the outside of the monitoring network with concentrations increasing see Graph Trend Summary Results Graphing and Report MAROS Plume Analysis summary e Review data based on qualitative knowledge of the site for example is there a reason for one well to be showing a sudden increase in concentrations having been persistently non detect e Review the parameters selected for the data consolidation and the analysis see Report MAROS SiteResults Version 2 1 A 11 63 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE E Wane Das eg end Bervesel ee D pi rk
457. uthors Ling M and Rifai H S University of Houston In MAROS the Modified CES method is used to determine the sampling frequencies at all sampling locations for each COC The Modified CES method is developed based on the Cost Effective Sampling CES Ridley et al 1995 from Lawrence Livermore National Laboratory LLNL The Modified CES method is designed to reduce the sampling frequency based on the analysis of time series sampling data in each sampling location considering both recent trends and long term trends of the concentration data In contrast to the Delaunay Method that is based on the spatial analysis the Modified CES method is an approach based on temporal analysis Its combined use with the Delaunay M ethod leads to a complete process of sampling optimization Cost Effective Sampling Cost Effective Sampling CES is a methodology for estimating the lowest frequency sampling schedule for a given groundwater monitoring location while it can still provide the needed information for regulatory and remedial decision making Its initial development at LLNL was motivated by the preponderance of sampling results which fall below detection limits at two of its restoration sites The fact that so many locations had never shown or had ceased for some time to show any detectable levels of contamination suggested that those groundwater monitoring wells were being sampled more often than necessary The CES method recommends three steps
458. valuation a decision matrix was used to determine the Concentration Trend category for each well as presented on Table 2 MANN KENDALL STATISTIC S The Mann Kendall statistic S measures the trend in the data Positive values indicate an increase in constituent concentrations over time whereas negative values indicate a decrease in constituent concentrations over time The strength of thetrend is proportional to the magnitude of the M ann K endall Statistic i e large magnitudes indicate a strong trend Data for performing the Mann Kendall Analysis should be in time sequential order The first step is to determine the sign of the difference between consecutive sample results Sgn x Xx is an indicator function that results in the values 1 0 or 1 according to the sign of xj xx wherej gt k the function is calculated as follows sgn Xj Xk 1 if xj Xk gt 0 Version 2 1 A 2 4 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE sgn x Xk 0 if X XX 20 sgn x Xk 1 if xj Xk lt 0 The Mann Kendall statistic S is defined as the sum of the number of positive differences minus the number of negative differences or n l n S L seno x k l j k 1 The confidence on the Mann Kendall statistic can be measured by assessing the S result along with the number of samples n to find the confidence in the trend by utilizing a Kendall probabili
459. ve 2 12 261 i 1 Step 4 Consulting the Durbin Watson table given N eter et al 1996 page 1349 with a 0 05 and n 16 in the column titled p 1 1 wefind d 1 37 Since D 1 31 lt d 1 37 there is significant serial correlation in this time series and procedures adjusting for seasonal effects and serial correlation must be used as shown in the following steps Step 5 The mean square error of the deseasonalized residuals is m Hj 2 e M i _ 12 261 N m 16 4 2 s Se G _ 1022 1 0 33 _ 9 359 and the Df associated with itis N 1 0 Y 16 1 033 Version 2 1 A 7 38 Air Force Center for October 2004 Environmental Excellence 1 022 Thus the standard error of the mean is AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE N m 16 4 mo ud 3 3 Step 6 Thelower confidence limit on the sample mean is therefore X oy 55 133 t 4 995 X0 357 7 33 2 78 X 0 357 6 34 mg L Since this lower confidence limit contains the background standard which is 6 5 mg L it can be concluded that the contaminant concentration at this well is within compliance Version 2 1 A 7 39 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Table A 7 16 Example data for adjusting seasonal effects and serial correlation Quarter Data Seasonal Deseasonalized Br ei i e e 1 mg L average Residuals e 1 6 74 1
460. velocity Version 2 1 A 8 2 Air Force Center for October 2004 Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE TABLE A 8 2 FREQUENCY DETERMINATION FOR SITES WITH GROUNDWATER FLUCTUATIONS AND MONITORED NATURAL ATTENUATION TTR Monitoring System Category E M L Close TTR 2 yrs Quarterly Quarterly Biannually Medium 2 TTR 5 Quarterly Biannually Biannually yrs Far TTR gt 5 yrs Biannually Biannually Annually TTR time to receptor distance to receptor seepage velocity DURATION MAROS uses a simple decision matrix to assess when the design of the groundwater monitoring network should be reassessed for reducing the scope of the system or to stop monitoring altogether Users can compare the projected duration of the sampling at their site to the suggested duration of monitoring evaluated based on the decision matrix below The matrix was developed based on engineering judgment and experience of the authors It is not based on any kind of statistical analysis If their site has groundwater monitoring planned for a significantly longer time period then some reduction in the monitoring duration could be applied subject to local and federal regulations The sampling duration at the site is determined by the Monitoring System Category assigned by the results from the combined Source and Tail Stability Category as well as the length of the sampling record availab
461. vent 10 Results by Considering All COCs User Name Meng State Texas to Sample Event 15 1 11 1994 12 19 1998 Number COC Averaged Well X feet Y feet of COCs Slope Factor Abandoned MW 1 13 00 20 00 1 0 259 O MW 12 100 00 8 00 1 0 165 MW 13 65 00 23 00 1 0 254 LJ MW 14 102 00 20 00 1 0 064 L1 MW 15 190 00 125 00 1 0 421 O MW 2 2 00 30 00 1 0 308 L1 MW 3 35 00 10 00 1 0 117 MW 4 55 00 37 00 1 0 165 MW 5 4 00 70 00 1 0 532 LJ MW 6 77 00 5 00 1 0 526 LJ MW 7 87 00 75 00 1 0 417 O MW 8 55 00 95 00 1 0 645 LJ Note the COC Averaged Slope Factor is the value calculated by averaging those Average Slope Factor obtained earlier across COCs to be conservative a location is abandoned only when it is eliminated from all COCs abandoned doesn t necessarily mean the abandon of well it can mean that NO samples need to be collected for any COCs When the report is generated after running the Excel module SF values will NOT be shown above MAROS Version 2 2002 AFCEE Thursday November 20 2003 Page l of MAROS Sampling Frequency Optimization Results Project Example User Name Meng Location Service Station State Texas The Overall Number of Sampling Events 15 Recent Period defined by events From Sample Event 10 To Sample Event 15 1 11 1994 12 19 1998 Rate of Change parameters used Constituent Cleanup Goal Low Rate Medium Rate High Rate BENZENE 0 005 0 0025 0 005 0 01 Un
462. ventions contains about 2 100 constituents Example names for common constituents can be found in Table A 1 6 Note if using the Access file for importing the name of the Access table should be ImportData as in the MAROS AccessimportTemplate mdb file TABLE A 1 1 REQUIRED FIELD FORMAT FOR EXCEL AND ACCESS IMPORT FILES SAMPLING RESULTS Column Number Field Name Description Name of the groundwater well sampled be sure all wells are spelled the 1 WellName same X coordinate of the well although not mandatory it is suggested that you enter 2 XCoord this field for graphing purposes Y coordinate of the well although not mandatory it is suggested that you enter 3 Y Coord this field for graphing purposes Compound measured mandatory entry Follow the ERPIMS format of the 4 Constituent naming convention found in the Excel template file included with software 5 SampleDate Date Sample was collected format mmy dd yyyy 6 Result Analytical result enter result as a number if non detect then leave blank 7 Units Measurement units for result choices mg L ug L ng L g L pg L 8 DetLim Reporting Limit detection limit same units as Result Flag N D for non detect must enter the detection limit or TR for trace amount must enter both detection limit and the result if there is a detect in the 9 Flags Result column leavethe flag blank ERPIM S Format T
463. vironmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE do distance between node N o and its natural neighbor N i 3 The SF is then calculated as qp C NC Max EC NC where EC estimated logarithmic concentration at node N 0 N Co measured concentration in logarithmic scale at node N 0 The magnitude of SF ranges from 0 to 1 not including 1 Value 0 means that the concentration at a location can be exactly estimated by its surrounding locations thus sampling at this location provides no extra information to our knowledge of the plume A value larger than O indicates the existence of estimation error The larger is the estimation error the larger the discrepancy would be between the estimated concentration and the measured concentration at a location Consequently it becomes more reasonable to keep sampling at this location so that the plume can be better defined In summary the larger the SF value of a location is the more important is this location and vice versa Our objectives in spatial sampling are to accurately map a contaminant plume and track the change in this plume It is clear that with more monitoring wells these can be achieved in a higher degree of accuracy Unfortunately there is always a trade off between degree of accuracy and budget The limitation of resources forces us to find a way to use as few monitoring wells as possible as far as certain degree of accuracy can be
464. well Although the software will calculate trends for fewer than four wells and a minimum of 4 sampling events the above criteria will ensure a meaningful evaluation of COC trends over time The minimum requirements described would apply only to well behaved sites for most sites more data is required to obtain an accurate representation of COC trends Sites with significant variability in groundwater monitoring data dueto water table fluctuation variations in groundwater flow direction etc will require more data to obtain meaningful stability trends Essentially the plume you are evaluating should be delineated with adequate consecutive sampling data to accurately evaluate the concentration trend with time Plume Stability Analysis Confirmation of the effective performance of monitored natural attenuation as a stand alone remedial measure requires the demonstration of primary lines of evidence i e actual measurement of stable or shrinking plume conditions based on evaluation of historical groundwater monitoring data For a delineated plume a stable or shrinking condition can be identified by a stable or decreasing concentration trends over time For this analysis an overall plume condition was determined for each COC based on a statistical trend analysis of concentrations at each well as described below STATISTICAL TREND ANALYSIS CONCENTRATION VS TIME Under optimal conditions the natural attenuation of organic COCs at any site
465. will not be used in this tutorial Empirical results should be developed on the basis of data from previous similar site studies eg plume a thon studies such as the Lawrence Livermore study the BEG studies and the AFCEE chlorinated database For this tutorial there are no additional modeling results The option No separate empirical evidence to be applied should be already selected External Plume trfonmation Empirical Results istept n fee Luginent Csl eote Ube epose spamat reene mom Select Next to proceed to the External ond aibi dadas Plumelnformation Complete screen C 4 External Plume Information Complete Screen indicates that the Modeling and Empirical Trend results have been entered This portion of the software is an optional utility which will not be used in this tutorial To proceed to the Long Term Monitoring LTM Analysis to weight the Plume Information and analyze the trends in the groundwater data select Trends Analysis Manitoring and Rewediatioe Dpimizatiun System MAROS External Plume information Complete a hit heen ence ky the nod t wd le NACE Mere n The PlumeAnalysis M enu will appear Click hereto proceed omini tn Sire Se Od TREND ANAL Y S S Version 2 1 A 11 37 October 2004 Air Force Center for Environmental Excellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE STEP 6 MAROS
466. xample User Name Meng Location Service Station State Texas Groundwater Flow Direction 0 degrees Distance to Receptor 1000 feet From Period 10 4 1988 to 12 19 1998 Selected Plume Well Distance to Receptor feet Centerline Wells MW 12 1090 0 MW 4 1135 0 MW 1 1177 0 The distance is measured in the Groundwater Flow Angle from the well to the compliance boundary Sample Event Effective Date Number of Regression Confidence in enterline Wells Coefficient 1 ft Coefficient BENZENE Sample Event 1 10 4 1988 3 2 88E 02 87 0 Sample Event 2 11 17 1989 3 4 25E 02 90 796 Sample Event 3 3 1 1990 3 3 14E 02 87 6 Sample Event 4 5 31 1990 3 2 77E 02 66 1 Sample Event 5 9 13 1990 3 4 45E 02 84 4 Sample Event 6 4 3 1991 3 5 06E 02 80 8 Sample Event 7 7 10 1991 3 4 57E 02 78 0 Sample Event 8 10 3 1991 3 3 83E 02 88 8 Sample Event 9 5 2 1992 3 3 53E 02 77 9 Sample Event 10 1 11 1994 3 4 33E 02 93 5 Sample Event 11 5 28 1996 3 6 76E 02 96 4 Sample Event 12 6 27 1997 3 4 44E 02 87 8 Sample Event 13 12 10 1997 3 4 10E 02 92 4 Sample Event 14 6 19 1998 3 2 83E 02 81 7 Sample Event 15 12 19 1998 3 7 29E 03 82 7 Note when the number of plume centerline wells is less than 3 no analysis is performed and all related values are set to ZERO Confidence in Coefficient is the statistical confidence that the estimated coefficient is different from ZERO for details please refer to Conference in Trend in Linear Regr
467. xcellence AFCEE MONITORING AND REMEDIATION OPTIMIZATION SYSTEM SOFTWARE Sampling Optimization D etailed Approach ee The Sampling Optimization M enu screen Sampling Optimization Menu accessed from the M ain M enu screen by dicking Sampling Optimization is the main menu for sampling optimization and data sufficiency analysis It allows 2 eni enn and Bermedbsalins Tighrmeratien System MIN The Sano Optincatan lait fr nen neni te vanos renta ane pres nog omg botri detarmiakien surgir tepar opttwzutien erc tee mitancy mayis Choose am wre et Noewitie manuel c cere sedi se ri rn o wap sacle iat the user to choose between performing opt t 2l PRAAN e Sampling Location Analysis Pa AE PE A MS e Sampling Frequency Analysis 5 be LAM EAR CREE wA e Data Sufficiency Analysis 5 Option 3 Data Suffictency Analysis 2 on T The functions accessed by each choice 2 gaua iwl are as follows e IT Sampling Location Analysis Determines sampling locations by the Delaunay method removing redundant sampling locations from the monitoring network and or add new sampling locations The theoretical basis of the Delaunay method is given in Appendix A 3 Sampling Frequency Analysis Determines the sampling interval for each sampling location by the Modified CES method The procedures used in the Modified CES method are given in Appendix A 9 D ata Sufficiency Analysis Evaluates the cleanup status and concentration tre

MAROS - phaas.net

Contents

Download Pdf Manuals

Related Search

Related Contents