Litter Analyst 2.0
Contents
1. not a number blanks e The data of items with the code 200 until 208 and 210 after 1 January 2010 are set to not anumber 2 Remove pollutants and faeces e The time series of the pollutants with the codes 108 until 111 are removed e The time series of the faeces with the codes 121 207 and 208 are removed 3 Make totals counts of all items code 400 of sources code 401 until 405 and materials code 406 until 415 4 Make clusters There are six clusters e Nets and ropes 300 contains the items with the codes 31 32 115 116 200 and 201 e Plastic polystyrene pieces lt 50 cm 301 contains the items with the codes 46 117 and 202 e All cartons tetrapacks 302 contains the items with the codes 62 118 and 204 e Other textiles 303 contains the items with the codes 59 and 210 e All Gloves 304 contains the items with the codes 25 113 and 203 e All metal oildrums 305 contains the items with the codes 84 205 and 206 5 Remove items included in the clusters in 4 6 Make user defined clusters These are defined in the sheet User defined Clusters in the file Litter Analyst_config xlsx in the directory config e Maximum number of clusters is 24 until column Z e UseaOor 1 for an item ina cluster e Anitem can only be used in one cluster e These clusters get the item numbers 500 and higher 7 Check for double survey dates If double survey dates occur only data2. 5 1 EVAIUAEION LADIES mamake abort E 11 5 1 1 SOLE S cecaveste sain vonelan a aekeked aeta iseen storten ennen etos 11 5 1 2 MELO eee ei natekenen eaten 12 5 1 3 Group 1 Summary of results from top X H items aaan oenen eereenneervnnen 12 5 1 4 Group 2 Summary of statistically significant trends aaneen oenen 13 5 1 5 Group 3 Summary of results from beach total COUNTS ccceccccesececeeecseeneceeenseeenees 14 5 1 6 Group 4 Summary of trends slopes of top X H items aaan eenennenne erven 15 517 Evaluation table of items versus those of sources and materials 15 5 2 End PO a soars ances east ets aed estas oe eee Seas adt en niee afmeten tibet Anne 16 5 3 End al EN 17 5 4 Trend DOM OU NN 18 55 Tend Judgement NSOE IM rentestanden deelen Bat nekineeee denn tee enebendenbdeaid 19 5 6 Trend analysis FOSU Seerne en E dna eeen mbo adsense nen 20 5 7 Dataden IN aple E T E N a N 21 5 8 Datasenes 0 ans denten eden eben eee ae ee eee 22 5 9 Vea DODO wter etc eeeteee denon Ae toate dota acnce annette an cesavaeohe eee tanesene 23 BO ASD enteren entente eeen ed beenen neen tadan acess 24 T MNRETEREMCOS 2 wd nternet kde hahhaha hen dedekebkdeenedinde 25 ADD ENE GEREN Ce aE EE teon oben Ea ee Cee botte ee ee tee 26 AMO lcastat 1 Manual of Litter Analyst 2 0 1 Introduction Litter Analyst is a standalone Windows program for statistical analysis of beach litter data It is developed
3. in the Matlab environment To enable the presentation of results of this program Excel 2007 or a later Excel version must be installed Litter Analyst can read OSPAR csv files of litter data execute a data cleanup and export them to OSPAR csv files or Excel spreadsheet files Furthermore it can perform statistical trend analysis using the Mann Kendall test and the Wilcoxon ranksum test It presents and exports evaluation tables of items sources and materials various plots of the litter data and trend results The program can be downloaded from http www amo nl com wordpress software download litter analyst More backgrounds can be found in the following report Baggelaar P K and Van der Meulen E C J 2014 Evaluation and fine tuning of a procedure for statistical analysis of beach litter data lcastat AMO October 30 2014 43 pages This report is available in the Help menu of Litter Analyst AMO Icastat 2 Manual of Litter Analyst 2 0 2 Input of Litter Analyst 2 1 Read and cleanup OSPAR csv file Once the program is started the main screen appears Choose File and then Read data file to read csv files The file may be a national or a regional file The national files can be downloaded from the OSPAR database The following 15 steps correct prepare and aggregate the litter data of a single csv file 1 Clean up e The data of items with the code 31 32 46 62 84 112 until 121 before 1 January 2010 are set to
4. of testing on a monotonic trend using the Mann Kendall test The p value is the two sided probability of observing this trend or a larger trend if the null hypothesis of no trend is true We consider it acceptable to use a confidence level of 95 for the testing Thus if the p value is less than 0 05 than we can say with 95 confidence that there is a Statistically significant trend If an estimated trend slope is negative that cell is green and if it is positive the cell is orange If the p value is less than 0 05 indicating a statistically significant trend that matrix cell is grey In evaluating the trend results one should bear in mind that the overall risk of detecting one or more Statistically significant trends when in reality no trends exist increases with the number of time AMO Icastat 12 Manual of Litter Analyst 2 0 series that is analyzed on trend If only one time series is analyzed on trend this risk is 5 because we test with 95 confidence However if two time series are analyzed on trend this risk increases to 9 8 according to the following formula Overall risc 1 95 where n is the number of time series that is analyzed on trend For 10 time series the overall risk is 40 1 for 15 time series it is 53 7 and for 20 time series it is 64 2 If for example the top X H list contains 15 items then the risk of detecting one or more statistically significant trends when in reality no trends exist is 53 7 Therefor
5. the estimated trend is Statistically significant it is determined by the absolute value of the ratio of trend and median and this judgement is Very small trend Small trend Moderate trend Large trend or Very large trend see figure 5 6 Figure 5 6 The judgement of a trend increase or decrease depends on the statistical significance of the trend and on the absolute value of the ratio of trend and median Trend significant trend medianl Judgement No gt 1 lt 5 Yes gt 5 lt 10 gt 10 lt 25 Very large trend AMO lcastat 20 Manual of Litter Analyst 2 0 5 7 Data density table Shown is an Excel spreadsheet with the number of surveys per year of all items of the beaches The file is saved in the temporary directory Figure 5 7 shows a part of the data density table of Belgium Figure 5 7 A part of the data density table of Belgium Koksijde St Andr Koksijde St Andr Koksijde St Andr Koksijde St Andr Koksijde St Andr Koksijde St Andr Koksijde St Andr Explanation of the parameters in the data density table Period Period of the trend analysis and data series of litter item Beach Name or code of monitoring beach or beaches Litter item Name or code of analyzed parameter litter item s Start data series Start of data series of litter item End data series End of data series of litter item Surveys Number of time series values Tot
6. 12 2013 A judgement can be a deterioration no trend or an improvement Figure 5 5 Trend judgement histogram of total counts materials and sources of the four Dutch beaches of 01 01 2002 31 12 2013 File Settings Explore He Trend judgement total counts materials and sources The Netherlands 01 01 2002 31 12 2013 GE Deterioration __ No trend EE improvement It is possible to show the trend judgement histogram of 1 All clusters 2 Materials 3 Sources 4 Total counts materials and sources On the X axis the items are shown with the number of beaches AMO Icastat 19 Manual of Litter Analyst 2 0 5 6 Trend analysis results The trend analysis results of all selected beaches are shown in an Excel spreadsheet Explanation of parameters in the trend results Maximum Maximum of the values of the data series Standard deviation Standard deviation of the values of the data series Coefficient of variation Ratio of the standard deviation to the average in Test Statistical test Mann Kendall MK Intercept Estimated intercept of the trend line Trend year Magnitude of the estimated trend expressed in counts per year Trend median Ratio of the estimated trend and the estimated median of the values of the time series Significant trend Statistical significance of the trend Yes or No Judgement Judgement of the trend This judgement is No trend if the estimated trend is statistically not significant and if
7. In presenting the results of aggregated data it must be made very clear that they can be only be regarded as representative for the group of beaches that are monitored and not for the population of national or regional beaches where these beaches come from because no probability sampling was used in selecting them An objective translation of the results of this group to a higher spatial level is only possible if beaches are selected with probability sampling Top X H selection of items The user can determine the Top X H list of aggregated items by choosing different systems for the ranking of items based on 1 The counts of items The ranking of items is based on the weighted average of the counts of an item in descending order the item with most counts has rank 1 2 The ranking of the scores of items The ranking of items is based on the sum of the scores of an item in descending order See the Appendix for the scoring system or see Conclusions of the Breakout Working Group on Assessment Criteria for OSPAR beach litter data Vigo 11 amp 12 November 2014 and Additional concept specifications Litter Analyst If there are items with the same score their weighted counts will determine the ranking of the items 3 The predefined item list If there are items with the same rank in this list their weighted counts will determine the ranking of the items Figure 4 4 The dialogue window to set the ranking system of items in the top X H
8. Litter Analyst 2 0 User Manual Litter Analyst 2 0 beta version until 01 07 2015 C Litter Analyst data The Netherlands csv File Settings Explore Help i ank i br TRE Litter Anatys E yAVersion April 15 2015 2 0 01 Litter Analyst 2 0 User Manual Client Rijkswaterstaat Dutch Ministry of Infrastructure and the Environment Contact dr Willem van Loon Authors ir Eit C J van der Meulen AMO drs Paul K Baggelaar castat April 15 2015 CONTENTS t Modu ON E gate ome antee card ueaaeca ees iota taiedoustanaeetaueades 2 2 MADOUIOR EIRCOM AM GIVSE cisaiacsi deseo oted athadeds lta ateda hedepncahors neten eb neee eea 3 2 1 R ead and cleanup OSPAR CSV Tle wsicinedsdevcdecitcwnsndvetinacodicmatne dee wei N emer 3 2 2 Make regional TINGS sanesna S N 4 3 lt OULDUE OF LIRTEPANAIVSE sss setters et reken tae eae ddr coast nator ee E ON 5 A SENES ONERE ANIV S ravenstein neet Ghaecaln cided EN 6 4 1 Default directories de r S hebhes set basdetelaneteteaken 6 4 2 Periods to analyze a eee sa nee ee 6 4 3 Aggregallonm condilion ienest a a a 7 4 4 TOP X H selection OF IR CIWS sscccsdcenesdeveed nn a a a a nalaten 8 4 5 kene Se o gt Cal mW oj aaea a a E 9 4 6 Font Ch aD MNCS andante eeb E E E cnt Seve deeb agai cosmaaart ines 9 4 7 Paden setings suraa da mwtedoteut a a oee der dbadandhnasea elen 10 gt JEXPIOFING datawith Litter ANALYSE tso tavern aeanae assis tage iarcteou staan aunties 11
9. ailability of at least 80 If for a specific quarter the data of an item is missing from more than 20 of the beaches the aggregated count of that item is set to a missing value for that quarter 3 Required availability of at least 75 If for a specific quarter the data of an item is missing from more than 25 of the beaches the aggregated count of that item is set to a missing value for that quarter Figure 4 3 The dialogue window to set the aggregation condition of the required availability of survey data 75 of survey data available 80 of survey data available 100 of survey data available Ok AMO lcastat 7 Manual of Litter Analyst 2 0 4 4 If the required availability is 100 this can lead to sparse time series with many missing values This can inhibit trend analysis due to an inhomogeneous distribution of the data In such cases it can be considered to set the required availability to 80 or 75 However take care in lowering the required availability as it may lead to biased results It is highly recommended to exclude beaches with a low data density in national and regional aggregation because they will strongly reduce the information content of the aggregated data Their data should be removed manually from the national csv file and their names should be removed or their weights should be set to zero in the configuration file worksheet Region Beaches Be cautious in presenting results of aggregated data
10. al counts Total of litter items found in that period Trend analysis Are all criteria for trend analysis fulfilled Zero Number of zero counts in the time series Zero Percentage of the zero counts in the time series Year Number of years without missing values in the time series Min years Is the length of the time series at least 4 5 years Min measurements Are there at least 5 data series values lt 50 zero Is percentage data series values equal to zero less than 50 lt 80 zero Is percentage data series values equal to zero less than 80 Homogeneous Are the data series values equally distributed over the period Unique counts Number of unique counts in the data series values Zero slope Number of zero slopes in the Mann Kendall test Zero slope Percentage of zero slopes in the Mann Kendall test Rank nr top X Rank number in the top X list of litter items on basis of their total counts Cum counts Percentage of the total counts of all items in the top X list with equal or higher Rank number in compare to all counts of all items in that period Top X counts Is the Cum counts lt the minimal total of percentage of total counts of all items of the top X list see 4 5 Total scores Scores given to the litter item on basis of the ranking system Rank nr top X scores Rank number in the top X list of litter item on basis of their total scores Cum scores See Cum counts accept that top X list is now bas
11. d item in two of the beaches score 18 the third most counted item on one beach score 8 6 on one beach score 5 and 7 on one beach score 4 In total the item receives a score of 125 and is ranked 1 for the aggregation of the 14 beaches Figure A1 An example of a simple ranking system applied to a situation of 14 beaches Each cell contains the score for 1 place on a beach 2 place on a beach etc Total Item OSPARID En score Nets and Nets and ropes 300 Nets and ropes 300 a an ES EEN Pancrease sell a nnn EEC MERKE Plastic Crisp as TT lE 0 E Plastic Other 48 TL Plastic Small_bags 3 7 Paste masapas 7 l sl ep EN EA EN EC Phstic Foam_sporselas Tj 4 3i 6 1 paseos el ele e Pasero el 2 pesters l 5 3 2 2 1 Phstic Strapping nds Bj 4 3 2 1 10 O E Ee i E AMO Icastat 26 Manual of Litter Analyst 2 0
12. e one should be cautious with the interpretation of the Statistically significant trends Fourth subgroup The fourth subgroup shows the number of items in the top X H list and the number of surveys in the analyzed period 5 1 4 Group 2 Summary of statistically significant trends First subgroup Number of statistically significant trends slopes Second subgroup This subgroup shows three versions of the ITI Item Trend Index for the three periods The ITI s are presented in the sequence ITT sum of slopes ITlhweighted average of slopes and ITlhweighted average of trend signs Each of these ITI s integrates the information about statistically significant developments of individual items The TlWweignted average of trend signs highlights the general direction of change of the items that have Statistical significant trends m n _ i A etende oftrendsigns _ 2 sign s i 1 where m is the total number of items the top X H items plus all the other items i the index of the item i 1 2 m n the counts of item in that assessment period N the total counts over all items in that assessment period s the estimated magnitude of the trend slope and sign s the trend sign If the trend is statistically significant the trend sign is set to 1 for s gt 0 and to 1 for s lt 0 And if the trend is not statistically significant or is not estimated this is the case for all the items that are not in the top X H list the tre
13. ed on total scores Top X scores Is the Cum scores lt the minimal total of percentage of total counts of all items of the top X list see 4 5 Rank nr top X predefined Rank number in the predefined top X list Cum predefined See Cum counts in which the top X list is predefined Top X predefined Is the Cum predefined lt the minimal total of percentage of total counts of all items of the top X list see 4 5 AMO Icastat 21 Manual of Litter Analyst 2 0 5 8 Harm list The Rank numbers of the harmful items with the fraction of counts more than 1 of the total counts Top X Is the Cum counts lt the minimal total of percentage of total counts of all items of the top X list see 4 5 of the chosen ranking system counts scores or predefined 2001 Number of surveys in a year Data series plot The user can choose between all cleanup items and shown is a data series plot In figure 5 8 we show an example of the data series plot of Nets and Ropes item code 300 of beach Sylt in Germany Figure 5 8 Data series plot of Nets and Ropes 300 of beach Sylt in Germany File Options Explore Help Data series plot Nets and ropes 300 Juist AMO lcastat 22 Manual of Litter Analyst 2 0 5 9 Year boxplot The user can choose between all cleanup items and shown is a year boxplot In figure 5 9 we show an example of the year boxplot of All Gloves item code 304
14. ems evaluation table This is because no top X H list is used but all five categories of sources or ten categories of materials are used instead For completeness the evaluation tables of sources and materials also present the summary of results of the beach total counts group 3 However this summary is exactly the same for all three tables because the beach total counts are the same at the levels of items sources and materials AMO Icastat 15 Manual of Litter Analyst 2 0 5 2 Trend plot The user can choose between all items of the selected beaches and presented is a trend plot If more than one item is chosen all selected trend plots are presented in a Word document In figure 5 2 we show an example of the trend plot of Crisp item code 19 of beach in Bergen in The Netherlands Figure 5 2 Trend plot Crisp 19 of beach Bergen in The Netherlands File Options Explore Help Trendplot Plastic Crisp 19 Bergen data values time series values Lowess trend line Result MK test is Judgement no trend The estimated trend 1 0909 units per year AMO lcastat 16 Manual of Litter Analyst 2 0 5 3 Trend palet A trend palet of the trend analysis is presented in different sheets of an Excel spreadsheet In figure 5 3 we show a part of the relative trend judgement of the aggregation of the Dutch beaches Bergen Noordwijk Terschelling and Veere Figure 5 3 Trend palet of the first 5 items in three periods
15. eriod The top X H items are sorted based on one of the three ranking systems see 4 4 The median is a better measure of central location than the arithmetic average in the case of a non symmetrical distribution Because non symmetrical distributions are predominant for counts of beach litter items it is advisable to use the median instead of the arithmetic average for evaluations Third subgroup If the number and temporal distribution of the survey data fulfill the criteria for trend analysis see below this subgroup shows the results of the trend analysis of the top X H items for the three periods otherwise the corresponding cells are left empty The following criteria for trend analysis are used e the series length is at least 4 5 years period between first and last measurement e the series contains at least 5 values e there is a more or less homogeneous distribution of the data in time o fora six years period the series has at least one value in each of the three two years blocks of the six years period o fora twelve years period the series has at least one value in each of the three four years blocks If these criteria are fulfilled the magnitudes slopes and statistical significances p values of the estimated trends of the top X H items are shown otherwise the corresponding cells are left empty The trend magnitude slope expressed in counts year is estimated with the Theil Sen estimator and the p value is the result
16. file The trend judgement histograms to a Word file A cleaned up OSPAR csv file after performing the steps 1 5 and 7 of 2 1 An Excel spreadsheet file including new clusters and item counts or a tai file for the input of Trendanalist a program for performing trend analysis after performing steps 1 I 7 of 2 1 The data density table to an Excel spreadsheet file The trend analysis results to an Excel spreadsheet file AMO lcastat 5 Manual of Litter Analyst 2 0 4 Settings of Litter Analyst The menu Settings offers the following possibilities e Default directories e Periods to analyze e Aggregation condition e Top X H selection of items e Length Top X H list e Font graphics e Hidden settings 4 1 Default directories 4 2 The user can set the default directories for 1 Data files 2 Result output files 3 Temporary files after closing the program temporary files will be deleted Figure 4 1 The dialogue window to set the directories of the litter data csv files the result files and the temporary files Data files C Litter Analystidata Results C Litter Analyst results Browse Temporary files C Litter Analystitemp Browse Ok Cancel Periods to analyze The user can define the start date and the end date of two periods to be analyzed The third period is automatically defined by the start date of the first period and the end date of the second period Statistical and trend analysis are appl
17. from the first date is used 8 Aggregate the items of beaches of a country or a region When a national data file is selected a selection screen for the aggregation of national beaches is presented The user can select the beaches for an aggregation of their items When a regional data file is selected see 2 2 Litter Analyst applies beach weighting in the aggregation of items of beaches see 4 7 The beach weights are defined in the sheet Region beaches in the file Litter Analyst_config xlsx in the directory config 9 Determine the statistical characteristics of all items of the three defined periods see 4 2 and perform the Wilcoxon ranksum test on a step trend of total counts AMO Icastat 3 Manual of Litter Analyst 2 0 2 2 10 11 12 13 14 15 Determine the data density table of the three defined periods Perform trend analyses on the time series of all items of the three defined periods Determine item trend indices ITI s of the three defined periods Determine the evaluation table of items of the three defined periods Determine the evaluation table of sources of the three defined periods Determine the evaluation table of materials of the three defined periods Also see 2 4 of our report Evaluation and fine tuning of a procedure for statistical analysis of beach litter data It can be found in the Help menu of Litter Analyst Make regional files Use this function to make regional cs
18. ied to each of these three periods The first two periods must fulfill the following two conditions 1 They are not empty 2 They do not overlap The program gives a warning if no trend analyses can be performed Trend analysis can only be performed if the period length is at least 4 5 years AMO Icastat 6 Manual of Litter Analyst 2 0 Figure 4 2 The dialogue window to define the two periods to analyze The third period is defined by the start date of the first period and the end date of the second period Start date period 1 at least 4 5 years 1 1 2002 End date period 1 31 12 2007 Start date period 2 at least 4 5 years 1 1 2008 End date period 2 31 12 2014 Ok 4 3 Aggregation condition The user can set the minimum percentage of beaches that is required for aggregation of beaches of a region For each quarter of a year the aggregated count of an item of a region is determined as the weighted average of counts of that item on the selected beaches of that region If a beach contains litter data of two or more surveys in the same quarter then for each item of that beach the median of the counts of these surveys is used as the count for that beach The user can choose between the following settings 1 Required availability of 100 If for a specific quarter the data of an item is missing from at least one of the beaches the aggregated count of that item is set to a missing value for that quarter 2 Required av
19. imated step trend of the beach total counts is shown otherwise the corresponding cell is left empty This p value is the result of testing on a step trend using the Wilcoxon ranksum test If an estimated trend or change is negative that cell is green and if it is positive the cell is orange If the p value is less than 0 05 indicating a statistically significant trend that table cell is grey AMO Icastat 14 Manual of Litter Analyst 2 0 5 1 6 Group 4 Summary of trends slopes of top X H items This group summarizes the results of all trends slopes of the top X H items regardless of their Statistical significances This summary can be more sensitive to a general change in one direction improvement or deterioration than the ITI s because these latter are only based on censored information the statistically non significant trends are set to zero First subgroup Percentage of negative slopes cell is green if the percentage is not zero and the percentage of positive slopes cell is orange if the percentage is not zero for the three periods Second subgroup Statistics of the estimated slopes for the three periods Presented are the minimum the 25 50 and 75 percentile and the maximum If a slope statistic is negative its cell is green and if it is positive its cell is orange 5 1 7 Evaluation table of items versus those of sources and materials The evaluation tables of sources and materials are somewhat different from the it
20. ise the percentage of zero slopes see Zero slope in the data density table can be too high to get reliable trend results with the Mann Kendall test Harmful items are defined in the sheet Ecological harm in the file Litter Analyst _config xlsx in the directory config The top X H list is displayed in the Evaluation tables of items see 5 1 and in the data density table see 5 7 Font graphics In a dialogue box the user can set the font the style and the size of graphical presentations in Litter Analyst A font change can help to make better graphical presentations for the export to a Word file AMO lcastat 9 Manual of Litter Analyst 2 0 Figure 4 6 The dialogue window to set font style and size of the graphical presentations Font Style Bold Plain Script MT Bold Bold Segoe Print Italic Segoe Script Bold Italic Segoe UI Segoe Ul Black Segoe UI Emoji Sample The quick brown fox jumps over the lazy dog 1234567890 4 7 Hidden settings In case of a csv file with national beach litter data Litter Analyst will show a dialogue window with the beaches that are available for the aggregation of items In case of a csv file with regional beach litter data of more than one country no dialogue window is shown and all beaches with non zero weight of the csv file will be automatically selected for aggregation Litter Analyst applies beach weighting in the aggregation of regional beach litter data from cs
21. list Based on item counts Based on item ranking scores Based on a predefined item list Ok For an item on an individual beach there is no difference between the ranking of the top X H on basis of counts 1 or scores 2 The beach weighting is defined in the sheet Region beaches in the file Litter Analyst_config xlsx in the directory config AMO Icastat 8 Manual of Litter Analyst 2 0 4 5 Length top X H list 4 6 The user can determine the length of the top X H list 1 The minimal percentage of total counts of all items that the top X list must represent 2 The minimal number of items in the top X list 3 The minimal number of harmful items in the top X H list The top X list is a list of the minimal number of top items that covers at least a certain minimal percentage of the total counts of all items The top X H list is the top X list including a certain minimal number of harmful items Figure 4 5 The dialogue window to set the length of the top X H list Minimal percentage of counts of items in top X list 80 Minimal number of items in top X list 15 Minimal number of harmful items in top X H list 5 Ok Cancel These three settings together determine the length and composition of the top X H list By setting two of the three settings at a low value the third setting will determine the list We advise not to include items with less than 1 of total counts of all items because otherw
22. m them And in 5 1 7 we describe the differences between the items evaluation table and the evaluation tables of sources and materials 5 1 1 Settings The settings are presented in the first four rows of the evaluation table These are the aggregation condition see 4 3 the ranking system see 4 4 the source csv file and parameters that determine the top X H list of items see 4 5 The latter is only applicable for top X H list items and not for the top X list of sources and materials In figure 5 1 we show an example of the settings in an evaluation table of items Figure 5 1 An example of the settings in an evaluation table of items Settings Source AMO Icastat 11 Manual of Litter Analyst 2 0 5 1 2 Metadata The metadata of the evaluation matrix are the name of the OSPAR country the name of the beach and the analyzed period The period to analyze can be set by the user see 4 2 5 1 3 Group 1 Summary of results from top X H items First subgroup Code and definition of each item of the top X H group for the three periods Second subgroup Descriptive statistics of the top X H items for the three periods This subgroup presents for each top X H item the median the arithmetic average the standard deviation all in counts survey and the coefficient of variation the ratio of standard deviation and average Also presented is the relative contribution of each item to the beach total counts over that p
23. nd sign is set to O The Tlsum of slopes presents the net change of these items m ITH om of slopes gt S j 1 The Thweighted average of slopes Presents the average slope weighted by the contribution of each item to the beach total counts and with all slopes set to O that are not statistically significant or are not estimated m n D a ls sientectagenace of slopes N S j 1 AMO Icastat 13 Manual of Litter Analyst 2 0 where s is the filtered magnitude of the estimated trend slope such that s if s is statistically significant and s Q if s is not statistically significant or is not estimated this latter is the case for all the items that are not in the top X H list It is important to realize that in deriving these ITI s all slopes that are not statistically significant or are not estimated this latter is the case for all the items that are not in the top X H list are set to 0 We refer to group 4 for an uncensored presentation of the characteristics of the slope estimates regardless of their statistical significances because that presentation can be more sensitive to a general tendency of change in one direction improvement or deterioration 5 1 5 Group 3 Summary of results from beach total counts First subgroup Descriptive statistics of beach total counts for the three periods Presented are the median the arithmetic average the standard deviation all in counts survey and the c
24. oefficient of variation the ratio of standard deviation and average Second subgroup If the number and temporal distribution of the survey data fulfill the criteria for trend analysis see 5 1 3 this subgroup shows the results of the trend analysis of the beach total counts for the three periods otherwise the corresponding cells are left empty If the criteria are fulfilled the magnitude slope and statistical significance p value of the estimated trend of the beach total counts are presented for the periods The trend magnitude slope expressed in counts year is estimated with the Theil Sen estimator and the p value is the result of testing on a monotonic trend using the Mann Kendall test If an estimated trend magnitude slope is negative that cell is green and if it is positive the cell is orange If the p value is less than 0 05 indicating a statistically significant trend that table cell is grey Third subgroup The third subgroup is only presented for the third and longest period It shows the magnitude step and the percentage change of the estimated step trend of the beach total counts by comparing the second period with the first The trend step expressed in counts survey and the percentage of change is estimated with the Hodges Lehmann estimator Only if for both the first and second periods the survey data fulfill the criteria for trend analysis see 5 1 3 the statistical significance p value of the est
25. of beach A Lanzada in Spain Figure 5 9 Year boxplot of All gloves 304 of beach A Lanzada in Spain File Options Explore Help Year box whisker plot All gloves 304 A Lanzada AMO Icastat 23 Manual of Litter Analyst 2 0 6 Help The menu Help contains the possibilities to view e This user manual e The report Evaluation and fine tuning of a procedure for statistical analysis of beach litter data e The version number of Litter Analyst AMO Icastat 24 Manual of Litter Analyst 2 0 7 References Baggelaar P K and Van der Meulen E C J 2014 Evaluation and fine tuning of a procedure for statistical analysis of beach litter data castat AMO October 30 2014 43 pages M Schultz D Fleet W van Loon and L Oosterbaan 2014 Joint proposal for a harmonized OPSAR beach litter assessment method Project plan RWS and Germany March 6 2014 OSPAR ICG ML Conclusions of the Breakout Working Group on Assessment Criteria for OSPAR beach litter data Vigo November 11 12 2015 AMO Icastat 25 Manual of Litter Analyst 2 0 Appendix A simple ranking system Example of a simple ranking system for determining the top X H items list If an item is the most counted item on a beach it receives a score of 10 the second most counted item receives a score of 9 the third one a score of 8 etc In the example below Nets and ropes is the most counted item on 9 of the 14 beaches score 90 the second most counte
26. of the aggregation of the Dutch beaches NaN Not a Number is given if the median of the series is zero the relative trend is the ratio of trend and median and when the median is zero the relative trend cannot be calculated AMO lcastat 17 Manual of Litter Analyst 2 0 5 4 Trend boxplot A trend boxplot of all trends is presented of the selected beaches of a period In figure 5 4 we show the boxplot of the four Dutch beaches Bergen Noordwijk Terschelling and Veere of the total counts the materials and the sources of 01 01 2002 31 12 2013 On the X and the Y axes are given the items and the number of beaches and the trend per year Figure 5 4 Trend boxplot of the total counts the materials and the sources of the four Dutch beaches of the period 01 01 2002 31 12 2013 File Settings Explore Help Trend year total counts materials and sources The Netherlands 01 01 2002 31 12 2013 S a 3 S It is possible to show the trend boxplot of 1 All clusters 2 Materials 3 Sources 4 Total counts materials and sources On the X axis the items are shown with the number of beaches AMO Icastat 18 Manual of Litter Analyst 2 0 5 5 Trend judgement histogram A trend judgement histogram of all trends is presented of a single period In figure 5 5 we show the trend judgement histogram of the four Dutch beaches Bergen Noordwijk Terschelling and Veere of the total counts the materials and the sources of 01 01 2002 31
27. v files The beach weights are defined in the sheet Region beaches in the file Litter Analyst_config xlsx in the directory config For each quarter the aggregated count of an item is determined as the weighted average of counts of that item on the selected beaches as follows n ae W bel 2w b 1 with ac the aggregated count c the count w the weight b the beach index n the number of beaches i the item index y the year index and q the quarter index q 1 2 3 or 4 AC yg AMO Icastat 10 Manual of Litter Analyst 2 0 5 Exploring data with Litter Analyst The menu Explore offers possibilities to view the following analysis results e evaluation table of items e evaluation table of sources e evaluation table of materials e trend plot e trend palet e trend boxplot e trend judgement histogram e trend analysis results e data density table e year boxplot e table of data series 5 1 Evaluation tables The evaluation tables of items sources and materials are presented in an Excel spreadsheet The file is saved in the temporary directory To enable a concise presentation of the results of the various statistical analyses we developed an evaluation table It can present for each separate beach four groups of results In 5 1 1 we describe the metadata of the evaluation table In 5 1 2 5 1 6 we describe the details of the four groups of results and indicate what statistical conclusions can be drawn fro
28. v files format OSPAR csv file The regions are defined in the sheet Region Beaches in the file Litter Analyst_config xlsx in the directory config To make the regional csv files 1 All national csv files in the data directory are read The countries are shown in the second column of the sheet Region Beaches All counts of all items of all beaches of all OSPAR countries are written to the csv file All countries csv All counts of all items of the selection of beaches of a region are written to the regional csv file with the name of the region The regions are defined in the third column of the sheet Region Beaches Use this function after a new download of csv files from the OSPAR database At startup and after a change of the data directory Litter Analyst controls if an update of the regional files is needed The 6 regions defined in Litter Analyst_config xlsx are 1 Se ae Arctic Seas Northern North Sea Celtic Seas Southern North Sea Bay of Biscay Iberian Coast AMO Icastat 4 Manual of Litter Analyst 2 0 3 Output of Litter Analyst Choose File and then Save or Save as to export The evaluation table of items to an Excel spreadsheet file The evaluation table of sources to an Excel spreadsheet file The evaluation table of materials to an Excel spreadsheet file The trend plots to a Word file The trend palet to an Excel spreadsheet file The trend boxplots to a Word
Download Pdf Manuals
Related Search