Home

User's Guide to FHAT20 - Ministry of Forests, Lands & Natural

1. Enter a comment that will be add ta the edited records _ Figure 23 The edit results dialogue box is used to modify FHAT20 predictions based on a set of user defined rules 54 May 2000 FHAT20 User s Guide 10 0 Controlling FHAT20 by Manipulating Tables and Files Certain aspects of FHAT20 operation can be controlled by modification of tables in MODEL MDB as well as the legend file MODEL MDE 10 1 Adding Variables to Include in Modelling The DISTLOOKUP table controls what variables are available to use in various modelling procedures The FieldName and TableName fields denote the name of the variable used in the database and the table that it resides in The DisplayName field provides the descriptive string of the variable that appears in the various dialogue boxes used for modelling The isGroup and isText fields denote whether the variable is tied to a fish group e g FishGrp DIST and whether the variable is text or numeric respectively The remaining fields determine whether the variable will be available for use in various modelling procedures ForFishDist variables can be used in fish distribution modelling Section 6 0 ForMKDE variables can be used in fish habitat capability modelling Section 7 0 ForStrat variables can be used in stratification for physical modelling Section 3 2 ForEdit variables can be used for editing rules Section 9 0 EditableVars variables can be edited Section 9 0 10 2 Contr
2. Figure 11 The stratification dialogue box is used to review and create rules that stratify data used to develop physical prediction models Keep in mind that there will be a strata class for all unique combinations of each variable class that you define For example if you defined 3 stream order classes two gradient classes and two confinement classes there would be 3 2 2 12 unique strata These 12 strata will be used to develop separate predictive relationships for the physical modelling which has a sample size limited to the number of reaches that were sampled for physical data If 36 sites were sampled you would on average have only 3 sites per strata More likely you would have some strata with 5 10 sites and a number of strata with no or few sites When you click on the combo box displaying existing strata groups the number of sites in each strata class will be shown in the bar graph You must trade off possible increased precision obtained by more detailed stratification against the danger of fitting models with limited degrees of freedom e g a linear regression based on two data points is fairly meaningless even though its value will be 1 The model will default to using the unstratified predictive relationship for any reach that falls in a stratification class that has a relationship based on less than 3 data points sites or a larger value if you specify it in the Channel Morphology dialogue box The stratification class for e
3. 4 Edit feature data to define whether a feature is an obstruction to each fish group 5 Model the fish group s range within the watershed 6 Model the fish group s habitat capability which is combined with the predicted range to estimate probability of fish presence and 7 Model FPC stream classification based on the predicted probability of fish presence and predicted channel widths iv May 2000 FHAT20 User s Guide Acknowledgements This project was funded by contracts from BC Fisheries to Ecometric Research Inc David Tredger and Tony Cheong scientific authorities on the project provided significant input into the design requirements of FHAT20 We thank Geographic Data B C for providing digital TRIM maps and Stu Hawthorn at BC Fisheries for providing a digital TRIM watershed atlas for one of the test watersheds Thanks to Dr Carl Walters at University of British Columbia for providing the bayesian sampling importance resampling algorithm May 2000 V FHAT20 User s Guide Table of Contents DISCLAIMER A T u W l u Sq l cuni dC UR aee ae ua INN ss III ABSTRACT erecta ees cce revo Gu fecbevev ee deed a daa oe deuda ee asa aeo coUe Vu evo a IV ACKNOWLEDGEMENTS Y eee tee ceo uocare up teo l nu eaque tana SEO E een Su t S V 1 05 INTRODUCTION I a T ee u ee edes eere uka eu 1 2 0 INSTALLATION
4. Predicted Channel Width m if the reach is not sampled Average measured channel width if the reach has been sampled Res_Phys Chan_width Distance to Bottom of System km Distance of reach km to the most downstream reach in the project area Network Dist_To_Bottom Sinuosity Ratio of reach length to the straight line distance Reach_Cards Sinuosity between the bottom and top of the reach Predicted Wetted Width Same as above but for wetted width Res_Phys Wet_width Predicted Bankfull Depth Same as above but for bankfull depth Res_Phys Bank_Full_Depth Probability of No Visible Channel The predicted probability that the reach is classified as a non visible channel Res_Phys Novis_prob Observed Non Visible Channel Denotes whether a sampled reach was a non visible channel Phys_Site No_Channel_Vis Group Upstream of Obstructions Reaches upstream of obstructions for the fish group False fish group is not upstream of an obstruction specific to that group Network Group _Dsbarr Group Downstream of Occurrence All reaches downstream of an observed occurrence of the fish group Network Group Ds Fish Pres Group Dist U S of Fish Occurrence Distance upstream km from nearest fish occurrence Network Group Dist US Fish Pres Group Not Present Upstream All reaches upstream of sampled reach where fish group was not found as long
5. Alternatively if the first fluvial reach downstream of a lake is predicted to be outside of the fish group s range then the lake upstream is also assumed to be outside of this range If you want to model the range of fish in lakes using this approach you must check the box labelled Predict range in lakes based on predicted range value in nearest downstream fluvial reach If you do not check this box your range rule must include a component for lakes e g isLake TRUBE or be based on variables that are available for both fluvial reaches and lakes e g MaxDSGrade DSBarr etc but not channel width FHAT20 users can build a series of such range rules for any fish group view the resulting ranges on the map compare the predicted ranges with known distribution limits from the reconnaissance survey and other information sources e g FISS and modify the rules until the desired range is May 2000 37 FHAT20 User s Guide achieved If we knew via sampling the upstream range limits for a fish group throughout the watershed there would be no need to go through such a modelling process However due to limitations on time and resources this method is the only way to estimate or interpret the range in parts of the watershed that have not been sampled Note that by combining range rules with obstruction information as in the examples given above it is possible to perfectly replicate the range limits that are known while at th
6. Channel Width 2 88 213 0 25 7 51 1 67 Bank Full Depth 9999 00 3883 00 95999 00 9999 00 000 Reach Gradient x 92 5 02 0 12 13 54 3 96 r Fish Abundance Classes Abundance Probability of Reference Sampled Sites sites electrofished within 4 100 mz x modelled distribution limits of current species group ps Area of Sample Site 100 m2 0 20 r Process PDF for Single Reach Recompute Abundance Classes 120 907500 00000 00000 0000 0000 000 000 000 000 000 0009 GRADIENT_20 Absent 120 907500 04532 00000 0000 0000 000 000 000 000 000 074 120 907500 04532 57672 0000 0000 000 000 000 000 000 00 8 120 907500 04551 00000 0000 0000 000 000 000 000 000 0 120 907500 04553 00000 0000 0000 000 000 000 000 000 0 x B Prob 4 Reach_ D 2 r Process PDF for All Reaches 0 8 1 15 20 25 Abundance 100 m2 Process PDF s For All Reaches Cancel Run Y Max X Max 25 v Reflect Figure 17 The fish habitat capability dialogue box is used predict reach specific habitat capability and probability of fish presence 7 3 Output Indicators from Habitat Capability Modelling There are seven output indicators that are produced by FHAT20 for each fish group that is modelled All predictions are saved to the RES_FISH table by reach in the model database This section describes what each of these fields represents Results can be viewed spatially by loading the a
7. Reconnaissance 1 20 000 Fish and Fish Habitat Inventory User s Guide to the Fish and Fish Habitat Assessment Tool FHAT20 Prepared by BC Fisheries Information Services Branch for the Resources Inventory Committee May 2000 Version 1 0 The Province of British Columbia Published by the Resources Inventory Committee Canadian Cataloguing in Publication Data Main entry under title Reconnaissance 1 20 000 fish and fish habitat inventory computer file user s guide to the fish and fish habitat assessment tool FHAT20 Available on the Internet Issued also in printed format on demand Includes bibliographical references ISBN 0 7726 4305 9 1 Fish stock assessment British Columbia Data processing Handbooks manuals etc 2 Fishes Habitat British Columbia Data processing Handbooks manuals etc BC Fisheries Information Services Branch II Resources Inventory Committee Canada QL626 5 B7R424 2000 333 95 611 09711 00 960244 5 Additional Copies of this publication can be purchased from Government Publications Centre Phone 250 387 3309 or Toll free 1 800 663 6105 Fax 250 387 0388 www publications gov bc ca Digital Copies are available on the Internet at http www for gov bc ca ric FHAT20 User s Guide Disclaimer Approval by the Ministry of any deliverable created by this model means only that the deliverables were provided in accordance with standard
8. that were not sampled A strata consists of a subset of reaches from the entire dataset defined by a set of remote sensed characteristics One or more variables can be used to define strata For example stream order could be used to define two strata those reaches with stream order lt 2 and those with order 22 A more complicated stratification scheme or rule also termed a stratification group would be based on two or more variables for example stream order and gradient The previous stream order classes could be subdivided into 3 additional classes with gradients 0 2 2 5 and gt 5 for a total of 6 strata In the physical modelling separate functions predicting width and depth probability of non visible channels will be fit for each strata When saving model results FHAT20 cycles through each unsampled reach in the dataset determines its strata based on the variables you included in the stratification group and then applies the appropriate model to predict its physical characteristics To define stratification groups select the Define Stratification Groups choice from the Modelling main menu item Fig 11 Previously saved stratification schemes will be displayed in the dropdown box in the upper left corner of the dialogue box When you select a scheme from this list the strata groups will be displayed in the list box in the upper right hand corner To create a new stratification scheme first click on the Remove butto
9. the standard deviation of a normal distribution function and theory dictates that an optimum window width can be determined by a simple formula that considers the variance of the data set The discussion so far has been restricted to a univariate context where the objective of the estimation technique is to estimate the underlying pdf of a single variable The kernel estimation techniques described above can be extended to include multiple variables with very little modification The key difference is that each observation in the data set is viewed as a vector of variables rather than a single data point where x and Xi are vectors of dimension d corresponding to the number of variables The multivariate Gaussian kernel function is as follows 1 x X fees 6 K t exp 0 5 t t 7 X X rs Conditional probability functions can be obtained by considering only the dimension of interest when calculating the pdf in this case abundance For example in a two dimensional case e g Fig 16 where abundance and a single habitat variable form the vectors x and Xi Equation 6 can be used to construct a 3 D graph showing the probability z axis of a given abundance value y axis in relation to the habitat variable x axis A conditional pdf P abundancelhabitat value can be obtained by taking a slice through the 3 D relationship at the habitat value of interest e g Fig 16b This is similar to calculating a un
10. CENE Fish Absent 5 oaeo 022 1059 05 195 0064 081 0 005 18 42 01 saa os sepa o 53 1 5 5 S6 e3 000 2 gt 5 lt 20 DO 51020 ng Figure 9 The channel morphology dialogue box is used to make reach specific physical predictions such as channel and wetted widths 1 CONF CODE EN CO 2 CONF_CODE UN NA To fit power functions to the x y data for each strata click on the button labeled Run Models When you do this model fit statistics will be displayed in the table at the bottom of the dialogue box and the fitted line the model will be shown as a green set of triangles in the x y scatterplot Fit statistics include sample size N constant A and slope B parameters of the power function the unexplained mean square error MSE the correlation coefficient R the percent of the variance explained by the model and the probability that the slope of the power function is not significantly different from zero Prob Low MSE high R and low Prob values denote good model fits to the data You should evaluate these statistics across a range of stratification schemes to develop the most predictive models possible When examining how well the model fit a particular data set you may notice outliers to the model that is blue points that are noticeably more distant from the fitted green line representing the model These outliers may be normal sites that represent the extreme end of th
11. FDIS 2 Errors in watershed codes are generally associated with the process of importing the watershed codes in the ILP table back into FDIS In some instances in headwater forks when the mainstem and tributary gets assigned a code which is opposite to what was identified this results in both a missing watershed code and a duplicate watershed code and reach number Newer versions of FDIS will check for this problem 8 May 2000 FHAT20 User s Guide Correction of situation b errors cannot be done automatically If a match between a corrected record in REACH_CARDS and a record in a site table occurs the import procedure has no way of knowing whether the identifiers were originally entered correctly in the site table via FDIS and should therefore now be linked to the record in REACH_CARDS with the original identifiers or whether the site identifiers were entered incorrectly from FDIS and should therefore be updated based on the corrected values specified in WSCODE_LOOKUP Thus when situation b occurs during the import procedure the TABLE_TYPE field in WSCODE_LOOKUP is set to the name s of the site table s where a record corresponding to the modified REACH_CARDS record exists It is up to the user to then manually check these cases using original maps UTMs etc to make sure that the records do correspond If they do not additional records must be added to WSCODE_LOOKUP by the user so that the records in the MODEL MDB site
12. File main menu item The EXPORT table contains a number of identifier fields to facilitate linkage back to FDIS and other BC Fisheries Inventory applications including e FDIS watershed code and reach_id e MODEL MDB watershed code and reach id equivalent to FDIS values unless the values were corrected in WSCODE_LOOKUP e NID NID_MAP May 2000 21 FHAT20 User s Guide e MAP e Easting and Northing UTM When you export the model results a table called CUREXPORTLOG is updated This table contains the rules SQL statements results and time stamps associated with the various modelling operations you completed This allows a third party to verify that model results in the EXPORT table were based on modelling steps completed in the correct sequence Maps displayed in FHAT20 can be exported as bitmap files Select the Dump Map Legend to Bitmap option form the File main menu item You will be prompted for a filename to save the map to and a separate filename to save the legend to You can import both of these files into another application and because the map and legend images are saved in different files you have flexibility in terms of where the legend is located on the final graphic You can also print the map and legend to the default printer by selecting the Print Map Legend option from the File main menu item 22 May 2000 FHAT20 User s Guide 3 0 Physical Predictions FHAT20 predicts channel
13. Further the resulting pdf may not reflect reality The fact that no fish are caught at a site does not eliminate the possibility that there may indeed be fish in the reach at a location or time other than the sample site and date and reflects a limitation in the sampling procedure If one were absolutely 60 May 2000 FHAT20 User s Guide certain that fish were absent from a reach then this approach may be appropriate But this certainty is not possible given the sampling procedure A more conservative approach that accounts for this sampling bias is to reflect the probability values associated with negative abundance values onto the positive line and then ignore all negative abundance values This approach is the preferred option suggested by the Silverman 1982 and is the default option adopted in the present algorithm In the Fish Capability dialogue box you have the choice to reflect ve values by checking the box labelled Reflect The main drawback of this approach however is that the probability of fish absence may be underestimated if in sampling we are indeed certain that fish are absent from a given site It will be useful to empirically evaluate both approaches by comparing their respective predictive capabilities Once the data are reflected the grid points are back transformed and the associated probabilities are integrated summed across each of the abundance categories as noted above The result is a probabilit
14. IS USED TO EDIT THE MAP DISPLAY OF DATA AND PREDICTED VARIABLBS puya h uo uya qa as 18 SCHEMATIC SHOWING THE RELATIONSHIP AMONG MODELLING STEPS IN FHAT 20 20 THE OPERATIONAL TRACKING DIALOGUE BOX DISPLAYS THE STATUS OF MODELLING PROCEDURES AND THE DATE TIME THE PROCEDURES WERE RUN cccccesssesececececccceccccccccccccccececescecceecseseseceseseseesessesesess 21 THE CHANNEL MORPHOLOGY DIALOGUE BOX IS USED TO MAKE REACH SPECIFIC PHYSICAL PREDICTIONS SUCH AS CHANNEL AND WETTED WIDTHS ccccccececececcccccccccceccccccecccccccccseccscsesseesesesecess 24 THE OUTLIER DIALOGUE BOX IS USED TO DISPLAY SITE SPECIFIC PHYSICAL DATA USED IN CHANNEL MORPHOLOGY MODELLING ALLOWING USERS TO EXCLUDE SPECIFIC DATA FROM THE MODELLING PROCEDURES reud e e eee reve ra t ouv buc e yk eye eee b Ode 25 THE STRATIFICATION DIALOGUE BOX IS USED TO REVIEW AND CREATE RULES THAT STRATIFY DATA USED TO DEVELOP PHYSICAL PREDICTION MODELS ccccccccecesececceccccccccecccececccececccesesesessescesceseesesesesess 28 THE NON VISIBLE CHANNEL DIALOGUE BOX IS USED TO PREDICT THE PROBABILITY THAT UNSAMPLED REACHES WILL BE NOT BE VISIBLE CHANNELS ccccccccccececccccccecceccecececccccsceseceesesesseseseeess 29 THE FISH GROUP DIALOGUE BOX IS USED TO REVIEW AND CREATE NEW FISH GROUPS USED IN FHAT20 MODELLING 32 THE
15. User s Guide When you decide on the best stratification scheme click on the button labeled Save Results to RES PHYS Table to apply the model to the unsampled reaches and to save the results to the RES PHYS table in MODEL MDB 3 4 Adding Physical Site Data to the Model Database Additional physical data not contained in FDIS can easily be included in the FHAT20 model database to improve the precision and accuracy of predicted channel and wetted widths bankfull depths and the probability of non visible channels Open the USER PHYS SITE table in Access and enter the watershed code and reach ids for the new sites as well as identifiers for the Site ID The latter field can be any numeric value make one up if a Site ID doesn t exist and make sure to enter unique values for different sites in the same reach Enter any of the following information currently used in FHAT20 for physical modelling CW avg average channel width in m WW avg average wetted width in m BFD avg average bankfull depth in m No Channel Vis is the channel visible Yes No You can enter values for other data fields in USER PHYS SITE for completeness but these data will not be used in any calculations in the current version of FHAT20 Since you presumably want to use these new data in the physical modelling make sure the UseForCW UseForWW and UseForBFD fields are checked The only circumstance when one of these fields should not be checked is if there are
16. applied to all reaches in the watershed only remote sensed characteristics from the REACH_CARDS table in FDIS and predictions made for each reach in the watershed excluding lakes e g channel width can be used as independent variables Note that range rules do not automatically produce a continuous fish range within a watershed Depending on what variables you use in the model there can be reaches downstream of the upstream limit that are not included in the range For example the following rule FPC_FISH_DSBARR FALSE AND GRADIENT 20 15 states that the FPC fish group must not have a barrier downstream specific to this group and cannot be present in reaches with gradients 215 If there is a reach of gradient 215 downstream of a reach with gradient 1546 there will be a hole in the fish range when viewing the results on the map fish will be distributed upstream and downstream of a reach which is not included in the range In the final computations of stream classification probability of presence is adjusted so that a continuous fish range is used to classify the stream Section 8 0 Fish range in reaches classified as lakes in FDIS is predicted based on the predicted range value for the closest fluvial reach downstream of the lake If fish are determined present in the first reach downstream of the lake then the fish range is assumed to extend upstream into the lake as long as the lake is within the modelled range limits
17. automatically overwrites any model results that meet the conditions of your rule The number of records that are modified based on your rule will be reported To save the rule to the model database type a name for the rule in the combo box at the top of the dialogue box labeled Existing Edit Rules and click on the Save Edit Rule button To delete an existing rule click on the Delete Edit Rule button May 2000 53 FHAT20 User s Guide ig Edit Modelling Results B X Existing Edit Rules for type new name ta create a new rule ProbPresFis Run Edit Rule Save Edit Rule Delete Edit Rule Fish Group FISH Select variable ta edit Unique values for Select variables to use in edit rule selected variable Prob of Fish Presence Parent Reach 075 of Fish Occurence Probability of Visible Channel Downstream Elevation Wetted Width False Magnitude Channel Width Wetland Most Probable Stream Class Parent Order Fish Distribution Reach Id Prob of Fish Presence _ Fish Distribution Bank Full Depth ls there a Lake Upstream Continuous Stream Class Does Reach Flow into a Lake Distance to Bottom of System Watershed Code Parent Reach Dz5 of Fish Occurrer Capability of Parent Reach Probability of WC Em Set value af Frob of Fish Presence i 00 r Rule Operators _ when following criteria area met Elea Enter Text Metwork FPC FISH PARENT DS FISH PRES True
18. characteristics and estimates of relative abundance from the sampled reaches to estimate capability in unsampled reaches Section 7 0 Fish range and capability predictions for Forest Practices Code fish are combined with channel width predictions to estimate the Forest Practices Code FPC stream classification S1 S6 for each reach 2 May 2000 FHAT20 User s Guide 2 0 Installation and General Operating Guidelines The FHAT20 computer program consists of 1 a relational database management system to retrieve data from FDIS and save modelling extrapolation results 2 a series of dialogue boxes and graphics to develop and examine various models predicting physical habitat characteristics fish range and habitat capability and 3 a series of computer algorithms that perform statistical and other extrapolation operations used in the modelling FHAT20 is a 32 bit application that will operate under Windows95 Windows98 or Windows NT and must be used in conjunction with FDIS version 6 5 or higher 2 1 Installation FHAT20 can be downloaded from the INTERNET via the BC Fisheries FTP site To obtain the installation program go to the FTP site FSHFTP ENV GOV BC CA You can do this through internet browser by using FTP FSHFTP ENV GOV BC CA pub outgoing FHAT20 as the address to get into the FHAT directory Click on FHAT20 ver1 0 May 11 2000 ZIP and you can then download it to your hard disk e g C TEMP Unzip the files into the tem
19. click on the button labelled Recompute Abundance Classes When you select a reach to process a pdf for the estimated sample site size will be shown in the text box labelled Area of Sample Site immediately above this button This site area is the product of the estimated wetted width for the reach and 10 times the estimated channel width or 100 m in length whichever is greater Note that the density limit defining the Not Present abundance class is simply 1 fish Site Area 100 m As sample site area increases the Not Present density limit decreases and so does the probability of fish absence Once you have examined the pdfs for a representative set of reaches and are comfortable with the model results you are ready to process all reaches in the data set and save these results to the RES FISH table in the model database To do this click on the button labeled Process May 2000 FHAT20 User s Guide PDFs for All Reaches in the lower right hand corner of the dialogue box This may take a few minutes depending on the number of reaches to process ig Multivariate Kernel Estimation of Probability Density Function PDF for Fish Abundance Pick Species Group ALL FISH of Sampled Sites zb Variable In Model Reach Val amp va _ Min StDev Onder 5 00 2 03 1 00 3 00 0 81 Upstream Elevation 9999 00 1691 28 9999 00 1809 00 5129 57 Abundance Wetted width 0 20 1 0 00 6 02 1 28 100 m2
20. in the dialogue box and double clicking on the variable Click on the watershed code in the table to highlight the stream reach on the map the selected reach will be displayed as a thick white line Physical Characteristics Order Magnitude Gradient Downstream Elevlatian Continement Code Channel Pattern Barriers Add Obstruction ta Selected Reach Select Record in Obstruction Grid Reach is Lake Reach is Lake Headed Watershed Code Reach Id Order Continement Code 120 907500 52900 00000 Figure 5 The search results dialogue box displays data and results for reaches selected on the map Zooming In on a Selected Area on the Map The zoom feature allows you to view a smaller area within the full study watershed shown on the map Simply click on the check box labeled zoom click on the map with the left mouse button and hold the button down while dragging the mouse over the portion of the map that you want to select When you release the mouse button the map will be redrawn showing only the selected area You can repeat this process to select an even smaller area to view or restore the map to its full size by clicking on the button labeled Draw Full Map You can view the reach breaks by clicking on the check box labeled Show Reach Breaks Overlay Second Layer Looking at Two Map Variables at One Time The overlay feature is very useful for comparing predicted results with observe
21. it active for the proper groups O Not Active 1 Active 4 Press the Process Obstructions button Available Watershed Codes 7120 907500 00000 00000 0000 0000 000 000 000 000 000 000 120 907500 04532 00000 0000 0000 000 000 000 000 000 000 120 907500 04532 57672 0000 0000 000 000 000 000 000 000 120 907500 04551 00000 0000 0000 000 000 000 000 000 000 120 907500 04559 00000 0000 0000 000 000 000 000 000 000 120 907500 04566 00000 0000 0000 000 000 000 000 000 000 Delete the Selected Obstructions Current Obstructions in DataBase ReclD 5 CODE REACH ID MEASURE TYPE HEIGHT LENGTH ACTIV 0000000 120 307500 0t 23 146 0000000 120 907500 5 0000000 120 907500 8 00000001 120 907500 8 0000000 120 907500 8 00000001 120 307 500 7t 0000000 120 307500 7t Oo000000 120 307500 5 0000000 120 907500 4 0000000 120 307 500 4 DD00000 _ 120 307 500 27 DD0D000 _ 120 907500 2 DD0D000 120 3077 500 27 000000 1 20 307 500 27 iini aot 1 20 307 500 27 OOOOC 120 907500 2 000001 120 907500 2 nnnnnnnr amzEnn a E 1 1 O FDIS u s vi O FDIS d s D FDIS d s C 0 n n 0 5 Road S d s Cs 5 d s Ci 5 d s 5 DV ne 5 d s O 5 u s 5 d s s 22 g 2 9 2 2 95 9 2 mmm 35 S S S S S S S S S S S 1 1 I D D S I ooo I o
22. m capability classes classes can be redefined by the user We essentially draw 42 May 2000 FHAT20 User s Guide such limits on top of the pdf e g as vertical lines like in Fig 16b and then compute their probabilities as the integral of the area under the pdf curve within each class A pdf can also be used to compute the probability of fish absence but this requires that a not present density limit be defined Intuitively one would expect this density to be 0 fish m i e if the density is greater than this value the site should be considered Present in status This would indeed be the case if the entire area of the reach were sampled In practice sample site size is less than the entire reach area and in the case of a reconnaissance level inventory sampled area is the product of the wetted width and 10 times the channel width or 100 m in length whichever is greater The minimum number of fish that can be caught within the site area to negate an Not Present class status is obviously one you can t physically catch less than one fish thus the not present density class limit must be 1 fish site area The not present density limit therefore decreases proportionally with the size of the site Given two reaches with equal densities one in a narrow channel small sample site and one in a wider channel larger sample site the not present density limit 1 site area will be higher in the small site compare
23. minimum you must model the range and capability of at least one fish group in order to predict FPC stream classes When you create a new fish grouping you need to define which obstructions are barriers to migration and model its range and capability To define a new fish group or examine what an existing fish group consists of open the Fish Groups dialogue box accessed from the Fish Groups option below the Modelling main menu item Fig 13 If you click on any fish group in the dropdown list box labelled Current Fish Groups a text box at the bottom of the dialogue box will show the rule a SQL statement that defines what species and life stages are included in the group e If you want to delete an existing fish group select the group from the list box and click on the button labelled Delete Selected Group e If you want to create a new fish group type in the new fish group name in the dropdown list box then Select the ages or life stages to be included in the fish group To add a particular age life stage or species select an entry from the one of the lists on the left side of the dialogue box and then click on the Add button and they will appear in the corresponding list box on the right side of the dialogue box If you want to remove one of the entries an age life stage or species select the entry from one of the right side list boxes and then click on the button labelled Remove When
24. missing values for some of the measures 1 a channel and wetted width were measured but no estimate of bankfull depth was taken You want to confirm that the KeepOnRebuild field is checked default condition This flag will ensure that the new data you enter will not be lost if you rebuild the database at a later date Ensure that the FDIS REC field is not checked default condition When user defined data is imported to the PHYS SITE table this field distinguishes FDIS data from user entered data After entering the data close the USER PHYS SITE table and select the Process Physical Site Data sub menu choice below the Utilities main menu item This will initiate a procedure to ensure that there are corresponding records in the Reach Cards table in MODEL MDB for each record in USER PHYS SITE If a user defined record has been successfully transferred to the PHYS SITE table the ValidSite field in USER PHYS SITE will be checked If this field is not checked after you have Processed the Physical Site Data then the watershed code and reach id for the site in question does not exist in the MODEL MDB version of Reach Cards and ValidSite will not be checked Edit the watershed code and reach 1d in USER PHYS SITE so that it corresponds to a record in Reach Cards and rerun the Process Physical Site Data procedure Note you will need to rerun the physical modelling procedures for these new data to have an effect on physical pr
25. model Section 7 3 Figs 18 19 52 May 2000 FHAT20 User s Guide 9 0 Editing Model Results You may want to modify model predictions after closer inspection of the results In some cases this can be accomplished by simply re running model rules for physical predictions fish range or fish habitat capability using different model variables and or conditions However there may be cases where you are comfortable with the overall model s and only want to modify specific records where model predictions seem inappropriate This can be accomplished by opening MODEL MDB in Access and manually changing the values of specific records Such manual editing may be adequate when only a few records need be modified and when the database is relatively small In other cases an automated editing procedure based on a series of user defined editing rules is required For example you may develop a fish range rule that limits fish to reaches with maximum downstream gradients lt 20 and to reaches with predicted channel widths gt 3 m However from field sampling or experience in other systems you may want your results to reflect the fact that many species can utilize the lowest reaches of small steep tributaries to larger mainstems where fish are present i e confluence areas You could edit the results using a rule that predicts potential fish presence if the reach is less than 3 m but is a tributary to a larger mainstem reach where fish were
26. predicted to be present 9 1 Automated Editing and the Edit Rule Automated editing of model results can be accomplished via the Edit Modelling Results dialogue box accessed from the Edit Results option below the Utilities main menu item Fig 23 To edit a variable select the variable from the list box on the left side of the dialogue box If you want to edit a variable that is fish group specific you also need to select the appropriate fish group from the dropdown combo box where the fish groups are listed When you click on the variable you want to edit unique values will appear in the list box immediately to the right You then need to select the independent variables you want to use in your editing rule Select these from the right most list box When you single click on a variable in this list box its unique values will be shown in the middle list box To include the variable in the editing rule double click on it and it will appear in the text box at the bottom of the dialogue box Set the appropriate conditions for that variable e g Parent Order gt 2 by typing the condition in the text box or by clicking on the math operators and double clicking on the unique variable values in the middle list box Finally set the new value that you want to change the predictions to in the small text box above the large text box where you are creating the rule To run the edit rule click on the Run Edit Rule button Running the rule
27. tables are updated These new records should contain the original watershed code reach id and RecID identifiers the FDIS_ values just copy their values and the new identifiers MODEL values that are used to link it to the correct REACH CARDS record In addition the ERROR TYPE field should be set to the appropriate code e g 4 if you are updating a record in the PHYS SITE table in MODEL MDB Correction of stream network errors detected by FHAT20 may require a significant amount of effort and depends completely on the consistency in the FDIS database The correction procedure can be streamlined by realizing the hierarchical nature of some of the errors You may observe a large number of records with incomplete lineage errors type 3 all resulting due to a small number of orphans error type 2 If you correct type orphan errors first all incomplete lineage errors will automatically be corrected by the import procedure The most efficient way to correct the data is to e Correct all orphan type errors type 2 and known problems with watershed codes type 0 first e then correct duplicate reach id errors type 1 which may involve changing watershed codes i e the reach id may be correct and the error resulted from an error in the watershed code e correct error types 4 7 and finally e check and make sure that records in PHYS SITE FISH SITE USER BARR and LAKES are correctly assigned to the new water
28. these 1000 models to predict its channel width A frequency distribution of predicted channel widths generated by weighting each predicted width by the May 2000 25 FHAT20 User s Guide likelihood of each of the 1000 models that was used to generate it is computed for each reach This distribution is then used to estimate the probability of each reach being in particular width classes e g 1 5 5 m as explained below To implement the bayesian algorithm click on the check box labeled Compute Uncertainty and then click on the button labeled Run Models Fig 9 When the computations have finished an example frequency histogram of widths will appear in the graphic adjacent to the x y plot This shows the uncertainty in channel widths for a theoretical reach with a known total upstream length You can adjust the parameters effecting the display of the frequency distribution e g the total upstream length maximum of x axis bin size in the frame labeled Visualize Bayesian Estimates of Uncertainty in Width Predictions If you change any of the parameters you must click on the button labeled Plot Test Reach to update the frequency distribution Note that there will be different distributions for each model strata that was fit The distributions will be relatively narrow low uncertainty for strata that have precise models not much scatter but will be relatively wide high uncertainty for imprecise models Below the frequen
29. to 100 and all other probabilities are set to zero Uncertainty in stream class predictions for each reach is summarized in the Uncertainty in S Class variable The variable compares differences in the probability of FPC stream classes S1 6 within a reach If there are fairly even probabilities across some or all of the classes the uncertainty is high The most uncertain situation is where all stream classes have the same probability 16 67 since there are 6 classes and the probabilities must sum to 100 in this case the uncertainty in S Class variable will equal 100 If one stream class contains a large probability and others therefore have low probabilities then the uncertainty is low The most certain situation is where the probability for one stream class is 100 and therefore 0 for the other classes in this case the uncertainty value will be 0 The formula for computing uncertainty is n n 2 100 i l i l 1 S S i Al where n 6 6 FPC stream classes The deterministic estimate of stream class follows the FPC guideline that reaches cannot be declared non fish bearing in status if they are downstream of fish bearing ones To reproduce this in the model the stream classification procedure generates a Continuous Probability of Presence variable where PoP values remain constant or increase as one moves in a downstream direction Fig 22 A user defined minimum probability of presence value is then o
30. were estimating a pdf for a low gradient reach fish abundance data collected from reaches of similar characteristics should be given greater weight when estimating the pdf Visualizing this graphically the multivariate kernel estimation essentially constructs a surface predicting probability as a function of fish abundance and the habitat variable Fig 16a based on all fish site abundance estimates the reference data set and then takes a vertical slice through this surface at the location of the reaches habitat value Fig 16b This approach can be extended using multiple habitat variables but cannot be visualized graphically since we are limited to viewing results in three dimensions See section 7 4 for computational and theoretical details of multivariate kernel density estimation May 2000 41 FHAT20 User s Guide b Low Medium High Gradient lt 1 deg Jed uoniodoJg Count of observations 0 T 0 0 200 400 600 Total Salmonids 100 m2 5 Gradient 5 7 allg 0 3 0 2 0 1 0 0 0 0 200 400 600 Count deg sed uoniodoJd Total Salmonids 100 m2 Abundance low high Figure 16 Relationship between habitat and fish abundance Graph a shows the bivariate distribution of total salmonids and stream gradient based on data from 77 coastal streams in British Columbia data from R Ptolemy BC Fisheries Graph b shows probability density functions of total salmonid fit
31. 500 120 307500 120 307500 120 907500 120 907500 120 907500 120 307500 120 907500 120 907500 120 307500 120 307500 120 907500 120 907500 120 307500 Poh lt lt fee a Figure 10 The outlier dialogue box is used to display site specific physical data used in channel morphology modelling allowing users to exclude specific data from the modelling procedures Channel width is used in conjunction with fish presence to determine the FPC stream classification S1 S6 Even with stratification the empirical power functions you have fitted to the data will no doubt show substantial scatter This means that there can be substantial uncertainty in the channel width predictions and it would be unwise to only use the most likely best fit width prediction in the decision making process FHAT20 uses a bayesian algorithm Walters and Ludwig 1994 McAllister et al 1994 to estimate the uncertainty in channel width for each unsampled reach When you compute this uncertainty the bayesian procedure essentially fits 1000 different power functions different parameter values for A and B to each data set each strata The relative likelihood of the data given each of these models combinations of the A and B parameters is computed and stored in memory Once these 1000 models and their likelihoods have been computed the procedure works its way through each unsampled reach and enters the total upstream length into each of
32. AND GENERAL OPERATING GUIDELINES 3 2 b INSPAEEATIONE ay t DL ED exe EN eir MR irn esr 3 2 2 FIRST TIME USE OF FHAT20 IMPORTING A FDIS DATABASE INTO FHAT 20 3 2 2 1 Error Checking Procedure esse eve et dte epe nearest 5 2 2 2 Error Correction visccccccccccssssecccccccccusssccccccccsusussecccceceuuuscceccscecauuuucceccscesuuuscecccsceuuaasecececesauaas 6 2 2 5 FHAT20 Maps Stick Diagrams or the Digital TRIM Atlas eese 11 2 3 CONTROLLING THE MAP DISPLAY ccccccccccceceeceeeeeceeeeeseeesesececeeeeeeeeeeeeeeeeeseeeseeeeeeueeueeeeueesaeaaaeaas 12 2 4 GENERAL MODELLING PROCEDURE eee I e I I eene nnne nn nnn n nnns nsn nsns asa asas asas asas anna ananas 19 2 5 EXPORTING RESULTS SAVING MAPS PRINTING MAPS cccccsssssceeccceesessseeececeecesssseeeeeeeeeeesnes 21 30 PHYSICAL PREDICTIONS u a aie e ERES e anao ae eo eia aa eU ERR Se 23 3 1 CHANNEL WIDTH WETTED WIDTH AND BANKFULL DEPTH PREDICTIONS 23 3 2 STRATIFICATION OF PHYSICAL PREDICTIONS a nnnn nnn nnn nnn n nana nana nnns nn 27 3 3 NON VISIBLE CHANNEL PREDICTIONS eee e ene nn enne nnn nnn n nnn n nnn nnns nnns nasa nasa nn 29 3 4 ADDING PHYSICAL SITE DATA TO THE MODEL DATABASE eee I ne en nn nnne n n
33. FHAT20 User s Guide You can view the error report at any time by selecting the Error Reporting option from the Utilities main menu item Fig 3 Error Report Imported Records Data Validation Rules Se SUC SLE Receh ziet Bs 7 Number of records that failed data validation rules but 2716 Number of Reaches imported into Model Reach Cards B80 included in modeling database Number of Lake Reaches imported into Model Lakes b Mumber of records deleted becaues they failed a data and Model Reach_Cards validation rule Records in WSCODE_LOOKUP that Require Fixing Fixed False Fixed True Total Records in WSCODE_LOOKUP Dupicate Reach ID Duplicate Reach ID Lake No Parent Match FH S SITE to Reach Cards No Match FISH SITE ta Reach Cards No Match Lakes to Reach Cards No Match LISER BARR to Reach Cards Figure 3 The error reporting dialogue box displays a record of errors and corrections made to the FHAT20 model database Correcting errors in watershed codes and reach ids for each record in the WSCODE LOOKUP table in MODEL MDB requires an understanding of the types of errors that can be trapped and handled by the importing procedure These errors are distinguished based on text in the COMMENTS field of WSCODE LOOKUP Table 2 It is important to note that in MODEL MDB the watershed code and reach id fields link records in REACH CARDS with records in the other tables contai
34. Group Prob High Group Probability of Presence The probability that reach is utilised by the fish group Res Fish Group Prob Pres Table 3 Con t Group Name Variable Name Description Table Field Stream Classification Most Probable Stream Class The most probable stream classification based on the combined probability of different stream widths and discontinuous fish presence a reach may have a lower probability of fish presence compared to an upstream reach Res_Fish Most_Prob_S Uncertainty in S Class A relative index of uncertainty associated with the Most Probable Stream Class A value of 0 denotes the minimum uncertainty 100 probability for one of the S Classes A value of 100 denotes maximum uncertainty all S Classes have an equal probability 16 67 because there are 6 classes Res Fish S_Uncert Sx Probability x 1 6 The probability that the reach is an Sx stream class e g S1 based on combined probabilities Res Fish Sx FPC Stream Class The stream class based on the most likely width Res Fish ConSClass with fish presence determined from FPC Fish Present FPC Fish Present Presence absence determined from continuous Res Fish ConFishPres probability of presence values combined with a minimum probability of presence limit defined by the user Continuous Probability of Presence Probability of fish pr
35. If two slices are taken through the bivariate distribution shown in Figure 16a this pattern can be seen more clearly Fig 16b Note how the proportion of sites in the medium and high density classes denoted by the vertical dashed lines declines as reach gradient increases from lt 1 to 5 7 40 May 2000 FHAT20 User s Guide The functional forms of habitat abundance data are certainly not simple To paraphrase Rice 1993 Although ecological theories can yield predictions of how animals should use habitats theory predicts in only general ways the shape of specific abundance habitat functions When abundance and habitat data are plotted the relationships commonly show combinations of thresholds floor and ceiling effects asymmetric ascending and descending limbs marked skewness or kurtosis differing variability in abundance at different positions along a habitat gradient and other diverse statistical problems Curvilinear models may fit the data better than linear models but they do not necessarily fit the data well To overcome these problems Rice 1993 suggested that a non parametric density approach be used to predict abundance from habitat data In particular Rice 1993 advocates the use of kernel density estimation to predict or forecast probability distribution functions pdf of fish abundance for a given set of habitat attributes A probability density function is simply a frequency histo gram showing the probabilit
36. MDB This field is set to TRUE for all reaches upstream of any obstructions for this fish group i e the obstruction is downstream of the reach in question and FALSE for any reaches downstream of the obstructions This new field can be viewed on the map by selecting a variable called Group Downstream of Obstructions from the Physical Variable Grouping As a safeguard you should always view a map of the obstruction results after you have edited and re processed the obstruction table It is very important that you include the Upstream of Obstruction variable when you build a fish range rule Section 6 0 or the modelled distribution limits will likely be too widely distributed in the watershed In some cases imprecision in the watershed codes may result in a tributary appearing to be upstream of a barrier in terms of how the model computes accessibility when viewing the Upstream of Obstructions theme even though on the hardcopy map the tributary is actually downstream of the barrier and should therefore be accessible If you are confident in the UTM value which determines the confluence location on the map for the stream in question or know from fish sampling that the confluence is downstream of the barrier then the barrier should be moved upstream by increasing its measure until it exceeds the confluence distance of the tributary s in question This can easily be accomplished in a few iterations Increase the value of th
37. N Y Walters C J and D Ludwig 1994 Calculation of bayes posterior probability distributions for key population parameters Can J Fish Aquat Sci 51 713 722 May 2000 57 FHAT20 User s Guide Appendix I Theoretical Background of Multivariate Kernel Estimation Kernel density estimation is analogous to constructing a frequency histogram However rather than assign counts to a set of predefined bins a count is assigned to each observation in a data set that falls within a specified range analogous to a bin defined by a window width x h x h and centred about the observation f x luo of X X falling in x h x h 1 n The scalar 2hn is used to transform the counts to estimates of probability such that the sum of all weights equals to one Silverman 1982 refers to Equation 1 as a naive estimator and is the simplest form of kernel density estimation Kernel density estimation can be more formally defined as 1 lt X oo bx 2 i l where K represents the functional form of the kernel In the case of the naive estimator the kernel function K is a simply a box defined as K t 05 if t lt 1 i 3 where x X x J 4 and x the observation of interest Xi all other observations in the data set The naive kernel estimator results in a pdf that is a discontinuous step function that does not meet the differentiability criteria of a true pdf A much sm
38. OBSTRUCTION EDITOR DIALOGUE BOX IS USED TO DISPLAY EDIT AND ADD MIGRATION BARRIERS TO THE FHAT20 MODEL DATABASE cccccsccecccccecccecccccccccccssecessceseceseeseeseseessess 34 THE FISH RANGE DIALOGUE BOX IS USED TO REVIEW AND EDIT RULES THAT PREDICT FISH GROUP RANGE IN THE MODELLED WATERSHED ccccccccsseseecsccceeesseccccccesuueeececcseesueesecccsscesuuaseceseeeeuuanseeceseees 39 RELATIONSHIP BETWEEN HABITAT AND FISH ABUNDANCE ccccccececececececccccececcccccceecececceccsesecsesesseseeess 42 THE FISH HABITAT CAPABILITY DIALOGUE BOX IS USED TO PREDICT REACH SPECIFIC HABITAT CAPABILITY AND PROBABILITY OF FISH PRESENCE cccccccccccccccccccccceccccccccccecccccceecesseseecsessesseessessesesess 45 SCHEMATIC SHOWING HOW THE PROBABILITY OF FISH PRESENCE POP IS COMPUTED FOR EACH REACH AS A FUNCTION OF MODELLED UPSTREAM DISTRIBUTION LIMITS PREDICTED BY FISH DISTRIBUTION RULES AND OBSERVED FISH OCCURRENCES ccccccececccececccccceccecccesccececcseseseessesesess 47 SCHEMATIC SHOWING HOW THE PROBABILITY OF FISH PRESENCE POP AND FISH DISTRIBUTION IS COMPUTED FOR STREAM AND LAKE REACHES IN RELATION TO FISH BEARING LAKES 48 THE STREAM CLASSIFICATION DIALOGUE BOX IS USED TO CLASSIFY REACHES INTO S1 S6 FPC STREAM CLASSES Em SCHEMATIC SHOWING HOW STREAM CLASSIFICATION PREDICTIONS ARE COMPUTED IN FHA T20 SCHEMATIC SHOWING HOW CONTINUOUS PROBABILITY OF FISH PRESENCEIS COMPUTED FROM THE PROBABILITY OF FISH
39. PRESENCE VARIABLE CALCULATED IN FISH HABITAT CAPABILITY MODEL 52 THE EDIT RESULTS DIALOGUE BOX IS USED TO MODIFY FHAT20 PREDICTIONS BASED ON A SET OF USER DEFINED desea ccvcbasd huyta us 54 May 2000 vii FHAT20 User s Guide List of Tables Ne viii DESCRIPTION OF SOME OF THE FIELDS IN THE WSCODE LOOKUP 6 DESCRIPTION OF ERROR CODES REPORTED IN WSCODE LOOKUP ORGANISATION OF VARIABLES DATA AND MODEL RESULTS THAT CAN BE DISPLAYED IN THE FHAT20 MAP INTERFACE cccccccsesesecececececececececececececececesesesesesess May 2000 FHAT20 User s Guide 1 0 Introduction The Reconnaissance Fish and Fish Habitat Inventory is a sampled based survey covering whole watersheds 1 all lakes stream reaches and connected wetlands within the watershed as defined from 1 20 000 scale maps and air photos This inventory is intended to provide information regarding fish distribution and relative abundance as well as stream reach and lake biophysical data for interpretation of habitat sensitivity and capability for fish production Anon 1998a While the reconnaissance inventory is intended to cover whole watersheds time money and personnel are not available to survey every stream reach and lake in the watershed therefore only a subset of reaches and lakes in the watershed is sampled However forestry planning processes require the devel
40. ach unsampled reach for the last variable you saved in the physical predictions form e g channel width is saved with the predictions in the WIDTH STRAT A field in the RES PHYS table 28 May 2000 FHAT20 User s Guide 3 3 Non Visible Channel Predictions Site assessment of some sampled reaches may reveal that the channel is not visible This could signify that 1 the mapping was incorrect and there was no channel where the map identified one or 2 the channel may flow subsurface In some watersheds a substantial number of reaches could be non visible channels FHAT20 predicts the probability of each unsampled reach being a non visible channel by stratifying the sampled reaches into different subsets based on remote sensed attributes that are available for all reaches Section 3 2 and computing the percentage of sampled reaches in each strata that are non visible These probabilities are then applied to unsampled reaches in the same strata The Non Visible Channel dialogue box accessed via the No Visible Channel option from the Modelling main menu item is very straightforward to use Fig 12 A list box at the top of the dialogue box shows different stratification groups When you select one the total number of sampled reaches in each strata are displayed along with the percentage of sampled reaches in this strata which were non visible channels You should examine these probabilities across different stratificatio
41. ady be displayed is best accomplished through the FHAT20 Legend Editor Section 2 3 However if you want to display an additional variable from MODEL MDB you must make a new entry in MODEL MDE for this variable The easiest way to do this is to copy an entry for a similar type of variable i e string boolean numeric from MODEL MDE to the bottom of the file and then modify it A typical entry would look like the following Display Name Order Table Name REACH_CARDS Field Name ORDER_20 Group Id 1 5 1 1 5 16777215 1 65535 2 16711935 3 255 4 16776960 5 The first 3 rows of the entry are self explanatory The Group Id row determines which of the four mapping subsets the uppermost combo box on the main FHAT20 form the variable is associated with Physical 1 Observed Fish Data 2 Fish Predictions 3 Stream Classification 4 The next line has four fields the number of categories in the display e g 5 the type of variable 1 numeric 0 text 1 boolean the lowest possible value in the dataset e g 1 and the highest possible value e g 5 The remaining entries specify the color for each category and the category number Since the colors are represented by long integers it is probably easier to modify the entry what variable number of bins etc and then edit the colors from the Legend Editor in FHAT20 once you have made the changes Note to reload the legend file MODEL MDE in FHAT20 you must close the
42. al FDIS identifier FDIS WS CODE The original watershed code from the FDIS database FDIS REACH ID The original reach id from the FDIS database MODEL WS CODE The corrected watershed code used in the MODEL database you must enter the corrected code in the table MODEL REACH ID The corrected reach id used in the MODEL database you must enter the corrected id in the table ILP Interim locator point number from FDIS ILP MAP Map sheet number associated with interim locator point COMMENTS User defined comments concerning errors and changes FIXED True False denoting that the corrected watershed code and reach id that you entered was accepted during the last import process IsLake True False denoting that the record is a lake and has a corresponding record in the MODEL Lakes table and the FDIS Lake Cards table TABLE TYPE If a reach in WSCODE LOOKUP has corresponding site data physical fish lake or obstruction the table name s is specified Users should check that the records in these tables do correspond with the modified watershed code and or reach specified in WSCODE LOOKUP ERRORDESCRIPTION String denoting the type of error and which tables are potentially affected ERROR TYPE An internal code an integer value see Table 2 used for error correction computations in FHAT20 during the importing process This field only needs to be set by the user to change a watershed cod
43. as there is not a fish occurrence for group upstream of this point Network Group US Fish Abs Group Parent D S of Fish Occurrence Reach is a tributary of a reach that is downstream of a known fish occurrence Network Group Parent DS Fish Pres Observed Fish Data Group Reach Presence Abundance Denotes fish presence if the site was not electrofished and density 22 100 m if the site was electrofished Based on a combination of variables in Model Fish Data Meth TotalNo amp Cpue Table 3 Con t Group Name Variable Name Description Table Field Fish Predictions Group Range The maximum range of the fish group in the watershed Res Fish Group Dist Group Most Probable Capability Class The most probable fish habitat capability class Res Fish Group Most Prob Class Group Probability of No Capability The probability that the reach has no capability that the abundance is less than 1 fish in the sample site area Res Fish Group Prob Absent Group Probability of Low Capability The probability that the reach has low capability Res Fish Group Prob Low Group Probability of Medium Capability The probability that the reach has medium capability Res Fish Group Prob Medium Group Probability of High Capability The probability that the reach has high capability Res Fish
44. ase the probability of presence for the inlet outlet reaches will be set to 100 e first reach of a tributary flowing into a reach that is downstream of a known fish occurrence In this case the probability of presence for the tributary reach will be set to 100 Since a pdf is not computed for reaches classified as lakes probability of presence is assumed to be 100 if the lake is within the fish group s distribution range otherwise it is set to 0 46 May 2000 FHAT20 User s Guide Upstream Limit of Fish Group Range in Watershed based on Distribution Model and Reach Break m Reach Break Only PoP 0 Po P 0 100 PoP 100 X Most Upstream Observed Fish 1st reach of tributary whose parent Occurrence and Reach Break reach is downstream of known fish gt occurrence PoP 10096 Downstream C Downstream of known fish occurrence PoP 100 Figure 18 Schematic showing how the probability of fish presence PoP is computed for each reach as a function of modelled upstream distribution limits predicted by fish distribution rules and observed fish occurrences If the reach is upstream of the modelled distribution limit then the probability of presence is set to 0 If the reach is downstream of a known occurrence then the probability of presence is set to 100 If the reach is upstream of a known occurrence but within the distributi
45. aximum values of the transformed abundance data are used to define a range of possible abundance values The observed data range is then expanded by 4 times the window width to accommodate extrapolated values and then divided into 150 equally spaced intervals The result is a new grid variable that is used to map the pdf A small scalar value is then computed and added to each grid point to ensure that at one of the grid points is equal to zero A corresponding variable is also created to store the probability values that will be associated with each of the grid points as the kernel function is calculated The grid variable is then used to create a matrix of data consisting of the grid points and the habitat variables used as the predictor of abundance The habitat variable values are identical for all of the grid point data This matrix forms the set of x vectors used in Eq 6 The transformed observations form the set of Xi vectors To improve computational efficiency all scalar calculations in the kernel function are moved outside the summation step The resulting kernel density estimation function is as follows To improve computational speed the summation component of Eq 10 is done only for those grid points that lie within a 5 range about the Xi of interest rather than the enter grid network This results in a small loss in accuracy but given the course resolution of the fish abundance categories this loss is negligible The scalar calculati
46. ccccccccecccccscccccececececsesescscsesesesesesssssesssssssesessssssssssseseseseesess 55 10 3 ADDING VARIABLES TO DISPLAY ON THE MADP ce eee 56 11 0 REFERENCES E 57 APPENDIX I THEORETICAL BACKGROUND OF MULTIVARIATE KERNEL UO 58 vi May 2000 FHAT20 User s Guide List of Figures 10 11 12 13 14 15 16 17 18 19 20 21 22 23 THE RELATIONSHIP BETWEEN FDIS DATA VARIOUS MODELLING STEPS PERFORMED BY THE USER IN FHAT20 AND THE MAJORITY OF THE FHAT20 MODEL DATABASE MODEL MDB WHERE MODEL DATA RULES AND RESULTS ARE SAVED cccccccccccscsecccececececececececececececececececececececececececececececececececesesevers 4 SCHEMATIC SHOWING THE PROCEDURE TO IMPORT DATA FROM FDIS INTO FHAT20 5 THE ERROR REPORTING DIALOGUE BOX DISPLAYS A RECORD OF ERRORS AND CORRECTIONS MADE TO THE FHAT20 MODEL DATABASE i upan dco ci ee Dese cd eene eo caeca eee nasi es Deco h eee ae ed ee vaca ed vba ao co ea eaae ee vo 7 THE MAIN FHAT20 DIALOGUE BOX USED TO DISPLAY AND QUERY REACH SPECIFIC DATA AND MODEL RESUETS 2 qapay eite ee tt eoe ud tear prae deo e COR E EAE EXER UP EA EB Ev heels Acie 12 THE SEARCH RESULTS DIALOGUE BOX DISPLAYS DATA AND RESULTS FOR REACHES SELECTED ON THE MAP anota ctii D cete t tc e ade a erac A oe ick a e a Poste dc seek eo D E CODE ERR Reb et conecto vet 13 THE LEGEND EDITOR DIALOGUE BOX
47. cific order Fig 7 1 Define stratification groups used to make physical predictions Section 3 2 Predict channel width wetted width bankfull depth optional and the probability of non visible channels for all unsampled reaches possibly using a stratified analysis Section 3 0 3 Define fish groupings Section 4 0 For each fish grouping 4 Editobstruction data to define whether a feature is an obstruction to each fish group Section 5 0 Model the fish group s range within the watershed Section 6 0 6 Model the fish group s habitat capability Section 7 0 which is combined with the predicted range to estimate probability of fish presence and 7 Model FPC stream classification Section 8 0 based on the predicted probability of fish presence and predicted channel widths A A change in data or modelling assumptions in any step requires that all steps following that point are reprocessed For example e if you modify the relationships predicting channel width and have fish range rules that depend on channel width you will need to rerun these rules i e repeat step 5 e If you modify channel width or wetted with predictions you must rerun the habitat capability calculations e If you modify channel width you must rerun the habitat capability and stream classification procedures e If you add or edit an obstruction that affects the range of a fish group you will need to rerun its range rule and since the habita
48. cy display parameters are a series of yellow boxes labeled 5 1 56 Fig 9 When you have computed the uncertainty for a set of models the values shown in these boxes display the probability that the test reach falls within each of the width classes associated with the 6 FPC stream riparian classification groups This probability is simply the area under the frequency distribution within a specific width range e g S2 5 20 When you have computed the uncertainty for a set of models and save results to the RES_FISH table a frequency distribution will be generated for each reach it will not be displayed to save on computational time but is generated internally and the probability of the reach being in each of the six channel width classes will be computed Eventually when you compute stream classification the probability of fish presence for each reach will be combined with its probability of being in each of the width classes to determine its FPC stream class designation See section 8 for more details on stream class computations A few simple rules to remember when using the Channel Morphology dialogue box e If you want to save channel width predictions you must first run the model s with the Compute Uncertainty box checked When the computations have finished click on the button labeled Save Model Results to Res Phys Table e If you want to save wetted width or bankfull depth predictions you do not have to run the mod
49. d sample data This feature allows you to look at two different variables on the map at one time First select a variable and display it on the map click on the check box labeled Overlay Second Layer and then select the second variable you want to display The second variable will be displayed as color coded squares shown in the center of each reach on top of the first variable Click on the Show Overlay Legend check box to bring up legends for both layers simultaneously May 2000 13 Table 3 Organisation of variables data and model results that can be displayed in the FHAT20 Map Interface The Group and Variable Name columns correspond to the items shown in the combo boxes labelled Group and Variable on the FHAT20 main window The Table Field column gives the table and field name where the data results are stored in MODEL MDB Group in the table is used to designate a fish grouping name e g FPC_FISH there will be multiple occurrences of fields preceded with Group one set of each fish group modelled Group Name Variable Name Description Table Field Physical Predictions Order Strahler stream order Reach_Cards Order_20 Magnitude Magnitude of reach of 1 order streams upstream Reach_Cards Magnitude_20 of reach Gradient Reach gradient Reach_Cards Gradient_20 Downstream Elevation Elevation metres above MSL of downstream node of reach Reach_Cards Downstream_Ele
50. d to the larger one Assuming equal pdfs densities in these two sites for arguments sake the probability of fish not present will be therefore be lower in the larger site because we are sampling a larger area relative to the small site FHAT20 accounts for this dynamic when computing the probability of not present for each reach Sample site size for each unsampled reach is estimated based on the predicted channel and wetted widths of the reach The minimum density below which the reach can be declared to be Not Present of fish is 1 fish divided by this estimated site area Large sites will have lower Not Present density limits than smaller sites A numerical example may help clarify these notions Say you have two sites in two different reaches with equal densities of 1 fish 100 m in both sites One site is 100 m and the other is 500m Thus one site contains only one fish while the other has five You classify a site as present in status if one fish or more is caught regardless of its size Thus under equal densities 1 fish 100 m in this example the probability of classifying the 500 m site as not present is 5 times less than the probability of classifying the 100 site as not present Thinking about it another way assuming you sample both sites with the same degree of effort per unit area you are five times more likely to catch a fish in the larger site since there are five times as many fish in your sampled area even tho
51. e Parent Stream A tributary of a parent stream may have a code specifying that its confluence is mid way up the parent e g 120 907500 63000 50000 located 50 of the way up its parent stream 120 907500 63000 However during inspection of the FHAT20 electronic map plotting variables determining which tributaries are not accessible to fish because of obstructions you may notice that the confluence of the child stream is actually further upstream that specified by the watershed code Assuming the map is correct such an error could result in erroneous predictions For example an obstruction in the parent stream located 60 of the way up its total length on the map would not limit access to the tributary because electronically the obstruction is upstream of the tributary confluence which is only 50 of the way up the parents total length To correct this type of error you need to estimate the correct watershed code for the tributary in question based on the actual proportional distance of its confluence relative to the parents total length e g 77596 A new record in WSCODE LOOKUP must be added containing the original FDIS and modified MODEL watershed codes e g 120 907500 63000 50000 and 120 907500 63000 75000 respectively Watershed Code Does Not Reflect Stream Hierarchy A stream may appear electronically to be a tributary of a particular parent stream based on its watershed code when in reality it is not a tributary of that pare
52. e functions for different sets of reaches For example while channel width will be correlated positively with upstream length we would expect unconfined reaches to be wider for a given upstream length than confined reaches Hence it is logical to develop different relationships for subsets of the entire dataset which we refer to as strata because you are stratifying the data into different sets Stratification can improve the accuracy and precision of the physical predictions assuming that there is a sufficient sample size to develop the empirical relationships within each strata FHAT20 provides a mechanism to e define these strata based on attributes in the REACH CARDS table of FDIS remote sensed attributes which are available for all reaches e build and evaluate models for each of these strata e estimate uncertainty in the predictions and e save the predictions to MODEL MDB for display and use in fish range habitat capability and stream classification modelling To predict widths and bankfull depths open the Channel Morphology dialogue box via the Modelling main menu item Fig 9 Select a variable to model Channel Width CW avg or Wetted Width WW avg Bankfull Depth ZBFD avg from the combo box at the top of the dialogue box labeled y axis The relationship between this variable and upstream length for the sampled reaches will be shown in the adjacent x y scatterplot as blue x s Below the scatterplot is a wh
53. e measure for the obstruction in question by x km reprocess the obstructions and view the Upstream of Obstructions field on the main form for the appropriate fish group If the problem tributary is still inaccessible increase the measure again and repeat the process until the desired results are achieved the tributary becomes accessible 36 May 2000 FHAT20 User s Guide 6 0 Modelling Fish Range in the Watershed FHAT20 uses fish range rules based on remote sensed reach attributes and obstruction data to model the potential range of a fish group within a watershed Fish range represents the maximum potential distribution of the fish group in the watershed This should not be confused with the computation of probability of presence in each reach which combines information on known ranges predicted potential ranges and habitat capability See Section 7 0 for details on how probability of presence is calculated A typical fish range rule might look something like the following MAXDSGRADE lt 20 AND ORDER gt 3 AND FPC_FISH_DSBARR FALSE that states that the range of the FPC_FISH fish group will only be in reaches that 1 have a maximum downstream grade less than 20 the maximum gradient of any reach below it is less than 20 2 have a stream order greater or equal to 3 and 3 do not have any obstructions downstream specific to the fish group as defined in the Obstruction dialogue box Because the range rules must be
54. e or reach id of MODEL MDB site data to assign it to another reach in the REACH CARDS record set 2 2 2 Error Correction To correct watershed code reach id errors open WSCODE LOOKUP in Access fill in the MODEL WS CODE and MODEL REACH ID fields with the correct values and save the table Since WSCODE LOOKUP will always retain the original FDIS FDIS WS CODE FDIS REACH ID and corrected MODEL WS CODE MODEL REACH ID values you will always be able to link the model results interpretive products back to the original FDIS database should the need arise When the corrections have been made to WSCODE LOOKUP close Access and select the Rebuild Database choice from the Utilities main menu item to re initiate the import process Rebuilding the database incorporates all the changes made in WSCODE LOOKUP into the new MODEL MDB database Hopefully all errors will be corrected during the MODEL MDB rebuilding process If errors are still reported repeat the sequence just described but this time only correcting records in WSCODE LOOKUP where the FIXED field FALSE Records where the FIXED field TRUE were corrected during previous rebuilding events This cycle can be repeated as many times as required The error reporting form displayed at the end of each import process always provides a summary of the current state of the data in terms of how many errors were originally detected and how many have been corrected 6 May 2000
55. e same time estimating the range limits in parts of the watershed that have not be sampled The key to matching known range limits is entering them as obstructions in the Obstruction dialogue box regardless as to whether these observed limits are caused by actual migration barriers or other factors that cannot be modelled by the remote sensed characteristics available in the reconnaissance dataset e g temperature One could also use obstructions to restrict the downstream range of a headwater type fish groups or species For example a range rule with Group _DSBARR TRUE would restrict the range of the fish group to reaches upstream of barriers for this group Review of existing fish range rules or entering new ones is done through the Fish Range dialogue box accessed via the Fish Range option under the Modelling main menu item Fig 15 Select the fish group to review or develop a range rule for from the combo box at the top of the dialogue box If you have previously saved rules for this fish group a list of them will appear in the combo box labeled Rule Name If you want to view one of these rules simply click on the name and the rule will appear in the white text box at the bottom of the dialogue box If you want to delete the rule click on the button labeled Delete Rule If you want to create a new rule enter a new rule name in the combo box labeled Rule Name Delete the contents of the text box showi
56. e uncertainty around the predictive relationship Alternatively you may know something about these sites which would motivate you to drop them from the model fitting because they do not reflect the 24 2000 FHAT20 User s Guide population of reaches you are trying to model For example there could be a problem with the measurements at a particular site difficult to determine the top of the bank or the site could be altered by human activity e g a rip rapped bank beside a road would not have a representative channel width You would want to exclude these types of sites from the model fitting exercise as they could bias your predictions and levels of certainty To exclude outliers click on the check box labeled Select Site Data to Include in Model A grid will appear with the list of sites that are used in the model fitting Fig 10 Green cells denote sites that are included in the modelling while red cells denote cells that have not been included To include exclude data from a site in the model fitting toggle the color of the cell by clicking on it with the mouse When you toggle a cell its corresponding reach will be highlighted on the map Sites included in the modelling are recorded in the UseFor CW WW _BFD fields in the RES PHYS table iw Define Data Set for Modeling Include or Exclude a data point click on it s cell Include Exclude WS CODE REACHD UPLEN 120 307500 120 907500 120 907
57. ecause there was a missing value for at least one of the variables used in the fish range model To save the rule itself to the RULES table in MODEL MDB the results are automatically saved when you run the rule click on the button labeled Save Rule To view the results from a range rule go to the main window and select the Fish Predictions variable group and the FISH Range variable To compare the modelled range with observed occurrences of the fish group select the FISH Reach Presence Abundance variable from the Observed Fish Data variable group Then click on the Overlay Second Layer check box on the main window and display the predicted fish range The predicted range will appear as dots on top of the observed range You will want to modify the range rules and possibly edit the obstruction table until 1 the modelled range limits do not exceed the limits of observed occurrences if those occurrences are known to be upstream range limits and 2 the modelled range limits extend at a minimum to all known occurrences Remember that you can double click on any reach on the map to look at its attributes This will allow you to determine what variable in the rule has caused the range to stop at a particular reach This information is helpful to modify the rule if required May 2000 39 FHAT20 User s Guide 7 0 Modelling Fish Habitat Capability Fish habitat capability is an index that measures the capability of stream
58. ect a watershed code or reach id in REACH CARDS in MODEL version by making changes in WSCODE LOOKUP and re running the import procedure you can create a situation where some records in FISH SITE have no match in REACH CARDS in the MODEL database This error can also arise from incorrect entry of watershed codes or reach ids when entering fish data in FDIS 6 Record in LAKES with no matching record in Reach Cards The LAKES table in MODEL MDB is initially populated based on watershed codes and reach ids in the FDIS table Lake Cards and these records are merged into MODEL MDB Reach Cards table When you correct a watershed code or reach id in REACH CARDS in MODEL version by making changes in WSCODE LOOKUP and re running the import procedure you can create a situation where some records in LAKES now have no match in REACH CARDS in the MODEL database 7 Record in USER BARR with no matching record in Reach Cards The USER BARR table in MODEL MDB is initially populated based on watershed codes and reach ids in the FDIS table FEATURE linked to S Site Cards When you correct a watershed code or reach id in REACH CARDS in MODEL version by making changes in WSCODE LOOKUP and re running the import procedure you can create a situation where some records in USER BARR have no match in REACH CARDS in the MODEL database This error can also arise from incorrect entry of watershed codes or reach ids when entering obstruction data in
59. edictions 30 May 2000 FHAT20 User s Guide 4 0 Fish Groups Fish groups are the basic unit used for modelling fish distribution and capability in FHAT20 A fish group is the sum total of all species and life stages that make up that group For example the ONCORHYNCHUS fish group consists of all salmonids of the Oncorhynchus genus Any reach where at least one fish belonging to this genus was caught or observed would be classified as present in status for the ONCORHYNCHUS group The relative abundance of ONCORHYNCHUS In any reach used for capability modelling would be the sum of all fish belonging to this genus caught by electrofishing divided by the site sample area When an FDIS dataset is successfully imported into FHAT20 three fish groups are automatically created ALL FISH All fish species and age classes found in the watershed FPC FISH All species of fish found in the watershed used by the Forest Practices Code to identify a stream as fish bearing Anon 1998c ONCORHYNCHUS All fish species and life stages belonging to the Oncorhynchus genus You can create additional fish groupings via the Fish Grouping dialogue box Note that a fish grouping can consist of a single species life stage or a combination of different species and or life stages This provides a very flexible framework for modelling specific species life stage combinations e g 0 rainbow trout through to broad management groupings e g all FPC species At the
60. els has been to rely on parametric multivariate statistical tools e g linear regressions that relate habitat attributes to an index of fish abundance The use of these tools require that certain assumptions be met regarding the form of the habitat capability function and the distribution of errors among habitat attributes and across the range of abundance measurements These assumptions however are rarely met James and McCulloch 1990 Rice 1993 and as a consequence tend to have weak predictive ability Of even greater concern are the estimates of certainty about these predictions Rice 1993 Confidence interval calculations require even stronger adherence to modeling assumptions Consider the distribution of fish densities obtained by electrofishing across a number of sites with a range of habitat qualities Fig 16a where in this example habitat quality is indexed by water surface gradient data from R Ptolemy BC Fisheries It is clear from Figure 16a that habitat factors do not directly control the abundance of animals but rather provide limits on maximum capability In low gradient reaches the majority of sites have low densities but a few sites are capable of high densities As gradient increases variability in densities is reduced and the high densities seen in reaches of low gradient are not attained Gradient does not have much effect on the mean density but does limit the ability of a site to produce moderate to very high densities
61. els with the uncertainty box checked only the best fit predictions for reach are saved to the database Once the best fit models predicting wetted width or bankfull depth have been computed for each strata click on the button labeled Save Model Results to Res Phys Table e 0 cases where some of the stratified models have low sample sizes you might want to base predictions on the unstratified model for these reaches Set the minimum sample size in the text box labeled Minimum Sample Size for Stratified Model to define this limit In some situations the model may predict a wetted width that exceeds a channel width based on stratified rules with few data points or incorrect data etc In such situations you may want to check the Ensure wetted width channel width box prior to saving the results The wetted width prediction for each reach will be compared to its predicted channel width and set to the channel latter value if wetted width exceeds channel width Note that you must predict channel and wetted widths for unsampled reaches as they are mandatory variables to perform the stream classification and habitat capability modelling steps respectively Fig 7 26 May 2000 FHAT20 User s Guide 3 2 Stratification of Physical Predictions Stratification groups are used to improve the precision of models predicting channel width wetted width bankfull depth and the probability that a channel will be non visible for reaches
62. esence defined on a continuous basis downstream reaches cannot have probabilities less than upstream reaches Res Fish ConPop FHAT20 User s Guide Controlling the Map Legend Each variable that can be mapped has a unique legend that controls how the data will be displayed on the map If you double click the mouse on the colored boxes in the legend the Legend dialogue box will be displayed Fig 6 Alternatively select the Edit Map Legend choice from the Utilities main menu item There are a number of parameters that you can change to adjust the legend To display a variable via color codes on a map a set of bins must be established for specific ranges of the variable e g bin 1 gradient 0 3 bin 2 gradient 3 7 bin 3 There is a corresponding color for each bin e You can alter the number of bins labeled number of strata in the dialogue box and manually set the lower and upper end of the range for the variable e Alternatively you can click on the button labelled Set Upper and Lower Ranges to have the program determine the minimum and maximum value for the variable e After you have made any of these edits click on the button labelled Ramp Breaks to ramp the breakpoints between the lower and upper range values e To change the colour associated with any bin click on a colour in the colour palette and then click on the coloured box adjacent to the bin e You can manually edit
63. f unconfined reaches only Repeat this procedure for other variables to include in the stratification scheme by adding additional variables to the lower left list box To save the new stratification scheme type its name in the dropdown list box in the upper left of the dialogue box and click on the Save Strata Group to Database button The stratification schemes will be saved to the ChanMorphStrata table in MODEL MDB May 2000 27 FHAT20 User s Guide q Define Stratification Groups for Analysis Existing Stratafication ABS Save Strata Group to Database Confinement Delete Strata Group from Database Browse Variables Available for Stratification D Linstratified 1 CUDE EN CO FE OC 2 CONF CODE UN NA Sample Size for Frequency Distribution 880 Upstream Length Order Gradient I Lenfinement Code __ Upstream Elevation Downstream Elevation Channel Pattern Gradient Group Pattern Group Basin Group Order Group zi CO EN WA NS UN Add Remove m Breakpoints for Selected Variable Variables Selected for Stratification Unique Values from all Reaches for of Breakpoints 2 Selected Variable CO Confined EN Entrenched Breaka JEN CO FC OC FC Frequently Confined Break 2 NA N Not Applicable pi NS Not Specified OC Occasionally Confined UN Unconfined Confinement Code
64. g area not just their distribution in sampled reaches and lakes These products must be interpreted from the sampled based inventory The Fish and Fish Habitat Assessment Tool FHAT20 is a computer program designed to analyze reconnaissance level inventory data to produce a set of standardized interpretive products FHAT20 is an extrapolation program used to estimate fish habitat characteristics fish presence and capability in unsampled reaches based on their remote sensed characteristics and models relating these characteristics to field based observations in the sampled reaches FHAT20 uses data stored in the Field Data Information System FDIS the standard reconnaissance inventory project database The end product from FHAT20 is a set of predictions of channel width and probability of fish presence for all reaches These predictions are used to estimate the most likely Forest Practices Code FPC stream class S1 S6 for each reach and the level of certainty associated with each prediction This user s guide documents the background theory and installation and operating instructions for the FHAT20 program There are seven basic modelling steps that must be followed 1 Define stratification groups used to make physical predictions Predict channel width wetted width bankfull depth and the probability of non visible channels for all unsampled reaches possibly using a stratified analysis 3 Define fish groupings For each fish grouping
65. h could not be computed FHAT20 uses the UTM coordinates from the REACH_CARDS table to create this stick diagram during the import process and saves this information to the file MODEL MIF in the project sub directory The stick map is actually a MAPINFO file that can be viewed in FHAT20 or MAPINFO When you view this map for the first time in FHAT20 after the import procedure you will note that the first reach of each stream hangs in the sense that it does not connect with its parent stream Such reaches will be represented by open circles at their upstream boundaries This occurs because FDISDAT MDB does not store the UTM coordinates for the downstream end for the first reach of each stream Some of these coordinates can be obtained from the 1 50 000 Watershed Atlas although coordinates may differ significantly from those on a 1 20 000 map and others may have been created by the ILP watershed code process If you have some or all of these first node coordinates create a comma delimited file with 3 fields watershed code 45 digits can contain hyphens or not and UTM easting and northing coordinates in the following format WS_CODE EASTING NORTHING 120907500000000000000000000000000000000000000 674071 5621503 120907500045320000000000000000000000000000000 674810 5622958 120907500045325767200000000000000000000000000 674344 5623477 From the Utilities main menu item select the Import First Nodes File option This will
66. ion limit reaches not predicted because of missing values for fields used in rule r False Negative n 33 reaches sampled for fish were predicted to have no fish when fish SaveRule Rani s Wetland were observed to be present Parent Order Delete Current Rule Probability of Non Visible Channel I watershed Code Reach ID Click On List to Insert Value Rule Operators Unique Values from all Values Where Fish Reaches Where Present Clear Enter Text 7 600001 E 02 0 1001 0 1063 0 111 0 111 0 116 r False Positive o of 3 reaches sampled for fish were predicted to have fish when na fish were observed to be present Watershed Code Reach ID 120 307500 21600 07 J 120 307500 21600 12 7 120 307500 21600 12 1 E Predict range in lakes based on predicted range value in nearest downstream fluvial reach Fish Range Rule Network ALL FISH DSBarr 0 AND Network MAXDSGRADE lt 20 120 907500 21600 12 1 120 907500 21600 43 1 120 907500 21600 43 4 Figure 15 The fish range dialogue box is used to review and edit rules that predict fish group range in the modelled watershed Fish range results are automatically saved to the RES FISH table in MODEL MDB under the field name Group_DIST where Group represents the name of the fish group Presence is denoted by a value of 1 absence by 0 and 9999 denotes that a prediction could not be made b
67. it is processed You want to ensure that the KeepOnRebuild field is checked This flag will ensure that the new data you enter will not be lost if you rebuild the database at a later date Enter a date in the Date field Enter any comments describing the site or data source in the Comments field Ensure that the FDIS_REC field is not checked default condition When user defined data is imported to the FISH_SITE table this field distinguishes FDIS data from user entered data After entering the data close the USER FISH SITE table and select the Process Fish Site Data sub menu choice below the Utilities main menu item This will initiate a procedure to ensure that there are corresponding records in the Reach_Cards table in MODEL MDB for each record in USER_FISH_SITE The FHAT20 table summarizing the fish data by fish group the MODEL_FISH_DATA table in MODEL MDB will then be rebuilt and Network type variables related to known fish occurrences will also be recalculated Note you will need to rerun fish range capability and stream classification models after the new fish data have been processed If a user defined record has been successfully transferred to the FISH_SITE table the ValidSite field in USER_FISH_SITE will be checked If this field is not checked after you have Processed the Fish Site Data then the watershed code and reach_id for the site s in question does not exist in the MODEL MDB version of Reach_Cards Edi
68. ite box containing a list of strata for the currently loaded stratification rule If you click on a string in the white list box the x y scatterplot will display the data for the strata you selected The May 2000 23 FHAT20 User s Guide first item in the list box 0 Unstratified always shows the entire sampled dataset If you want to load a different set of strata that had previously been defined select one from the combo box labeled Stratification If you want to define a new set of strata select the Define Stratification Groups choice below the Modelling main menu item Section 3 2 below Predict Channel Morphology Mm x Y Anis Y A X B Uncertainty in width predictions Ew ava A Axis 3 ji UPLEN 4 A S E T A ah m Stratification CW 3 A Prob Confinement avg j AR E a 90 x x 1 Run Models 1 m J Compute Uncertainty 0 j i Bayes fon 10 20 30 40 dn 05 t 132 25 t t5 A 5 progress UPLEN avg Save Results for Current Variable Only Minimum sample size for B Ensure meted channel width 3 stratified model Select Site Data to Include in Model D 7 LInstratified Visualize Bayesian estimate of Uncertainty Plot Test Reach Upstream length for test reach krn fi Histogram Maximum m 5 Ex NA gt Pub f Histogram Bin Size m or DENN XIV ae er 0977 0322 045 0235 4916 03
69. itical step in using FHAT20 is importing data from FDIS Section 2 0 describes how this importing procedure works and what to do when errors in the FDIS database are detected The remaining sections provide details on the following FHAT20 modules 1 A physical habitat module which predicts channel width wetted width bank full depth and the probability of non visible channels in all non sampled reaches Section 3 0 2 Afish grouping module which allows users to summarize species life stage specific presence and abundance data collected from the reconnaissance survey into management relevant fish groupings Section 4 0 These groupings are then used in all subsequent fish distribution and capability modelling An example of a fish grouping would be all species identified under the Forest Practices Code FPC that are used to classify a stream as fish bearing 3 Anobstruction module that allows users to visualize and edit fish obstruction information collected during the reconnaissance survey and from other surveys These obstructions are used to restrict the range of the fish grouping within the watershed Section 5 0 May 2000 1 FHAT20 User s Guide 4 A fish range module that allows users to predict a fish group s range in the watershed based on observed fish occurrences obstruction data and rules that use predicted or remote sensed information Section 6 0 5 A fish habitat capability module that uses remote sensed reach
70. ivariate pdf but using the multivariate kernel function Eq 6 with the habitat variable set to the value of interest This rational can be extended to multiple habitat variables The use of conditional probability functions to develop estimates of abundance from habitat data is analogous to a non parametric form of multiple regression analysis Rice 1993 No a priori assumptions are made with respect to data distributions and functional form thus avoiding the statistical difficulties of regression techniques in general Further predictive equations are not developed Rather abundance predictions are in the form of pdfs that are constructed from the data itself The computational algorithms follow that described by Silverman 1982 All abundance and habitat variables are checked for symmetry using the g statistic for that describes the level of skewness Variables that have g values greater than 1 are log10 transformed to restore a level of symmetry The variables are then standardized to have an average of 0 and a standard deviation of 1 Transforming the variables in this way allows the window width to be calculated using a May 2000 59 FHAT20 User s Guide simple relationship that has proven to be robust under a wide range of data conditions Window width is calculated as follows Optimum Window Width 41024 y nV 9 A computational grid on to which the pdf is mapped is then created using the abundance data Minimum and m
71. jo Figure 14 The obstruction editor dialogue box is used to display edit and add migration barriers to the FHAT20 model database Any characteristics of the feature such as its type height length and any comments that were entered in the FEATURE table will also be displayed To view a map of the features listed in the table select the Features theme from the Physical Characteristics table on the main form In the Obstructions table the column labeled ACTIVE determines whether an obstruction has he potential to be a migration barrier for at least one fish group The remaining columns on the right side of the table are labeled after the fish groups TRUE FALSE 1 0 values denote 34 May 2000 FHAT20 User s Guide whether the feature is a migration barrier for that group You must set these values for all fish groups for any obstruction that is active ACTIVE 1 To navigate in the grid click on the appropriate record and move to specific columns by using the left and right arrow keys Note the following conventions e If the ACTIVE field is set to 0 then the obstruction cannot be a migration barrier for any of the fish groups e g FPC FISH FALSE e Ifthe obstruction is a migration barrier for any of the fish groups then the ACTIVE field must be set to 1 e The ALL FISH group represents all fish species This fish group must therefore have the widest range in the watershed com
72. l estimate the pdf for see step 3 below and the average minimum maximum and standard deviation of values in the reference data set sites that have been sampled for fish within the watershed range of the fish group When you click on a row in the table the graphic to the right of the table shows the relationship between this habitat variable and fish abundance in the reference data set a two dimensional version of Fig 16a Each cell in the graphic shows the percentage of sample sites in a particular fish abundance habitat value combination The cells are color coded red high percentage light gray low percentage based on the legend shown at the top of the graphic labeled of sampled sites The white text box to the right of the legend can be edited followed by hitting the return or enter key to rescale the legend To determine which variables to include in the model click on each row in the table and examine the values of cells in the adjacent graphic Habitat variables that show strong patterns between abundance and the habitat attribute should be included in the pdf model Select the reach to compute the pdf for The frame labeled Process PDF for Single Reach on the right side of the dialogue box contains a unique list of watershed codes When you click on a watershed code in this list the Reach_ID combo box below the list will be updated showing all reaches for this stream Select the reach from this combo box to comp
73. les i eb BON Rues Res Phys Distribution S Classes Capability a o 7 o oO c o m UJ o mi Ja CR CRT C RC CR FDIS Database i Channels Model Database aig aa saa ae i Capability Modelling Process lt _ lt lt Models 2 Figure 1 The relationship between FDIS data various modelling steps performed by the user in FHAT20 and the majority of the FHAT20 model database MODEL MDB where model data rules and results are saved A May 2000 FHAT20 User s Guide 2 2 1 Error Checking Procedure When you first load FDISDAT MDB FHAT20 automatically checks FDIS data for inconsistencies in the watershed codes and reach ids These inconsistencies must be corrected because the watershed code and reach id fields are critical to determine the connectivity of reaches and streams within the watershed the upstream and downstream neighbours of each reach Connectivity is used to compute a number of variables in the MODEL MDB database for example the total stream length upstream of each reach used to predict channel width or whether the reach is upstream of an obstruction or downstream of an observed fish occurrence The error checking procedure also detects missing values for any FDIS variables that can potentially be used in FHAT20 modelling procedures The first step that the FHAT20 import procedure completes is the combination of the Reach_Cards and Lake_Ca
74. n schemes The most predictive stratification is the one which provides the most contrast in probabilities among strata i e some strata will have very high probabilities of a reach being a non visible channel while other strata will have low probabilities while still maintaining sufficient degrees of freedom in each strata class The model will default to using the unstratified predictive relationship for any reach that falls in a stratification class that has a relationship based on less than 3 data points If an unsampled reach cannot be placed in one of the strata because it has a remote sensed value that is outside the ranges of the stratification classes the default unstratified relationship will be used to predict its probability of being a non visible channel The strata class for each unsampled reach is saved in the NOVIS_STRATA field in the RES PHYS table LM Non isible Channel ioj x Stratification UpstreamLenath gt Save Results to RES PHYS Table J Unstratified t 1 UPLEN 2 2 UPLEN 10 3 UPLEN 50 4 UPLEN 1000 Non Visible n 41 Figure 12 The non visible channel dialogue box is used to predict the probability that unsampled reaches will be not be visible channels If you need to define a new stratification scheme access the Stratification dialogue box from the Define Stratification Groups option from the Modelling main menu item Section 3 2 May 2000 29 FHAT20
75. n the map when viewing the Features theme you can move directly to the appropriate row in the obstructions table by clicking on the Select Record in Obstruction Grid button on the Search Results dialogue box May 2000 35 FHAT20 User s Guide Once you have added a new record s to the obstruction table and defined the obstruction location via its measure or reach_id you will need to fill out the other columns in the table You can add notes about the obstruction and the source of the information in the Comments column Note that obstructions imported from the FEATURE table in FDIS are identified by the string FDIS preceding the rest of the text in the COMMENT field in the obstructions table Set the KeepOnRebuild field to True 1 to ensure that the obstruction s you have entered will not be lost if you rebuild the model database at a later date Set the values for the ACTIVE column for new records to 1 if you have added an obstruction presumably it is an obstruction to at least one fish group You must then set the values for all fish group columns If you want to delete a feature select if from the obstruction table and click on the button labeled Delete the Selected Obstructions When you are done editing the obstruction table click on the button labeled Process Obstructions This will calculate a field called Group_DSBarr where Group fish group name in the NETWORK table in MODEL
76. n to remove any variables from the list box at the bottom left hand corner of the dialogue box these were variables included in the currently selected stratification group Then click on a variable from the list of available variables A histogram will be displayed to the right of the list showing the distribution of values for the selected variable and the number of reaches from the total dataset with non missing values You do not want to select a variable to be included in the stratification scheme if there are many missing values You can use the histogram to determine appropriate breakpoints for each strata class for this variable To include a variable select it and click on the Add button The variable will now appear in the list box at the bottom of the dialogue box If you then select this variable again from the lower list box a list of unique values in the entire dataset will be presented to the right You then need to define the number of bins classes or breakpoints for this variable Click on a unique value and then on the text box for a particular breakpoint to populate that text box You can also enter values manually Note for string variables you can have multiple values for a single breakpoint For example if you used stream confinement code you might define confined and unconfined classes The former would consist of entrenched confined frequently confined and occasionally confined reaches while the latter would consist only o
77. ned for missing values DataCheck allows the user to modify the range of legal values that a field can have and what to do if a particular record does not meet these criteria Records that fail the data screening procedure for a given field are either deleted or the missing value is replaced with a missing value flag 9999 for numeric fields or NS for text fields The is Dropped field in the BadData table specifies which of these two actions will be taken s May 2000 FHAT20 User s Guide 2 2 3 FHAT20 Maps Stick Diagrams or the Digital TRIM Atlas A key component of FHAT20 is the ability to view some FDIS data and all model predictions on a digital map of the study area Two types of maps can be used by FHAT20 1 a stick diagram where each reach is a straight line between the upstream and downstream reach coordinates and 2 a TRIM atlas which is a digital version of the hardcopy TRIM maps with watershed codes and reach measures the distance of each reach break from the confluence of each stream for each reach in the FDISDAT database Stick Map The stick map represents each reach in the study area as a single straight line Such diagrams do not contain bends lake boundaries and other features that will be shown on the TRIM atlas maps Stream reaches are shown as straight lines while reaches classified as lakes 1 with records in the FDIS Lake_Cards table are shown as fatter lines or filled circles if the lake lengt
78. ng the last rule displayed Then select a variable from the list box on the right side of the dialogue box If you single click on a variable in this list the adjacent list boxes will be populated with 1 unique values for all reaches in the dataset and 2 unique values for reaches where the fish group was observed to be present The latter list will provide some guidance on maximum or minimum attribute values to use as part of the rule Double click on the variable to add it to the rule The variable name as it appears in the model database MODEL MDB will now appear in the lower left text box Select an operator e g lt gt and a value from the list box of unique values e g NETWORK MAXDSGRADE lt 20 Your rule will often contain more than one variable so you will have to combine them with the or AND operators see the example rule above The LIKE and NOT LIKE operators are used to evaluate text variables such as confinement code e g REACH CARDS CONF CODE LIKE FC All rules should contain the statement Group _DSBARR FALSE if there are known barriers for the fish group being modelled otherwise the modelled distribution could extend upstream of these obstructions After a fish range rule has been developed it must be applied to all reaches in the dataset by clicking on the button labeled Run Rule As FHAT20 cycles through all the reaches in the dataset it inputs the remote sen
79. ning site data PHYS SITE FISH SITE USER BARR LAKES Making a change to a watershed code or reach id in REACH CARDS via modification of WSCODE LOOKUP may result in two situations a Arecord in a site table may no longer have a corresponding record in REACH CARDS in which case error Types 4 7 will be reported in WSCODE LOOKUP or b Arecord in a site table may match a record in the corrected REACH CARDS table but the site data does not belong to that REACH CARDS record and instead is actually associated with another reach May 2000 7 FHAT20 User s Guide Table 2 Description of error codes reported in WSCODE LOOKUP Code Error Description 0 Invalid Watershed Code Watershed code is invalid e g 000 999 NA 1 Duplicate Reach ID Two records in the FDIS table Reach Cards have the same watershed code and reach id This likely resulted from a duplication resulting from the importing of watershed codes into FDIS or an incorrect specification of reach id or watershed code when data were entered in FDIS 1 Duplicate Reach ID reach is a Lake A record in the FDIS Reach_Cards table matches same watershed code and reach_id a record in the FDIS Lake_Cards table In this case the isLake field in WSCODE_LOOKUP will be TRUE This error type is distinguished from the previous error type only because it may be helpful to know which record is the lake when attempting to determine the correct
80. nn nnn 30 4 0 FISH GROUPS ua a auqa waqa Gua eene rosa ed ee aee Z wS ano eo aas eV eoa ease 31 4 1 ADDINGEISH SITE DXATA etude RERO 33 5 0 OBSTERUCTIQONS 5 ette eee erae eo seo ge voe qo ene 1 eene esas eu pee ue epa ent da cue pU sasay 34 6 0 MODELLING FISH RANGE IN THE WATERSHEL 37 7 0 MODELLING FISH HABITAT CAPABILITY 40 7 1 OVERVIEW OF METHOD USED TO PREDICT FISH CAPABILITY eee eee nennen nnn 40 7 1 2 Prediction of Capability Classes based on a PDF sess 42 7 2 USING THE FISH HABITAT CAPABILITY DIALOGUE BOX 43 7 3 OUTPUT INDICATORS FROM HABITAT CAPABILITY MODELLING 45 8 0 FPC STREAM CLASSIFICA TION 49 9 0 EDITING MODEL RESULIS 53 9 1 AUTOMATED EDITING AND THE EDIT ee ee e nnn nn nnn nnn nnn nnn nn nnn anna anna nn 53 10 0 CONTROLLING FHAT20 BY MANIPULATING TABLES AND FILES 55 10 1 ADDING VARIABLES TO INCLUDE IN 55 10 2CONTROLLING DATA CHECKING 0 cccccccc
81. nt stream or vice versa For example 120 907500 63000 50000 0123 would electronically be treated as a tributary of 120 907500 63000 50000 However on inspection of the map the stream in question may actually be a tributary of 120 907500 63000 Its watershed code needs to be modified in WSCODE LOOKUP to reflect its actual position in the stream hierarchy and its proportional confluence distance relative to its parent stream 120 907500 63000 57000 During the importing routine FHAT20 catalogues records with missing values for variables that can be used in the modelling process There will be a unique record in the BadData table of MODEL MDB for each field with a missing value for a given reach 1 there can be multiple records per reach If you want to enter valid values for the records with missing values identified in BadData you must enter them in FDIS and then re import the data using the Rebuild Database option from the Utilities main menu item Fig 2 Note that it is possible to continue modelling with missing values however FHAT20 cannot make predictions for reaches with missing values for a variable if that variable was included in one of the model s The table DataCheck in MODEL MDB contains the list of variables in FDIS that will be screened by the data checking procedure during the import process You can add variables to this list or deactivate variables by unchecking the IsActive field so they are no longer scree
82. olling Data Checking The DATACHECK table allows users to control how variables are screened for missing values during the importing process see Section 2 2 The FieldName and TableName fields denote the name of the variable that will be checked and the table it resides in The isText field denotes whether this variable is text or numeric in format The remaining variables determine how the variable is screened and what happens when missing values are found MinLegalVal the minimum value the variable can have e g gradient 20 MaxLegalVal the maximum value the variable can have DefaultV al the value that the variable will be set to in MODEL MDB if it is outside of the range specified by MinLegalVal and MaxLegalVal DeleteRecIfBad determines whether a record with a missing value will be deleted if it is outside of the range specified by MinLegalVal and MaxLegalVal IsActive determines if the variable will be screened during the import process This variable allows you to maintain the checking parameters in the DataCheck table but to not screen the variable by setting isActive to false May 2000 55 FHAT20 User s Guide 10 3 Adding Variables to Display on the Map The ASCII text file MODEL MDE located in the project data sub directory controls what variables can be displayed on the map and how they are displayed 1 the legend Modifying the legend e g what colours are used how many categories for variables that can alre
83. on in the stream If only the reach id is specified and the measure field distance in km from the obstruction to the most downstream end of the stream i e its confluence is blank the model will assume that the obstruction is located at the upstream end of the reach you specify and the measure will be computed accordingly Alternatively you can enter the measure and the model will use this value as the barrier location After processing the obstruction table the reach id corresponding to the measure will be shown in the table Set the measure value to zero and the reach id value to a legal entry if you want the model to use the upstream location of the specified reach as the obstruction position The second way of adding a new obstruction is to double click on the desired obstruction location on the map from any theme in the main window When you do this a Search Results dialogue box will appear with a list of reach attributes If you have selected the correct reach to add the obstruction to click on the button labeled Add Obstruction to Selected Reach The obstruction will then appear as a new entry at the top of the table and the reach id column will automatically be populated You then have the option of specifying a specific measure for the obstruction or to allow the model to assume that the obstruction is located at the upstream end of the reach If you double click on a reach with an obstruction when viewing existing obstructions o
84. on limits of the fish group the probability of presence is generally 100 X where X is the predicted probability that the capability in the reach is less than 1 fish over the typical sample site area for that reach probability of not present May 2000 47 FHAT20 User s Guide Upstream Limit of Fish Group Range in Watershed based on Model Results and Reach Break PoP 096 Fluvial Reach d Unsampled lake outside of distribution limits PoP 0 PoP 0 a Unsampled lake outside of fish range XS Unsampled lake immediately upstream of reach within fish range Outlet range is within fish range but lake is outside of range so PoP 096 Ew 100 X 9 Inlet reaches to lake within fish range PoP 100 PoP 100 X Inlet reach to lake within fish range PoP 100 Unsampled lake within fish range PoP 100 IU First reach of tributary where parent reach is downstream of a known fish occurrence PoP 100 Sampled lake which is fish bearing Downstream known fish occurrence Reaches downstream of known fish occurrence Outlet reach of lake within fish fish bearing lake PoP 100 range and downstream of known fish occurrence PoP 100 Figure 19 Schematic showing how the probability of fish presence PoP and fish distribution is computed for stream and lake reaches in relation to fish bearing lake
85. ons only need to be calculated once 1 m 0 5 t 1 f x On an 4 exp 0 5 t t 10 Because the probability values will be extremely small numbers 10 16 10 32 there is a high risk of numerical error when scaling the pdf using the appropriate scalar function Another error is introduced because of the limited range of the grid which cut off the very extreme ends of the tails The cut off tails are not significant in practical terms as they tend to be extremely small values 10 32 but they never the less affect the numerical integration of the pdf at the computational level Another problem is that the scalar formula applies to the full multivariate pdf and not the conditional pdf of interest To avoid these complications the scalar is estimated by integrating the pdf estimate as calculated in Eq 9 and then dividing the probability of each grid point by the integrand The result is a pdf that integrates fully to one with a 10 10 order of magnitude for accuracy Because the gaussian kernel is not bounded in any way negative abundance values will be possible There are two ways to deal with these negative abundance data The first is to simply accept the negative values and include their integrand as part of Fish Absence probability estimate From a numerical perspective the disadvantage of this approach is that the pdf on the positive line will not integrate to one and therefore will not be considered a true pdf
86. oother pdf estimate can be obtained by considering other functional forms of the kernel The most common and intensively studied kernel is the Gaussian normal distribution function where 1 0 5 E K t 5 This is the kernel function used in the present algorithm The resulting pdf estimate is continuous and integrates to one thus meeting all the criteria of a true pdf One of the advantages of kernel density estimation over simple frequency histograms is that only one parameter must be set to construct a pdf To construct frequency histograms two parameters are required the bin width and origin Both parameters can have dramatic effects on the shape of 58 May 2000 FHAT20 User s Guide the resulting pdf estimate Kernel estimates require only one parameter the window width h and will therefore always be a more robust estimate of the underlying pdf The choice of window width h can be done subjectively through graphical assessments based on a reference to a standard distribution or done more objectively through cross validation techniques Variable window widths are also possible where h is varied depending on the local density of data The use of variable window widths is a class of density estimation techniques that are referred to as adaptive kernels To maximize computational speed window widths in the present algorithm are calculated based on a reference Gaussian distribution Window width in this case is analogous to
87. opment of products showing the extent of fish distribution or stream channel widths for the entire planning area not just their distribution in sampled reaches and lakes These products must be interpreted from the sampled based inventory The Fish and Fish Habitat Assessment Tool FHAT20 is a computer program designed to analyze reconnaissance level inventory data to produce a set of standardized interpretive products FHAT20 is an extrapolation program used to estimate fish habitat characteristics fish presence and capability in non sampled reaches based on their remote sensed characteristics derived from 1 20 000 scale maps and air photos and models relating these characteristics to field based observations in the sampled reaches FHAT20 uses data stored in the Field Data Information System FDIS the standard reconnaissance inventory project database The end product from FHAT20 is a set of predictions of channel width and probability of fish presence for all reaches These predictions are used to estimate the most likely Forest Practices Code FPC stream class 51 56 for each reach and the level of certainty associated with each prediction This user s guide describes how to use FHAT20 and the assumptions and methods of its modelling approaches Section 2 0 describes how to install FHAT20 and provides an overview of the steps that you must follow to develop standard interpretive products of fish habitat distribution and capability A cr
88. pared to any other fish group and would therefore be affected by the smallest number of obstructions relative to other groups In other words the ALL FISH obstruction field should only be set to 1 if it is an obstruction to all species and life stages found in the watershed There will be at least two circumstances in which you will want to add additional obstructions beyond those imported from FDIS 1 You wish to include additional information on obstructions obtained from sources other than the reconnaissance survey e g FISS information and 2 Aknown upstream limit for a fish group that was not caused by an obstruction e g temperature channel morphology cannot be modelled by the remote sensed reach characteristics gradient order width used in the fish range rules Assuming you have confirmed that there are no fish in this stream or reach by sampling you can simulate this range limit by adding an obstruction and then using the obstruction variable as part of the distribution rule See Section 6 0 This allows you to exactly reproduce the known upstream limit of a fish groups range in a particular stream There are two ways of entering a new obstruction The first is to click on a stream in the watershed code list above the obstruction table When you do this you will be prompted as to whether you want to add this new record to the obstruction table Once it has been added you must specify the location of the obstructi
89. perations performed A dialogue box accessed from the Show Operational Tracking choice below the Utilities main menu item displays this information Fig 8 The dialogue box consists of a grid with rows for each modelling step and columns for each of the currently defined fish group When the data is first imported all the cells in the grid will be purple denoting that none of the operations have been performed As you begin to perform various modelling tasks the appropriate cells on the grid turn green and the date time stamp that the task was performed is shown If you redo a particular model step e g predict channel width all modelling procedures that depend on that operation will be out of date These steps will need to be redone and are depicted as red cells on the grid 20 May 2000 FHAT20 User s Guide ig Operation Tracking Iof x W Operation Tracking ANY_FISH FPC_FISH Strata Rules No Vis Channel Channel Width Bank Full Depth Obstructions Fish Distribution Stream Class Results Edited Fish Group Added Figure 8 The operational tracking dialogue box displays the status of modelling procedures and the date time the procedures were run You can easily navigate from one modelling procedure to another by double clicking on individual cells in the grid If you attempt to perform a task out of sequence that is perform a task that depends on a previous model step that has not been run purple cell o
90. porary directory One of these files is named SETUP EXE Run SETUP EXE to initiate the FHAT20 installation program There is a file available for download suitable for creating a disk setup version for computers not hooked up to the internet see the README TXT file for directions to this and any updates When the installation process is complete you will need to create at least one sub directory below the directory where you installed the program e g C VFHAT20 WSHD1 The FDIS database that you want to use to develop an interpretive product using FHAT20 should be copied or moved to this sub directory There should be one sub directory for each FDIS dataset that you wish to analyze A typical directory structure would be as follows DIRECTORY CONTENTS SOURCE OF CONTENTS CAFHAT20 FHAT20 program Installation program C VFHAT20 WSHD1 FDISDAT MDB Copy from WSH1 FDIS directory of WSHDI project CAFHAT20NWSHD2 FDISDAT MDB Copy from WSHD2 FDIS directory of WSHD2 project 2 2 First Time Use of FHAT20 Importing a FDIS database into FHAT20 To start FHAT20 double click on FHAT20 EXE from the Windows Explorer Select the File Open Database menu item move to the directory that you copied moved the FDIS database to e g C FHAT20 WSHD1 and load FDISDAT MDB The first time you run FHAT20 on an FDIS dataset a set of procedures will be initiated to read the data check it for errors and omissions and create a ne
91. ppropriate field from the Fish Predictions table in the main FHAT20 map window Each field name is preceded by the fish group name e g FPC_FISH ALL_FISH generically referred to as Group in the field descriptions below See Table 3 to find the text descriptions shown in the main window which correspond to these field names Group_DIST The maximum distribution range of the species within the watershed based on obstructions and user defined distribution rules 1 potentially present 0 2 Not Present 9999 could not be predicted because of missing data for input variable Group_Prob_Not Present Low Medium High The probability that fish will be not present in a reach or in low medium or high abundance classes based on habitat capability results May 2000 45 FHAT20 User s Guide Predictions are based on the pdf for unsampled reaches and sampled ones where fish were not found by electrofishing or where electrofishing was not conducted Note that the abundance classes represent habitat capability if a site is above an obstruction it may still be given a high capability value and low Prob_NotPresent value even though it is currently not accessible but could be if the obstruction was removed and does not contain fish see the Group_Prob_Presence field description For sample sites where the fish group was found by electrofishing the measured density is used to determine the appropriate capability class se
92. program and re open it Also note that if you re import the FDIS data any changes that have been made to MODEL MDE will be overwritten so it is probably a good idea to keep a copy of the modifications you make to MODEL MDE in a separate file 56 May 2000 FHAT20 User s Guide 11 0 References Anon 19982 Reconnaissance 1 20 000 fish and fish habitat inventory Standards and procedures B C Ministry of Fisheries Fisheries Inventory Section for the Resource Inventory Committee Anon 1998b Reconnaissance 1 20 000 fish and fish habitat inventory Data Forms and User Notes B C Ministry of Fisheries Fisheries Inventory Section for the Resource Inventory Committee Anon 1998c Forest Practices Code Fish stream Identification Guidebook Forest Practices Code of British Columbia Act Operational Planning Regulation James F C and C E McCulloch 1990 Multivariate analysis in ecology and systematics panacea or Pandora s box Annu Rev Ecol Syst 21 129 166 McAllister M K Pikitch E K A E Punt and R Hilborn 1994 A bayesian approach to stock assessment and harvest decisions using the sampling importance resampling algorithm Can J Fish Aquat Sci 51 2673 2687 Rice J C 1993 Forecasting abundance from habitat measures using nonparametric density estimation methods Can J Fish Aquat Sci 50 1690 1698 Silverman B W 1986 Density estimation for statistics and data analysis Chapman and Hall New York
93. r is out of date red cell FHAT20 will warn you and stop you from performing the operation If you want to disable this toggle the Operation Tracking check box off or deselect the Operational Tracking choice below the Utilities main menu item Due to the structure of the grid some operations appear fish group specific but are not Note that there are physical prediction rows channel width wetted with probability of non visible channel bank full depth for each fish group depicted on the grid but these events are not fish group specific When you compute one of these variables you will see that the appropriate cell in the grid is changed for all fish groups If you edited a model result section 9 0 and later recalculate that result the Results Edited cell will show that the edit rule is out of date The edit rule itself is not out of date however this tells you that you had previously edited the results from a particular operation have since recomputed the operation but have not re edited the result The grid alerts you to this fact but does not block you from performing additional operations that depend on the result This is logical as you may not need to edit the new result 2 5 Exporting Results Saving Maps Printing Maps Model results can be saved to the EXPORT table in MODEL MDB for analysis in other applications and display in FDIS Map To export results select the Export Results option from the
94. rds tables from FDIS Fig 2 into a table called REACH_CARDS in MODEL MDB The importing procedure then loops through all reaches in this new table and checks for errors in watershed code and reach_id fields Errors are reported in the MODEL MDB table WSCODE_LOOKUP Table 1 When errors are found FHAT20 continues with the import procedure but only includes clean reaches where no errors were detected Following the import procedure a dialogue box will appear reporting on the numbers of different types of errors that were encountered Although it is possible to continue with various modelling steps described below to develop interpretive products based on only the clean reaches it is recommended that the user correct the errors following the procedures outlined below masses assesses assess User fixes errors via changes to WSCodes_Lookup re FDIS Reach_Cards Lake_Cards Reach_Cards Lake_Cards FDIS Site Data Check network WSCodes Lookup WSCodes Lookup ws_codes reach_id s Reach caras Cards Phys_Site Screen illegal Fish_Site variable values DataCheck gt User Barr Lakes Figure 2 Schematic showing the procedure to import data from FDIS into FHAT20 May 2000 5 FHAT20 User s Guide Table 1 Description of some of the fields in the WSCODE LOOKUP table Fields not listed below are self explanatory Field Name Field Description RECID An origin
95. reach to support fish Capability can be assessed for a single species and life stage or it can be an aggregate measure for all fish within the reach Capability can rarely be measured in an absolute sense unless all fish within the reach can be caught More likely a consistent sampling approach is used to provide a relative index of fish abundance in various reaches and the relative capability among reaches is then be compared Estimation of absolute capability implicitly assumes that the reaches are fully seeded i e abundance is not limited by juvenile recruitment to the reach More realistically a relative comparison of capability among reaches must assume that all such reaches are seeded to the same extent The Fish Habitat Capability component of FHAT20 uses fish density estimates fish 100 m of sampled habitat estimated by one pass electrofishing as an index of fish habitat capability in sampled reaches This assumes that catchability is the same in all reaches The user builds various non parametric models that relate remote sensed reach attribute information such as stream width order and gradient to fish abundance estimates in reaches sampled by electrofishing These models are then applied to unsampled reaches to predict their fish habitat capability using remote sensed reach attribute information as input 7 1 Overview of Method Used to Predict Fish Capability The traditional approach to developing habitat capability mod
96. reach_id or watershed code for one of the records 2 No Parent stream from FDIS orphan With the exception of the mainstem every stream watershed code in the FDIS database must have a stream that it flows into hereafter referred to as its parent stream This error code denotes that a record a unique watershed code and reach id does not have a parent stream and that it is not the mainstem it is therefore an orphan reach 3 Stream eventually flows into a stream with no parent incomplete lineage The current stream flows into a parent stream it is not an orphan or the mainstem but there is a break in the network a non parent situation somewhere between its parent and mainstem 4 Record in PHYS_SITE with no matching record in Reach_Cards The PHYS_SITE table in MODEL MDB is populated based on records in the FDIS table S SITE CARDS When you correct a watershed code or reach id in REACH CARDS MODEL version by making changes in WSCODE LOOKUP and re running the import procedure you can create a situation where some records in PHYS SITE have no match in REACH CARDS in the MODEL database This error can also arise from incorrect entry of watershed codes or reach ids when entering site data in FDIS 5 Record in FISH SITE with no matching record in Reach Cards The FISH SITE table in MODEL MDB is populated based on watershed codes and reach ids in the FDIS table Fish Form When you corr
97. read in the CSV file you created and add these records to the FIRSTNODE table in MODEL MDB This process will add the new coordinates to the stick map eliminating some of the hanging tributaries and making the map more readable TRIM Watershed Atlas Map If a digital TRIM Watershed Atlas is available for the study area with watershed codes that correspond to those in FDISDAT MDB a method exists to bring TRIM linework into the model However as the TRIM Watershed Atlas is not available provincially and this procedure is under development it has not been included in this manual The Load Map File and Load Lakes File under the Edit Legend selection in Utilities are used in this process For information on the status of the TRIM Watershed Atlas and the procedure to import TRIM linework into the model please email the BC Fisheries Information Services Branch at gems2 gov bc ca May 2000 11 FHAT20 User s Guide 2 3 Controlling the Map Display The main FHAT20 form window consists of a digital map of the watershed that is used to display reach specific data or model results using a color code system Fig 4 Also shown is a legend depicting what the color codes represent and a set of combo boxes that are used to select various data model results to display Data results are organized into four functional variable groupings shown in the upper combo box Within each grouping there are a
98. s X is the predicted probability that the capability in the reach is less than 1 fish over the typical sample site area for that reach probability of not present 48 May 2000 FHAT20 User s Guide 8 0 FPC Stream Classification All fluvial reaches in the model database can be classified according to the FPC stream classification system Anon 1998c To perform a FPC stream classification open the FPC Stream Classification dialogue box from the Stream Classification sub menu choice under the Modelling main menu item Fig 20 Classification will depend on predicted and observed average channel widths combined with predicted and observed estimates of probability of fish presence Thus the first step in classifying the reaches is to select a fish group to be used as the basis for determining fish bearing status Since FHAT20 computes the fish group specific probability of fish presence PoP for each reach on a continuous scale 0 100 Section 7 3 Figs 18 19 you must also specify a minimum probability of presence below which a reach will be classified as being non fish bearing in status Note that specifying a value of 096 default condition provides the most conservative predictions in that any PoP value gt 0 will lead to a fish bearing classification for the reach Once the minimum PoP value has been specified in the dialogue box click on the Compute Stream Classification button to perform the calculations Click on the Save
99. s and specifications of this procedure to the acceptance levels implicit in Ministry quality assurance procedures Users are cautioned that interpreted information on this product developed for the purposes of the Forest Practices Code Act and Regulations for example stream classifications is subject to review by a statutory decision maker for the purposes of determining whether or not to approve an operational plan The statutory decision maker is typically the Ministry of Forests District Manager except in areas of joint approval where it is the Ministry of Forests District Manager and the Designated Environment Official May 2000 iii FHAT20 User s Guide Abstract The Reconnaissance Fish and Fish Habitat Inventory is a sampled based survey covering whole watersheds as defined from 1 20 000 scale maps and air photos This inventory is intended to provide information regarding fish distribution and relative abundance as well as stream reach and lake biophysical data for interpretation of habitat sensitivity and capability for fish production While the reconnaissance inventory is intended to cover whole watersheds time money and personnel are not available to survey every stream reach and lake in the watershed therefore only a subset of reaches and lakes in the watershed is sampled However forestry planning processes require the development of products showing the extent of fish distribution or stream channel widths for the entire plannin
100. sed or predicted attributes for each reach into the model and determines whether the fish group is potentially present in the reach This populates a field called Group DIST in the RES FISH table in MODEL MDB If your rule is overly restrictive the predicted range may not extend to reaches where the fish group was observed false negatives If your range rule is too liberal the predicted fish range will extend into reaches that were sampled but where the fish group was not found false positive If you are confident that the fish sampling was accurate that failure to find fish really meant that no fish ever use this reach you may make the rule more restrictive A list of false positive and false negative reaches will be shown in the dialogue box after a rule has been run If you want to review the physical characteristics of these 38 May 2000 FHAT20 User s Guide reaches simply double click on them in the grids to highlight their location on the map You then have the option of double clicking on the actual reach on the map to see the attributes of interest Fish Range Rules Ea Fish Groups Available JALL_FISH Rule Name 5 ampleRule1 Run Rule Avalible Variables Double Click on list to Add Variable to the Rule Upstream Length Upstream Length Upstream of an Obstruction Fish Occurence Upstream Confinement Code Upstream Elevation Downstream Elevation Rule Results 07 teaches of ggg within distribut
101. series of variables which are shown in the combo box labeled Variables each variable generally corresponds to a unique field in tables within MODEL MDB Table 3 provides a listing of the functional data groupings and their fields and what each field represents To view results for any field simply select the appropriate grouping and field from the combo boxes im 1 20 000 Fish amp Fish Habitat Assessment Tool March 31 2000 File Modelling Utilities r Map Controls Tables Physical Characteristics Fields in T able Order Fish Groups Associated with FISH Variables ALL FISH z Missing gt lt 1 Ei gt lt 2 EE 2and 3 gt Gand lt 4 gt 4 lt 5 Draw Full Map Reach Breaks OverLay Second Layer Show apela Legend Figure 4 The main FHAT20 dialogue box used to display and query reach specific data and model results 33 May 2000 FHAT20 User s Guide The mapping display has a number of features to facilitate data review and analysis Point and Click to get Reach Attribute Information Double click on any reach with the left mouse button to bring up the Search Results dialogue box Fig 5 A table shows the watershed code and reach id for the reach you selected and the value for the current layer on the map You can then find the value for other reach attributes by selecting the variable group from the combo list box
102. shed codes or reach ids in REACH CARDS check all records with values in the TABLE TYPE field in WSCODE LOOKUP If there are errors follow the procedure specified in the previous paragraph add new records to WSCODE LOOKUP Never correct incomplete lineage errors they can only be eliminated by correcting orphan errors and are only provided in WSCODE LOOKUP in case you run the modelling procedures without completing all corrections thereby providing a catalogue of reaches missing from the analysis Note it is possible to run the modelling procedures with errors in the watershed codes and reach ids however none of these reaches will be included in the model analysis At the very least you have a record of these reaches in the WSCODE LOOKUP table all records with the FIXED field FALSE However predictions for other reaches still included in the analysis may be effected by the deletion of the problem reaches due to network connectivity aspects of the modelling Thus running the model with missing reaches is not recommended especially if they are parents and have reaches flowing into them 1 not first order headwater reaches May 2000 9 FHAT20 User s Guide Errors in the watershed code that do not result in orphan streams are not detected during the FHAT20 import procedure There are two types of such errors that could affect modelling results Watershed Code Does Not Reflect Position of Confluence Along th
103. statistics button to copy the summary statistics from the stream classification the table in the dialogue box to the SclassStats table in MODEL MDB i FPC Stream Classification x Pick Species Group ALL FISH Reaches with probability of fio will classified as non fish presence less than bearing in status FPC Stream Reaches Stream Reserve Zone Management Zone Total Riparian Class Length Km amp rea Hal Area Management Area Hal 58s B 00 00 00 0 0 Figure 20 The stream classification dialogue box is used to classify reaches into S1 S6 FPC stream classes Results for the stream classification are presented in probabilistic and deterministic formats Fig 21 The probabilistic approach computes the probability of each FPC stream class Probability of S 6 occurring in any reach rather than generating a single most likely class May 2000 49 FHAT20 User s Guide This probability is generated by multiply the fish group probability of presence PoP for a reach by the probability of the appropriate width class for S1 S4 reaches and 1 PoP times the probability of the appropriate width class for 55 96 reaches In all cases 51 56 probabilities sum to 10096 The highest probability across the six stream classes is found and saved to the Most Probable Stream Class variable for each reach For sampled reaches the observed channel width is used to set one of the 51 54 and 55 56 width class probabilities
104. t capability calculations are dependent on the fish range you will also have to rerun the capability calculations e If you rerun habitat capability for a fish group you will need to rerun the stream classification if it was run using the same fish group The habitat capability modelling contributes to the calculation of the probability of fish presence variable which is used in the stream classification procedure e If you define a new fish grouping you must complete steps 4 6 and possibly 7 for this group e If you modify the original data in FDIS you must re import the data and reprocess all the results May 2000 19 FHAT20 User s Guide Fo area Stratification Rules m 7 a 4 Y 4 Non Visible Wetted Width Obstructions Bankfull Depth Channel Width Channel _ n a ES Y Y hae gt Fish Group Range U L Fish Group Habitat Capability Stream Classification Figure 7 Schematic showing the relationship among modelling steps in FHAT20 Solid lines denote fixed relationships e g channel width is used to predict stream class while dashed lines show relationships that depend on whether particular variables are used in later modelling steps channel width depth may be used in predicting fish range but it is not mandatory To assist you in following the correct order in the modelling procedures FHAT20 keeps track of the sequence and time stamps of all modelling o
105. t the watershed code and reach id in USER_FISH_SITE so that it corresponds to a record in Reach Cards and rerun the Process Fish Site Data procedure May 2000 33 FHAT20 User s Guide 5 0 Obstructions Obstructions are used by FHAT20 to limit the distribution of a fish group in the watershed through the process of running a fish range rule Section 6 0 All obstructions used in FHAT20 are stored in a table called USER_BARR in MODEL MDB When FDIS data are first imported into FHAT20 USER BARR is populated with all the records from the FEATURE table in FDISDAT MDB thus it will contain features that are not obstructions and you can delete these from the table if you wish You can examine these records 1 the contents of USER BARR by opening the Obstruction Editor dialogue box accessed from the Obstructions option under the Modelling main menu item Fig 14 A table will show each feature whose location is identified by a watershed code reach id The feature measure distance upstream from the confluence is also computed by assuming that the feature is located at the top of the reach it resides in the measure value can be adjusted to reflect its actual position in the reach as described below is Obstruction Editor Add Obstruction From List 1 Double Click on Watershed Code in List of Available Codes 2 Edit Reach Id Measure in the Obstructions Grid 3 Edit the Obstruction to make
106. t to 100 all other classes set to 0 For sampled sites were the fish group was found but not by electrofishing the pdf is used to estimate the probability of abundance classes but the _Prob_Low class is the sum of the _Prob_Low value and whatever probability was estimated for the Not Present class _Prob_NotPresent and the Prob_NotPresent class is then set to zero since fish were sampled in this reach we know Prob_NotPresent must equal zero Group_Most_Probable_Class The most likely capability class This is computed by finding the maximum probability across the not present low medium and high categories Group_Prob_Presence Probability of fish group presence This prediction combines habitat capability and fish range results Probability of presence is used in conjunction with channel width to predict FPC stream class as described in Section 8 0 below If the reach is beyond the fish groups range in the watershed Group_DIST 0 then this value is 0 This applies to both stream and lake reaches If the reach is within the group s range and downstream of a sampled reach where this fish group was found then the value is 100 If the reach is upstream of such a point but within the fish group distribution range then the value 100 Group_Prob_NotPresent as predicted by the pdf Figs 18 and 19 The only exceptions to this rule are for e Fluvial reaches that are inlets or outlets to lakes within the fish groups range In this c
107. table in Access and enter the watershed code and reach ids for the new sites as well as identifiers for the Site ID The latter field can be any numeric value make one up if a Site ID doesn t exist but make sure to enter unique values for different sites in the same reach The following fields must be populated Species Use FDIS codes Stage or Age Populate one or both of these fields using FDIS codes TOTALNO Total number of fish caught METH Type of gear employed Use FDIS codes EFFORT of seconds of electrofishing Populate with missing value 9999 if other gear type is employed AREA m of area electrofished Set to missing value for gear types other than electrofishing METHOD NUMBER of times a particular type of gear was deployed With the exception of the EFFORT and AREA fields all other fields correspond to FDIS standards Regarding METHOD NUMBER if 10 minnow traps were deployed at a site and only one species age stage was caught there should be 10 records with METH MT with separate records for METHOD NUMBER 1 10 If two electrofishing locations were fished at a site each location would be distinguished by a different METHOD_NUMBER If more than one electrofishing pass was completed on a particular date only enter data from the first pass do not enter the average or the total across passes If the site was resampled on a different date enter separate records for each date the data will be averaged when
108. ted to distributions taken from two habitat quality classes as indexed by stream gradient Dashed lines denote low medium and high abundance classes The pdf estimation approach is ideally suited for data intensive situations such as the Reconnaissance 1 20 000 Fish and Fish Habitat inventory Predictions and uncertainty are totally dependent on the input data If there are strong relationships between habitat attributes and fish capability in the sample data the kernel estimation will produce tighter narrower pdfs that are noticeably different between reaches with different habitat attributes If the relationship between fish capability and habitat attributes is weak often the case then this will be reflected by wide pdfs high uncertainty in capability which look more or less the same in all reaches 7 1 1 Prediction of Capability Classes based on a PDF A probability density function is really just a plot of the probabilities of a range of fish densities for a particular reach e g 0 600 fish 100 m in Fig 16b This pdf must be summarized into simpler terms so that results for all reaches can be examined spatially on a map and summarized in tabular statistical formats One way of summarizing a pdf is to break the distribution into broader fish abundance classes and then estimate the probability for each of these broader classes In FHAT20 we divide each pdf into low 5 fish 100 m medium 5 20 fish 100 m and high 220 fish 100
109. the bin breakpoints by editing the values in the breakpoint text boxes If you wish to edit the legend for another variable simply select it from the list If you are editing the legend for a variable currently displayed on the map you must click on the button labelled Apply Legend to see the changes Note that any changes you make to the legend are saved to a file called MODEL MDE and effect the display in all subsequent sessions iw Legend Editor Data Fields Mast Probable Stream Class Categorical Map Scale Fish Groups ALL FISH Number of Strata Color Palette Lower Range Upper Range l Legend Controls Ramp Breaks Set Upper and Lower Ranges Choose Map File Load Map File Load Lakes File Lakes Visible Break paints TIT Apply Legend Figure 6 The legend editor dialogue box is used to edit the map display of data and predicted variables i8 May 2000 FHAT20 User s Guide 2 4 General Modelling Procedure A series of specific steps must be followed to develop standard interpretative products of fish habitat distribution and capability using FHAT20 As you develop models and apply them modelling rules and results are saved to various tables in the FHAT20 database MODEL MDB The relationship between FDIS data various modelling steps and the FHAT20 database is shown in Figure 1 There are seven basic modelling steps that must be performed in a spe
110. ugh the densities are the same The bottom line is that under equal densities the probability of not present is inversely proportional to the size of the sample site 7 2 Using the Fish Habitat Capability Dialogue Box The Fish Habitat Capability Dialogue box can be accessed from Fish Habitat Capability sub menu choice below the Modelling main menu Fig 17 To compute pdfs of fish capability follow these steps 1 Select a fish group from the combo box at the top of the dialogue box By default when the dialogue box loads the currently active fish group is displayed Note that before you can select a fish group to compute capability pdfs you must first predict the fish group s range within the watershed via the Fish Range dialogue box When you select a fish group a reference data set used to compute the pdf is populated based on all sites sampled for fish within the fish group s watershed range The number of sites included in this reference set is shown in a yellow box labeled of reference sample sites May 2000 43 FHAT20 User s Guide 44 Select habitat variables to include in the multivariate kernel estimation variables from the table immediately below the fish group combo box You can choose multiple habitat variables To select the variable click on the In Model column to toggle the X on variable selected or off For each habitat variable the table shows the habitat value for the reach that you wil
111. ute its pdf An alternate way to select a reach is to close the Capability dialogue box to expose the map on the main window Double click on the desired reach This will bring up a dialogue box showing the reaches attributes If you close this dialogue box and open the Capability box again you will note that the reach that you selected on the map will now be the currently selected reach in the Capability dialogue box Its pdf will be displayed as well as its pdf statistics The pdf computed for a selected reach will be shown in the graphic in the lower left portion of the dialogue box The pdf graphic shows the probability 0 10096 of different fish abundance classes for a continuous range of fish abundance classes 100 You adjust the y and x axis maxima of the graph by editing the values in the text boxes below the graphic followed by the hitting the return key The program automatically computes the probability of different fish abundance classes based on the pdf 1 it computes the area under the pdf curve within fixed ranges on the x axis as in Fig 16b The fish densities that define each abundance class and their probabilities are shown in a frame immediately above the pdf graphic labelled Fish Abundance Classes If you want to alter the abundance class breakpoints e g gt 20 fish 100 m is the breakpoint for the High Abundance class simply edit the values in the appropriate text boxes and
112. vation Confinement Code Confinement of reach Reach_Cards Conf_code Channel Pattern Channel pattern of reach Reach_Cards Cptn_Code Features Location of Features from FDIS Features table or user defined User_Barr Reach is a Wetland Denotes whether reach is a wetland Reach_Cards Wetland Reach is U S of a Wetland Denotes which reaches are upstream of wetlands Reach_Cards USofWetland Reach is a Lake Denotes whether the reach has a corresponding record in the Lakes table Reach_Cards IsLake Reach is Lake headed Denotes whether a reach has a lake upstream Network IsLakeHeaded Reach flows into Lake Denotes whether a reach is an inflow stream to a lake Network ReachFlowsIntoLake Upstream Length Length km of stream upstream from the Network Uplen downstream end of each reach Maximum Downstream Gradient The maximum gradient between the reach and the Network Maxdsgrade first reach at the most downstream end of the watershed Parent Order The Strahler order of the parent stream of the first reach of its tributary Network Parent_order Sampled for Physical Fish Data Was the reach sampled for physical parameters or fish True False Network Sampled_Phys Network Sampled_Fish Table 3 Con t Group Name Variable Name Description Table Field Physical Predictions Predicted Channel Width
113. verlaid on the Continuous PoP map to generate a deterministic map of continuous fish presence termed FPC Fish Present The final stream classification termed the FPC Stream Class is computed as either the most likely width class among 1 54 ranges if the reach is fish bearing in status FPC Fish Present True or the most likely width class among 55 56 ranges if the reach is not fish bearing in status FPC Fish Present False All results are saved to the RES_PHYS table copied to the EXPORT table when results are exported and can be viewed on the map 50 May 2000 Probability of Fish Presence PoP PoP 1 PoP PoP must be stable or increase in a downstream direction from any reach Continuous Probability of Fish Presence Probability of Different Channel Probability of Each FPC Width Classes CWx Stream Class s EEE EE Most Probable Stream Class Find most likely channel width class Uncertainty in Stream Class lt s d User defined minimum PoP FPC Fish Present True False FPC Stream Class Figure 21 Schematic showing how stream classification predictions highlighted in bold are computed in FHAT20 FHAT20 User s Guide 50 Figure 22 Schematic showing how continuous probability of fish presence values not enclosed in boxes is computed from the probability of fish presence variable values enclosed in boxes calculated in fish habitat capability
114. w Access database MODEL MDB which will be used for all future modelling sessions In subsequent sessions you will still have to move to the appropriate subdirectory and load FDISDAT MDB but if you had previously created MODEL MDB successfully FHAT20 will actually open MODEL MDB MODEL MDB contains imported FDIS data as well as modelling rules and results that you develop May 2000 3 FHAT20 User s Guide FHAT20 uses 7 tables from FDISDAT MDB to create MODEL MDB database REACH CARDS S SITE CARDS FEATURE FISH FORM FISH GEAR SPECS FISH NET SPECS and FISH EF SPECS Fig 1 MODEL MDB contains Model Fish Data i Build Fish i Group_Rules Distribution c a subset of information from FDIS used for modelling new variables computed from information in FDIS e g total stream length upstream of each reach a set of default rules used to define fish groupings and model fish range in the watersheds predicted results of channel width wetted width bankfull depth probability of non visible channels fish range and capability computed from previous modelling sessions Fish_Summary Corset watershed 3 Reach Cards S Site Cards codes and reach Fish iBonm Saag idm Fish_Gear_Specs 5 WsCode_Lookup Lake_Cards Fish_Net_Specs Fish EF Specs ty Adda i Reach Cards ei S Phys Site Define Stratification Models to Predict Width Ru
115. width wetted width and bankfull depth for all unsampled reaches and the probability that these reaches are non visible channels 3 4 Channel Width Wetted Width and Bankfull Depth Predictions FHAT20 predicts channel and wetted width and bankfull depth in unsampled reaches as a function of their upstream drainage area and empirical relationships developed from sampled reaches When FDISDAT MDB is first imported into FHAT20 the total length of stream upstream for each reach is computed by summing the LENGTH field in REACH CARDS Total upstream length is strongly correlated with drainage area which is a good predictor of some channel characteristics in areas of similar unit discharge m sec km The ratio of stream length to drainage area km stream km drainage area should be relatively consistent within an inventory area and will be a function of rainfall surficial geology and the detail that was used to represent stream lines on the TRIM maps Parameters of the power function predicting channel width wetted width or bankfull depth Y as a function of upstream length UPLEN Y a UPLEN b are estimated from the sample data where widths and depths have been measured in the field and applied to unsampled reaches The model s is fit by a least squares procedure on log transformed variables Rather than estimate a single relationship for each variable channel width or wetted width etc FHAT20 allows the user to develop separat
116. y y axis of different fish densities classes x axis The solid lines shown in Figure 16b are pdfs of fish abundance in high gradient lt 1 and low gradient 5 796 quality habitat Perhaps the biggest advantage of the use of a pdf to estimate habitat capability in this application is that it does a good job of capturing the uncertainty in capability predictions FHAT20 uses multivariate kernel density estimation to compute the pdf of fish capability for every unsampled reach in the FDIS dataset based on a comparison of their attributes e g gradient order width elevation relative to those in the sampled reaches where fish abundance estimates were measured The simplest way of constructing a pdf for an unsampled reach would be to assemble density estimates from all sampled sites within the fish groups watershed range and plot this as a histogram The kernel estimation method would simply draw a smoothed curve through this histogram e g Fig 16b and we could compute various statistics of interest e g median value 95 confidence limits from this distribution If we employed this approach we are essentially saying that the capability of all reaches in the watershed are identical Clearly we can do better than this We know intuitively for example that abundance in a lower gradient stream tributary to a mainstem will tend to be higher for many species than abundance in a steep gradient stream in the headwaters of the watershed So if we
117. y value that one of the four abundance categories is appropriate for a given site May 2000 61
118. you have defined the fish group click on the button labelled Process Records Rules defining a fish group are saved in the GROUP_RULES table in MODEL MDB e It is important to remember that if you add a new fish group you must go to the Obstruction dialogue box section 5 0 and determine which entries in the obstruction table are migration barriers for this fish group before you can run fish range and capability models May 2000 31 FHAT20 User s Guide ig Build Fish Grouping For Modeling Current Fish Groups FPC_FISH Run and Save Rule Delete Rule All Values Selected Values for Group Rule Age eda gt l Remove x or gt Le ALL STAGES TES a aur Juvenile Remove lt ALL SPECIES Kokanee Add gt EE RB Rainbow Trout Remove c Remove c HESULTS Rule That Makes Up This Fish Grouping Grouping FPC FISH Species AMD FISH SITE SPECIES RB OR FISH SITE SPECIES A amp CT OR FISH_SITE SPECIES C Age Es Figure 13 The fish group dialogue box is used to review and create new fish groups used in FHAT20 modelling procedures 32 May 2000 FHAT20 User s Guide 4 1 Adding Fish Site Data Additional fish data not contained in FDIS can easily be included in the FHAT20 model database to improve the fish range capability and stream classification predictions Open the USER FISH SITE

User's Guide to FHAT20 - Ministry of Forests, Lands & Natural

Contents

Download Pdf Manuals

Related Search

Related Contents

User's Guide to FHAT20 - Ministry of Forests, Lands &amp; Natural

Contents

Download Pdf Manuals

Related Search

Related Contents

User's Guide to FHAT20 - Ministry of Forests, Lands & Natural