GWSDAT User Manual
Contents
1. Nitrate Sulphate Toluene TPH Time Series Plot Options Conc Trend Smoother I Conc Linear Trend Fit T Show Legend I Scale to Conc Data F Log Conc Scale I Overlay GW levels I Overlay NAPL Thickness Well Trend Plot TPH in M W9 inti pore EEEN AAA 05 Mm a h Figure 4 GWSDAT user interface Conc GreyScale C Conc Terrain Ciscles Conc Topo Circles Conc GreyScale Circies NAPL Circles Groundwater Flows C None Same Length Weighted Length Piot Options T Show Well Labels I Scale colours to Data Show Conc Values I Show GW Contour Overlay ShapeFiles FF Plume Diagnostics Display table G Trend C Threshold Absolute C Threshold Statistical Edit Thresholds Subset Plot Colour Key Show an Reds Greens E White Non Detects TCA Tenueyy s TVUSMO GWSDAT User Manual v2 1 7 GWSDAT User Interface Following successful model fitting the GWSDAT user interface will be dis played see Figure 4 Note Occasionally the GWSDAT user interface ap pears hidden underneath any windows that are currently open The user can locate GWSDAT from the windows taskbar at the bottom of the screen The GWSDAT user interface is a stand alone point and click graphical user interface GUI which the user can interact with in many ways The following sections will explain the interface in more detail 7 1 GWSDAT Spatial Plot Plot Type TPH 04 Apr 20092. A Appendices A 1 GWSDAT Software Architecture A 2 The Mathematics behind GWSDAT A 2 1 Spatiotemporal Solute Concentration Smoother A22 Plume Diagnostics gt coso becco kecen i sa A 2 3 Groundwater Flow Calculation A 2 4 Well Trend Plot Smoother A 3 Converting a CAD drawing to a Shapefile List of Figures O ON DOO FP WwW DY H Rh e Se Se a m OC b gt eS GWSDAT data input example ccoo eR Oe eee we GWSDAT Mon Menik ne dos eS EO ee ee eS GWSDAT options form 4 6 6 45 54 ea Re Ee GWSDAT user interface i sca sierra HH BO GWSDAT Sparel Fisk 246 be 644 Se eR EERE EEA GWSDAT Visualisation Options o o Example spatial plot with plume diagnostics GWSDAT Well Trend Pit s cs ss eR KR Re we GWSDAT Trend amp Threshold Indicator Matriz GWSDAT Well Reporting Report Generation Example GWSDAT GW Well Reporting Report Generation Example Example Estimate Plume Boundary plot Example Plume Animation plume diagnostic output GWSDAT Spline Smoothing Selection Illustration Acknowledgements The authors gratefully acknowledge the many different people who have will ingly contributed their knowledge and their time to the development of GWS DAT The authors wish to express their gratitude to Adrian Bowman Ludger Evers and Daniel Molinari from the department of Statistics Universit
3. The plume mass is calculated on a mass per unit aquifer depth basis e g kg m To calculate the total plume mass the user must multiply this value by the aquifer depth The plume center of mass x y is defined as the mean location of the concen tration distribution within the plume boundary region D The x coordinate of the plume Centre of Mass is evaluated by numerical calculation of Jf xs a y dedy A i x y dxdy 5 where x and y are the spatial coordinates and x y represents the predic tions of the spatiotemporal solute concentration smoother evaluated at time t In a completely analogous manner the y coordinate of the plume Centre of Mass is evaluated as follows JJ ys x y dzdy Y JJ x y dxdy 6 In the event that multiple plumes are detected then the above quantities are calculated for each individual plume and aggregated together The individual plume areas and masses are summed to calculate the total over all plumes The aggregate average plume concentration and aggregate plume centre of mass is calculated by taking a weighted average of the individual quantities 43 GWSDAT User Manual v2 1 A 2 3 Groundwater Flow Calculation For a given model output interval the Groundwater GW flow strength and direction are estimated using available GW level and well coordinates data The model is based on the simple premise that local GW flow will follow the local direction of steepest descent hy
4. checkbox located on the right hand side of the plot This superposes isobars of smoothed groundwater elevation data on top of the solute concentration plot e Overlay Shapefiles This controls whether to overlay a site plan XYLENE 02 Aug 2005 to 01 Sep 2005 ug l 120 gt 5000 5000 3000 1500 100 80 60 40 20 20 40 60 80 100 120 140 Plume Mass 0 0091555 kg m Plume Area 793 71 m 2 Figure 7 Example spatial plot with plume diagnostics e Plume Diagnostics This controls whether to calculate and display plume diagnostic quantities from the predictions of the spatiotemporal solute concentration smoother see Figure 7 The delineated plume 19 GWSDAT User Manual v2 1 is displayed with a solid red contour line which also includes a label displaying the plume boundary threshold value The plume centre of mass is displayed with a red cross and the plume mass and area printed at the bottom left margin of the spatial plot Note in order for the correct plume diagnostics units to be used the Coord Units field in the Well Coordinates table must be specified see Section 3 2 More details about plume diagnostics can be found in Section 7 7 and Appendix A 2 2 7 2 Well Trend Plot BENZENE in MW 01 o o o N o o F w 9 aan E Sy z a wu 2 E o 5 4 N fi 3 wE e E o o o 2003 2004 2005 2006 Date Figure 8 GWSDAT Well Trend Plot The well trend plot enables the
5. o A IN A 500 fi we 200 fi TOLUENE ug l 100 1 50 fi Date TOLUENE in MW 03 o o ute z o a gt 3 ao oS wW zZ Sond a a 3 ep Y o o a o 24 o o o T T 2003 2004 2005 Date T 2006 925 92 0 Groundwater Elevation us 955 96 Groundwater Elevation 950 945 TOLUENE ug l TOLUENE ug l TOLUENE in MW 02 o 8 y 8 N a 8 _ El a gJ 3 o gJ R o 84 T T T T 2003 2004 2005 2006 Date TOLUENE in MW 04 o o slo o ma 87 oy o Te a gt 4 e gt P a els o a r N d e a o ve 3 T T 2003 2004 Date 924 926 92 8 Groundwater Elevation 92 2 93 90 93 95 Groundwater Elevation 93 85 93 80 Figure 11 GWSDAT GW Well Reporting Report Generation Example the Page Up and Page Down buttons to toggle between plots if the number of wells selected is greater than 4 The current checkbox options selected in the Time Series Plot Options see Section 7 2 are respected Figure 11 is an example GW Well Reporting output e GW Well Reporting gt PPT This selection is identical but redi rects output directly to Microsoft PowerPoint 7 6 Spatiotemporal Diagnostics The spatiotemporal Diagnostics menu located at the top of the GWSDAT user interface Figure 6 provides options to assess the goodnes
6. All three animation methods display data for the current solute selected in the Solute listbox Figure 6 and maintain the current spatial plot options 25 GWSDAT User Manual v2 1 e Contour Animation This selection produces a series of spatial plots generated in a separate window from the first to the last model out put interval Once plotting is complete the user can toggle forwards and backwards between the different model output intervals using the Page Up and Page Down buttons on the keyboard e Contour Animation gt PPT This selection is identical but ad ditionally generates a PowerPoint slidepack of spatial plots for each model output interval More PowerPoint output options are given in Section 7 5 e Contour Animation gt HTML This selection generates a html an imation page of spatial plots in the user s internet browser During the short delay whilst html output is being generated it is recommended that the user not use any of the GWSDAT functionality To start the animation simply press the play button gt at the bottom left of the page The animation will play the spatial plots in time series order to build a movie of the historical solute concentration dynamics The html animation can be viewed independently of GWSDAT and hence provides an excellent media for communicating results to individuals who do not have direct access to GWSDAT 7 5 Report Generation The Report Generation menu located
7. Conc Terrain This option overlays the predictions of the spatiotemporal solute con centration smoother for a particular model output interval using a terrain colour scheme The above example Figure 5 displays the spatio temporal solute concentration smoother for TPH Total Petrol Hydrocarbons on the 3rd May 2009 Please note that the output of the spatio temporal trend smoother is always given for the latest date in the displayed output interval The dark green colours indicate low solute concentration and the colours are gradated through yellow and brown to almost white to illustrate increasing estimated solute con centrations The concentration values can be read off from the key on the right hand side of the plot As the user iterates through time steps it may be noticed that the area covered by the spatiotemporal solute concentration smoother changes This is due to the fact that spatiotemporal predictions are only gen erated between interpolated data and are not extrapolated to regions where no data exists which could potentially lead to erroneous re sults For each time step the area of the contour is calculated only from the collection of wells for which the monitoring period spans the current model output interval GWSDAT generates predictions in the convex hull region dictated by these wells The convex hull see http en wikipedia org wiki Convex_hull may be visualised as the expected boundary if an elastic band was pla
8. at the top left of the GWSDAT user interface Figure 6 provides options for automatically generating a series of report plots e Current Plot gt PPT Left clicking on any of the plots in the GWS DAT user interface see Figure 4 results in an expanded plot in its own separate window This selection copies the current plot into a Microsoft PowerPoint slide e Current Plot gt WORD Left clicking on any of the plots in the GWSDAT user interface see Figure 4 results in an expanded plot in its own separate window This selection copies the current plot into a Microsoft Word document 26 GWSDAT User Manual v2 1 e Latest Snapshot This selection generates a sequence of plots relating to the latest model output interval For each solute the spatial plot for the latest model output interval is generated This is followed by the latest Trend Threshold Absolute and Threshold Statistical Indicator Matrices The plots are generated one after each other To toggle between each of the plots use the Page Up and Page Down buttons on the keyboard e Latest Snapshot gt PPT This selection is identical but additionally generates a PowerPoint slidepack of the plots described immediately above GWSDAT Example e BENZENE e TOLUENE e XYLENE 2003 2004 2005 2006 ii fi i 1 1 1 1 L 1 Mw 09 MW 10 MW 11 105 4 F 10 ang H 103 4 H 102 4 H por 10 Y Ya H A 3 MV4 05 MYW 06
9. data by a method of substitution In accordance with general convention the default option is to substitute the non detect data with half its detection limit e g ND lt 50ug l is substituted with 25ug l For a more conservative choice select the alternative of non detect data to be substituted with its full detection limit e g ND lt 50ug l is substituted with 50ug 1 5 4 Spatiotemporal Modelling Resolution Modelling Resolution This option controls the resolution of the spa tiotemporal solute concentration smoother see Appendix A 2 1 The user can select between either a default resolution or a higher resolution model fit In most instances there will be little difference in the modelling results between the two settings However in some rare circumstances with com plex data sets it may well be necessary to use the higher resolution setting Please note it takes approximately 3 4 times longer to fit a higher resolution model 5 5 NAPL Handling Method An additional pop up box will be displayed after the GWSDAT options box if the input contains NAPL data i e NAPL is entered in the constituent field Selecting Yes to the question Do you wish to substitute NAPL values with maximum observed solute concentrations forces GWSDAT to recognise NAPL data in the input dataset as indicative of high dissolved 10 GWSDAT User Manual v2 1 solute concentrations This option has been added to provide the user with a more
10. gt This serves as a very useful method for outlier detection In addition the analysis may be skewed if data are input from monitoring wells with disparate construction or screened in different aquifer systems Another important point to consider is that the quality of the spatiotem poral smoother is directly influenced by quality of the underlying data In general data originating from sites with many evenly spatially dis tributed wells with a long time history leads to better quality smoother predictions The converse of a small number of wells or poor well lo cation network design e g wells located in almost a straight line or short monitoring history will lead to less reliable smoother predictions particularly at the edges In summary the spatiotemporal solute concentration smoother plot is provided to help the user visualise the distribution of solutes and as an aid to risk based decision making However for the reasons stated above the predictions should be interpreted with care and a more detailed eval uation may be necessary to understand observed trends and outliers Further methods for assessing the goodness of fit of the spatiotemporal smoother can be found in Section 7 6 For more details on the spatiotem poral smoothing algorithm please see Appendix A 2 1 The following sections describe the various spatial plotting options see Fig ure 5 16 GWSDAT User Manual v2 1 7 1 1 Plot Type e
11. in ArcMap after their conversion to shapefiles prior to upload into GWSDAT useful for removing gridlines etc A Uncheck the CAD layer used to produce shapefile to remove image from view window 46 GWSDAT User Manual v2 1 Ensure exported Shapefile is selected and visible in view window C Click Start Editing on the Editor toolbar above the map D Use the arrow pointer to select lines and press delete on the keyboard to remove from drawing Select UnDo from Edit Toolbar in case of errors Editor gt Stop Editing Click Yes to save edits Repeat data export as detailed in steps 6 9 above and re save as new shapefile 47
12. input data set for illustrative purposes The columns fields in the Historical Monitoring Data input table are as follows e WellName The name or identifier of the well or soil boring being sampled Well names must be consistent and unique For example MW 1 and MW would be treated as different wells e Constituent Here enter the name of the solute type e g Benzene Toluene Again in the same manner as WellName please ensure that the name of a solute is consistent and unique for all entries The iden GWSDAT User Manual v2 1 tifiers GW and NAPL are reserved for Groundwater elevation mea surements and NAPL thickness data respectively see further details below SampleDate The date at which the well was sampled not the date the results were returned from laboratory analysis Please use a cal endar date format the preferred format is dd mm yyyy and do not include time data Result According to the constituent type the result may be concen tration groundwater elevation or NAPL thickness data Concentration data The concentration of the constituent is entered here Non detect values should be entered as either lt X or ND lt X where X is the detection limit as specified by the laboratory For ex ample if the detection limit is 100ug l then either lt 100 or ND lt 100 is acceptable The non detect threshold value must be specified i e ND on its own is
13. m in Figure 13 To estimate total plume mass the user must multiply this value by the estimated plume thickness using same units as those entered in the CoordUnits field Additional information on how plume parameters are calculated can be found in Appendix A 2 2 7 8 Saving and Loading a GWSDAT Session The user can save their GWSDAT session at any time by selecting Save Session from the File menu located at the top left of the GWSDAT user interface see Figure 6 This is particularly useful for large data sets which take a long time to run a GWSDAT analysis The GWSDAT session is saved with a GWSDATData file extension All the current display settings e g Colour Scheme Plot Options Current Solute are saved This GWSDAT Data file can be forwarded to another user and viewed once again exactly as 1t was saved To open a previously saved session select Load Session from the GWSDAT Excel Menu see Figure 2 Note Occasionally the GWSDAT user interface appears hidden underneath any windows that are currently open The user can locate GWSDAT from the windows taskbar at the bottom of the screen 34 GWSDAT User Manual v2 1 References 1 10 Raftery A Madigan D and Hoeting J Bayesian model averaging for linear regression models Journal of the American Statistical Associa tion 92 179 191 1997 N Ahuja and B J Schacter Pattern Models John Wiley amp Sons New York 1983 Julia A Az
14. not permissible In the absence of known detection limits the user must substitute a sensible value e g the lowest detected value for the solute in the data set Groundwater level data is entered here as an elevation above a com mon datum e g metres or feet above sea level or some other common reference height Please ensure that all groundwater measurement en tries have the same units e g metres or feet and that the Constituent field is set to GW In the presence of NAPL please ensure that the groundwater elevation has been corrected for NAPL density See Figure 1 for an example of how to enter groundwater height data NAPL thickness data is also entered here Please ensure that all NAPL thickness entries have the same units e g feet or metres and that the Constituent field is set to NAPL If no NAPL is present do not add a NAPL entry with zero thickness simply omit from the table Where NAPL is recorded in soil borings that do not reach the water table the NAPL thickness should be entered as zero Well location markers for soil borings or wells where NAPL has been recorded are highlighted in red Units Here select from the drop down listbox see Figure 1 the cor responding units Solute concentration data can either be mg l or GWSDAT User Manual v2 1 ug l For groundwater elevation and NAPL thickness data please set to one of mm cm metres inches feet or leve
15. realistic picture of the area of impacted groundwater in the event that NAPL in wells prevents direct measurement of CoC concentrations Before using this function the user should however be confident that dissolved CoCs are derived from the observed NAPL and not from a different source Solutes flagged as Electron Acceptors see Section 3 1 are omitted from the NAPL substitution process 6 Data Validation Procedures In the majority of cases the default data processing options as displayed in Figure 3 will be acceptable To continue the GWSDAT analysis press the OK button This initialises a series of data validation procedures GWS DAT will report back by means of a messagebox on any data anomalies These include e Incorrect Date formatting e Unspecified detection limits e Missing or non unique well coordinates Data input case sensitivity issues Solute concentrations entered as zero detection limit must be speci fied An excellent way to spot errors in GWSDAT input data is to use the Excel Auto Filter on the input tables and inspect the unique filtered values in the drop down listbox Following successful data input GWSDAT will begin the model fitting pro cess see Appendix A 2 Model fitting progress is indicated to the user by a commented progress bar indicator 11 GI Fie Animations Report Generation Spatiotemporal Diagnostics Plume Diagnostics Visualisation Options Ethylbenzene
16. the CTRL key down whilst selecting and shift clicking can be used to select ranges If only one solute is selected then the plotting behaviour is modified such that the detect and non detect data points are coloured black and orange respectively Furthermore if the Display Trend Smoother is checked in Figure 6 then the corresponding trend smoother with 95 confidence intervals are overlaid as thin black lines onto each graph In order to avoid output which is too busy to be comprehensible the maximum number of wells that can be displayed in the same plot is 16 i e 4X4 matrix If the number of wells selected exceeds this then a series of well reporting plots will be generated in the same window To toggle between these plots use the Page Up and Page Down buttons Well Reporting gt PPT This selection is identical but redirects the output directly to Microsoft PowerPoint GW Well Reporting This selection is similar to Well Reporting but additionally allows the overlay of groundwater level and or NAPL thickness on each plot with the axis being placed on the right hand side The time series of observed groundwater level is represented by open circles joined by a black solid line The functionality works by iterating through the selected wells and pasting the Well Trend Plots into a 2X2 matrix In an identical manner to Well Reporting use 28 GWSDAT User Manual v2 1 TOLUENE in MW 01
17. the well or soil boring Well names must be identical to those specified in the Historical Monitoring Data table Hint It is better to name wells using the convention of MW 01 rather than MW1 so that plots in GWSDAT are correctly ordered e XCoord The x coordinate value for the corresponding well e YCoord The y coordinate value for the corresponding well GWSDAT User Manual v2 1 e Aquifer The aquifer field allows the user to associate wells or soil borings with particular subsurface features e g aquifers sub strata in the event that data from these needs to be modelled separately The user can enter the name maximum of 8 characters of the aquifer or sub stratum or select a letter A G from the drop down listbox The aquifer field can also be used to partition the dataset from a large site in the event that mutiple unrelated plumes are present or if wells are clustered with large gaps in between On initiation of a GWSDAT analysis the user is asked to select an aquifer subsurface feature to analyse Note Plots generated using data associated with particular subsurface features have the feature name appended to the title e g Shallow aquifer If the user leaves the aquifer flag as blank no such appending will occur e CoordUnits Select from the drop down listbox the length scale of the well coordinates Either leave this field blank or select metres or feet The units specified in this field are us
18. to 03 May 2009 i ug l Conc Terrain gt 5000 Conc Topo 5000 Conc GreyScale 224480 3000 Conc Terrain Circles 1500 C Conc Topo Circles 2 800 Conc GreyScale Circles NAPL Circles S 400 Groundwater Flows 200 C None 100 C Same Length y 75 Weighted Length ea 50 Plot Options N M Show Well Labels 25 10 I Scale colours to Data V Show Conc Values o 5 I Show GW Contour Ss 0 M Overlay ShapeFiles 551640 551660 551680 551700 F Plume Diagnostics Figure 5 GWSDAT Spatial Plot The GWSDAT spatial plot see Figure 5 is for the analysis of spatial trends in solute concentrations groundwater flow and if present NAPL thickness It displays the locations of the monitoring wells black solid dots together 13 GWSDAT User Manual v2 1 File Animations Report Generation Spatiotemporal Diagnostics Plume Diagnostics Solute Solute Conc Units Ethylbenzene C ng l Nitrate E Sulphate ug l al Time Steps Toluene TPH noe y Time Series Plot Options Select Monitoring Well IV Conc Trend Smoother BENS Sea Sis IBH3 l Conc Linear Trend Fit IGDBH101 GDBH102 I Show Legend GDBH103 e Scale to Conc Data GDBH104 GDBH105 M Log Conc Scale mwi T IMW10 Overlay GW levels MW101 a T Overlay NAPL Thickness Figure 6 GWSDAT Visualisation Options with the well names and actual measured solute concentration values detect data is disp
19. 51683 81 224448 57 MW103 TPH 3 11 2009 162 551688 30 224470 95 MW6 JEthylbenzene 03 11 2009 ND lt 1 ugl 551698 19 224468 99 MW6 Toluene 3 11 2009 ND lt 1 _ ug 55168293 224440 03 Mwe TPH 03 11 2009 ND lt 5 Jut 551687 43 224453 03 SGS4 P1 _ Ethylbenzene 03 11 2009 ND lt 1 _ ug l 51695 93 22446453 SGS4 P1 Toluene 03 11 2009 ND lt 1 551688 24 22442570 SGS4 P1 _ TPH 03 11 2009 ND lt 5 ug l 551677 34 224427 38 SGS4 P3__ Ethylbenzene 13 11 2009 ND lt 1 ug l 551667 54 224429 89 SGS4 P3 Toluene 03 11 2009 ND lt 1 Ju 551662 76 224443 96 SGS4 P3 TPH 03 11 2009 6 a 551658 17 224435 39 SGS3P1__ Nitrate 02 11 2009 57 12 i 7 551684 93 22445203 SGS3P2__ Nitrate 02 11 2009 62 21 e z 551658 40 224436 14 SGS3 P3 Nitrate 12 11 2009 65 18 mm 551664 47 22445461 A SGS4 P1__ Nitrate 02 11 2009 88 6 bas 551664 47 22445461 A SGS4 P3 Nitrate 02 11 2009 3 55 inches 7 551664 47 22445461 A MW102 Toluene 02 11 2009 ND lt 1 ju 55164546 22444097 _A Mw102__ TPH 02 11 2009ND lt 5__lug l 551645 46 224440 97 MW Toluene 2 11 2009 ND lt 1 _ ugl 551714 21 22445387 _A MW7 TPH 02 11 2009ND lt 6 551714 21 224453 87 A Figure 1 GWSDAT data input example 3 1 Historical Monitoring Data Input Table Each row of this table corresponds to a unique combination of well sam pling date and solute type Groundwater and NAPL Non Aqueous Phase Liquids gauging data may also be entered in this table Figure 1 displays an example GWSDAT
20. Ground Water Spatio Temporal Data Analysis Tool GWSDAT Version 2 1 User Manual by Wayne R Jones Wayne W Jones shell com Michael Spence I Mike SpenceAshell com Shell Global Solutions UK Brabazon House Concord Business Park Threapwood Road Manchester M22 ORR Statistical Consultant Statistics and Chemometrics PTD TASE Environmental Consultant HSE Technology Soil and Groundwater PTD HSGW Contents Introduction Installation Instructions Data Input 3 1 Historical Monitoring Data Input Table 3 2 Well Coordinates Table a 24 24 co 2 3 3 GIS ShapeFiles Table ee ee GWSDAT Main Menu Data Processing Options 5 1 Model Output Interval 2 lt 4 ew dee ee sans 5 2 GW Level Aggregation Method 5 3 Non Detect Handling o o e cocs da nye de aea daa aai 5 4 Spatiotemporal Modelling Resolution oo NAPL Handling Method sc 54404 cas caasa aa Data Validation Procedures GWSDAT User Interface TL GWSDAT Spatial Plot e si scos sa mra adare aa RA Fla Plot OpHOnS c o e q seocu aa a o ee eS He Well o is ds eee eee a a 7 3 Trend amp Threshold Indicator Matrix LL PU os eG eh ERA 7 5 Report Generation 64 2 6b we bee eee 10 10 10 10 11 7 6 Spatiotemporal Diagnostics T Pume Diagnostics seu La e 7 8 Saving and Loading a GWSDAT Session References
21. MWW 07 MYY 08 2 H 105 E porn en 5 pes E 1093 e E res A pu 2 Ard AA o rar merce f es _ A FP 10 o ce xf i al lt A j a le v rA A Xy j 100 eee ee der ee Y sr es 3 MVV 01 MVWV 02 MVWV 03 MW 04 mui i y ooo oeeo eee A 100 4 e Ass e pese pos T T T T T T 2003 2004 2005 2006 T T T T T T T T 2003 2004 2005 2006 Sampling Date Figure 10 GWSDAT Well Reporting Report Generation Example e Well Reporting This selection generates a matrix of graphs display 27 GWSDAT User Manual v2 1 ing time series solute concentration values on a well by well basis In contrast to the Well Trend Plot Section 7 2 it is possible to overlay different solute concentration values within the same graph Figure 10 is an example Well Reporting output The colour key at the top identifies each solute and the name of each well is displayed in a ban ner at the top of each of the individual time series graphs The Well Reporting output provides a very concise method of visualising a lot of data The choice of which solutes and wells to include together with the choice of whether to use a log scale for the solute concentration val ues is selected by the user from a series of interactive listboxes which are generated when the Well Reporting option is initiated Multi ple solutes and wells can be selected or deselected by holding
22. NE Plume Threshold Conc 10ug 1 0 018 0 020 0 022 y f 1 e y 0 016 L Plume Area m 2 0 014 L 7 0 012 L e 0 010 1 1600 1550 1500 1450 1400 Plume Threshold Conc 10ug l T 2004 2005 2006 Date 2003 1350 T T T T 2003 2004 2005 Date 2006 Concentration ug l 55 50 45 40 35 30 Example Site Average Plume Concentration XYLENE Plume Threshold Conc 10ug I T T T T 2003 2004 2005 2006 Date Figure 13 Example Plume Animation plume diagnostic output This function produces a series of spatial plots incorporating plume diagnostics data from the first to the last spatiotemporal model output interval In addition a time series plot of plume mass plume area and average plume concentration see Figure 13 is generated The user can also select to export the complete time series of plume diagnostics data to a Microsoft Excel csv file e Plume Animation gt PPT This selection exports a series of spatial plots incorporating the plume diagnostics data directly to Microsoft PowerPoint 33 GWSDAT User Manual v2 1 Note Plume parameters are calculated based on the units of length entered in the CoordUnits field in the Well Coordinates table Section 3 2 if no units are entered then relative changes in plume parameters are plotted in dimensionless units Note also that plume mass is calculated per unit of plume depth e g kg
23. arrows in Figure 5 display the estimated direction and relative hy draulic gradient of groundwater flow at monitoring points across the a site This is calculated from the combination of well coordinates and recorded groundwater elevations for this particular model output interval see Ap pendix A 2 3 for more details The Groundwater Flows radiogroup posi tioned to the right hand side of the plot allows the user to choose either not to display groundwater arrows or direction only arrows or both direction and relative strength arrows 7 1 3 Plot Options e Show Well Labels This controls whether to display well names labels immediately below the well locations e Scale colours to Data By default the colour key of solute concen trations is subdivided as shown in Figure 5 By using the same sub divisions the spatiotemporal solute concentration smoother plots can be directly compared between different model output intervals This 18 GWSDAT User Manual v2 1 control will produce a new colour key whose subdivisions span the con centration predictions for the current model output interval only e Show Conc Values This controls whether to display actual sam pled concentration values immediately above the well locations If the data is identified as a NAPL measurement the value will be displayed as NAPL in a red font e Show GW Contour To add contour lines of groundwater level data check the GW Contour
24. below the user specified threshold value the cell is coloured dark green If the upper 95 confidence interval is not below the threshold value the corresponding cell is coloured dark red In the event that this cannot be calculated e g no data or our confidence in the trend smoother estimate is poor then the cell is coloured grey If the current concentration value is classified as non detect then the corresponding cell is coloured blue The Subset Plot button located to the right see Figure 9 generates a user specified subset of the the Trend and Threshold Indicator Matrix plot in its own separate window The Show listbox allows the user to filter the Trend and Threshold Indicator Matrix plot according to the different colours For example if the user selects red then the plot will only display the corresponding rows and columns which contain a red entry These functions are particularly useful when there exists a large number of wells and or solutes The Colour Key button generates a graphic displaying the colour key ex plained above for the Trend and Threshold Indicator Matrix plot 7 4 Animations The animations menu located at the top left of the GWSDAT user interface Figure 6 provides three different methods for generating an animated series of solute concentration plots Section 7 1 These movies give the user an insight into the historical dynamics of site groundwater solute concentrations and distributions
25. ced around the loca tions of these wells e Conc Topo This function is identical to Conc Terrain but uses a topograghic colour scheme which gradates increasing solute concentra tions through blue green yellow and beige e Conc Terrain Circles This selection overlays terrain colour coded circles located at the wells which have been monitored within the cur rent model output interval The size of the circles scales with the log of the observed solute concentration values and the solute concentration range can be read off from the colour key to the right of the plot 17 GWSDAT User Manual v2 1 e Conc Topo Circles This selection is identical to Conc Terrain Circles but uses a topographic colour scheme Hint In the presence of poor well location network design or limited data then it is recommended the user select either the Terrain Circles or Topo Circles plot type e NAPL Circles This selection displays the observed NAPL thick nesses within the current model output interval as size scaled and colour coded circles NAPL thickness ranges are read off from the colour key on the right hand side of the plot Colours are gradated from dark red through yellow to almost pure white to illustrate increasing NAPL thickness The location of wells which have recorded NAPL in any part of their monitoring history are coloured with red solid dots instead of the usual black solid dots 7 1 2 Groundwater Flows The blue
26. dling spatial objects R package http cran r project org web packages maptools index html H B Mann Nonparametric tests against trend Econometrica 13 245 209 1945 A I McLeod Kendall Kendall rank correlation and Mann Kendall trend test R package http www stats uwo ca faculty ain Wood S N Generalized Additive Models An Introduction with R Chapman amp Hall CRC 2006 Wood S N Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models Jour nal of the Royal Statistical Society Series B Statistical Methodology 73 3 36 2011 A A Oloufa Triangulation applications in volume calculation Journal of Computing in Civil Engineering 5 1 103 119 1991 Eilers P and Marx B Generalized Linear Models with P Splines in Advances in GLIM and Statistical Modelling L Fahrmeir et al eds Springer New York 1992 Edzer J Pebesma and Roger S Bivand Classes and methods for spatial data in R R News 5 2 9 13 November 2005 R Development Core Team R A Language and Environment for Sta tistical Computing R Foundation for Statistical Computing Vienna Austria 2008 ISBN 3 900051 07 0 http www R project org Joseph A Ricker A practical method to evaluate ground water con taminant plume stability Ground Water Monitoring and Remediation 28 4 85 94 2008 Barry Rowlingson Peter Diggle adapted packaged for R by Roger Bi vand pcp fu
27. draulic gradient For a given well a linear plane is fitted to the local GW level data Li a bx cyi i 7 where L represents the GW level at location x y Local data is defined as the neighbouring wells as given by a Delauney triangulation http en wikipedia org wiki Delaunay_triangulation 2 of the monitoring well locations The gradient of this linear surface in both x and y directions is given by the coefficients b and c Estimated direction of flow is given by 6 tan 5 8 and the relative hydraulic gradient a measure of relative flow velocity is given by R VP E 9 For any given model output interval this algorithm is applied to each and every well where a GW level has been recorded A 2 4 Well Trend Plot Smoother The well trend plot smoother is fitted using a nonparametric method called local linear regression This involves solving locally the least squares problem mi a gt Vi a Bla 2 w x z h 10 where w x x h is called the kernel function A normally distributed probability density function with standard deviation h is used as the kernel h is also called smoothing parameter that controls the width of the kernel 44 GWSDAT User Manual v2 1 function and hence the degree of smoothing applied to the data the higher the value of h the smoother the estimates Within GWSDAT local linear regression is deployed using the R package sm 5 6 and the ban
28. dwidth is selected using the method published in 9 45 GWSDAT User Manual v2 1 A 3 Converting a CAD drawing to a Shapefile System requirements ArcGIS comprising ArcMap ArcEditor ArcCatalog 10 LE Open ArcCatalog from the Start Menu Start gt All Programs gt ArcGIS gt ArcCatalog In ArcCatalog navigate to ArcMap globe with magnifying glass icon When ArcMap opens a screen will pop up Select A New Empty Map then click OK Go to File gt Add Data positive sign with yellow triangle under neath gt Select site CAD drawing saved as a dxf file gt Add Click on the symbol to expand the sub layers of the dxf file e g Polyline Polygon Multipatch Point Right click on required layer e g Polyline or an edited amp exported shapefile to open the drop down menu On drop down menu select Data gt Export Data On Export Data pop up menu choose Select All Features This Layers Source Data and select the folder you wish to save the shapefile into then click OK Click Yes to add the exported data as a new layer Repeat steps 6 9 to convert all the layers required to produce the base map in GWSDAT into shapefiles Add the shapefiles into GWSDAT one by one see Section 3 3 to pro duce the complete base map image The next section details how to edit layers
29. e in the observations with the consequent loss of smoothness To overcome this hurdle the objective function modified with the addition of a term that penalises the lack of smoothness of the fit 39 GWSDAT User Manual v2 1 The objective function now takes the form S a y B x al Al Dall where A is a non negative smoothing parameter and D is the m 2 x m matrix 1 2 1 0 0 0 0 0 0 1 2 1 0 0 0 0 D 0 0 1 2 1 0 0 0 0 0 0 0 0 1 2 1 The additional term in the objective function Dal ay 2a2 a3 am 2 20m 1 Am controls the smoothness of the fit by applying penalties over adjacent coeffi cients By minimising the new objective function for a given value of A we obtain the least squares estimator of the parameters B B D D B y Consequently the fitted values are given by Ba B B B D D By Hy 2 When A 0 the expression for the estimator of the parameters amp boils down to the classical solution in linear models theory As A gt oo the fitted function tends to a linear function Figure 14 shows the effect of penalisa tion it forces the coefficients to yield a smooth pattern The fitting process of a function using B Splines is pictured with and without penalisation to gether with the basis functions the columns of the B matrix The left plot results from not penalising A 0 the term in the objective function that accounts for the smoothness it ca
30. ed in the calculation of plume mass balance parameters e g plume area and solute mass for further details see Section 7 7 on plume diagnostics 3 3 GIS ShapeFiles Table A site plan can be superposed over plots of concentration distribution NAPL thickness and groundwater elevation see Figure 5 Site plans are imported into GWSDAT in the form of shapefiles see http en wikipedia org wiki Shapefile for more information A shapefile is actually a collection of several files typically created using ARC GIS e Filenames shp The GIS Shapefiles Table is used to store the filepath to site plans which need to be in shapefile format The user can either enter the file location manually or use the Browse for Shapefile function in the GWSDAT Excel menu see Figure 2 for interactive file selection Only the location of the main shapefile file ending with a shp extension needs to be specified in this table the associated data files e g dbf sbn sbx shx will be picked up automatically provided they are in the same folder see example in Figure 1 It is possible to overlay multiple shapefiles up to a maximum of seven GWSDAT User Manual v2 1 4 GWSDAT Main Menu In MS Excel 2007 or later the GWSDAT menu is located under the Add Ins tab User Manual taser Data Fle CE Blank Data Input Template 17 GWSDAT Analysis if Basic Example LG Load GWSDAT Session E Comprehensive E
31. end where the solute concentration is esti mated to no more than double or half in the next two years Light red and light green indicate that solute concentrations will no more than double or half in the next year respectively Dark red and Dark green indicate stronger upward and downward trends respectively In the event that the trend cannot be calculated e g no data or our confi dence in the trend smoother estimate is poor then the corresponding cell is coloured grey Blue cells represent non detect data As an exam ple consider Figures 8 and 9 It can be seen that the trend at the end of the current model output interval grey vertical line in Figure 8 for Benzene at monitoring well MW 01 is decreasing The correspond ing cell in Figure 9 top left has been coloured light green to illustrate this e Threshold Absolute This assesses if the observed solute concen tration values for all well and solute combinations are below a user specified threshold value default value of 500 ug l within any given model output interval The threshold value is depicted as a horizontal dashed red line and the end of the current model output interval is de picted by a vertical grey line in the Well Trend Plot see Figure 8 The Threshold Absolute Indicator Matrix compares the observed concen tration values with the threshold value If any observed concentration values within a model output interval are above the threshold value t
32. groundwater elevation data within each 3 month interval so that a larger dataset is available for the plotting of elevation contours by local linear regression Note that both the monthly and quarterly model output options only ag gregate the dataset used to plot groundwater elevation contours The solute concentration dataset is not aggregated in time because the spatiotemporal model from which concentration contours are generated does not require this i e the underlying spatiotemporal model used to generate the solute concen tration smoother plots does not vary with the data aggregation interval Note that if no monitoring data is present within a particular monthly or quarterly interval then GWSDAT will not generate a groundwater elevation contour or spatio temporal solute concentration smoother plot This is to avoid producing potentially misleading spatial plots far away in time from any actual data GWSDAT User Manual v2 1 5 2 GW Level Aggregation Method In the event that there are multiple groundwater elevation measurements from the same well within a given output interval the user can select how to use this data The user can select to calculate either the Mean Median Min or Max groundwater elevation Again this choice does not affect the spatiotemporal model used to generate the solute concentration smoother plots 5 3 Non Detect Handling Non Detect Handling Method GWSDAT handles non detect
33. hen the corresponding cell is coloured dark red If the concentration values within a model output interval are all below the threshold value then the corresponding cell is coloured dark green In the event that no data exists then the cell is coloured grey If the current concentration value is classified as non detect then the corresponding cell is coloured blue To change the threshold values press the Edit Thresholds button This generates a data editor where the user can manually change the threshold limits in ug l for the different solutes There is no need to save the changes simply close down the data editor to automatically update threshold limit values The current solute threshold limit values are displayed directly below the columns of the traffic light plot e Threshold Statistical This assesses if current solute concentration 24 GWSDAT User Manual v2 1 levels for all well and solute combinations are below a user specified threshold value with a statistical degree of confidence Again the threshold value is depicted as a horizontal dashed red line and the end of the current model output interval is depicted by a vertical grey line in the Well Trend Plot see Figure 8 The Threshold Statistical Indicator Matrix looks at the intersection of the end of the current model output interval vertical grey line and the trend smoother solid blue line If the upper 95 confidence interval upper dashed blue line is
34. ilistic framework known as Bayesian after Rev Thomas Bayes whose rule represents the pivotal theorem in this approach see 7 1 15 Under this paradigm is not considered to be a fixed unknown quantity to be estimated but rather a random variable whose value may vary within a given range This behaviour is described in probabilistic terms which assign a measure of confidence or probability to each of the values A may take on The Bayesian framework allows to compute the probability that the random 41 GWSDAT User Manual v2 1 variable A may take a particular value conditional to the fact that y has already been observed This probability indicated as f A y is known as the posterior distribution of A Bayes rule states that f Ay x f y A f A where stands for proportional to f y A is known as the likelihood function and expresses the conditional probability of observing data y given that the true value of the parameter is A f A is known as the prior distribution of the random variable A and comprises our prior beliefs on its uncertainty The optimal value of A is the one that maximises the posterior distribution and is computed using numerical methods A 2 2 Plume Diagnostics GWSDAT calculates plume diagnostic quantities from the predictions of the spatiotemporal solute concentration smoother see Appendix A 2 1 In com mon to 3 20 numerical methods are employed to integrate out the plume diag
35. intervals are very large i e very low confidence in the trend smoother fit are coloured grey instead of blue and are disregarded from the Trend and Threshold Statistical matrix plot calculations see Section 7 3 The advantage of this nonparametric method is that the trend estimate is not constrained to be monotonic i e the trend can change direction More details of this nonparametric smoothing algorithm are given in Appendix A 2 4 e Conc Linear Trend This displays a traditional linear time series trend estimate green solid line together with 95 confidence intervals green dashed lines to the log of historical solute concentrations values This is equivalent to fitting an exponential decay growth model on a linear scale The statistical significance of this trend is assessed by means of the well established Mann Kendall trend test 12 21 GWSDAT User Manual v2 1 The Mann Kendall p value and the estimated solute concentration half life is displayed immediately below the main title of the well trend plot Users should be aware that individual well half life values should not be used to estimate the plume half life If the Mann Kendall p value is below 0 05 then the estimated trend is deemed statistically significantly different from 0 i e there is indeed trend present in the data A p value above 0 05 should be interpreted as there is no evidence to suggest that trend is present e Show Legend This con
36. iz C J Newell Meng Ling Hanadi S Rifai and J R Gonzales Maros A decision support system for optimizing monitoring plans Ground Water 41 3 2003 A W Bowman and E Crawford R package rpanel simple control panels version 1 0 5 University of Glasgow UK R package www stats gla ac uk adrian rpanel Adrian W Bowman and Adelchi Azzalini sm Smoothing methods for nonparametric regression and density estimation R package http www stats gla ac uk adrian sm A W Bowman and A Azzalini Applied Smoothing Techniques for Data Analysis the Kernel Approach with S Plus Illustrations Oxford Univer sity Press Oxford 1997 Denison D Holmes C Mallick B and Smith A Bayesian Methods for Nonlinear Classification amp Regression John Wiley amp Sons New York 2002 Paul H C Eilers Demr Milieudienst Rijnmond and Brian D Marx Flexible smoothing with b splines and penalties Statistical Science 11 89 121 1996 C M Hurvich J S Simonoff and C L Tsai Smoothing parameter selection in nonparametric regression using an improved akaike infor mation criterion Journal of the Royal Statistical Society Series B 60 271 293 1998 Duncan Temple Lang RDCOMClient R DCOM client R package http www omegahat org RDCOMClient 35 GWSDAT User Manual v2 1 11 12 13 14 15 16 17 3 Nicholas J Lewin Koh and Roger Bivand maptools Tools for reading and han
37. l Units must be specified for each entry All entered groundwater elevation measure ments must have the same units Likewise for NAPL thickness e Flags Four flags are available in GWSDAT v2 1 that can be used to modify the way in which certain types of data are handled by the software The E Acc Electron Acceptor NotInNAPL and Redox flags are used to identify input data types which are to be omitted in the event that the user activates the NAPL substitution function see section 5 5 Note that it is only necessary to flag one data row in this way for all rows containing that constituent to be excluded from NAPL substitution see Figure 1 The fourth flag Omit can be used to exclude individual data entry rows from the GWSDAT analysis 3 2 Well Coordinates Table The Well Coordinates Table is used to store the coordinates of groundwater monitoring wells or soil borings For most of the purposes of GWSDAT modelling it is only the relative distances between wells which are important This means any arbitrary cartesian coordinate system can be used as long as well coordinate values have an aspect ratio very close to 1 i e a unit in the x coordinate is the same distance as a unit in the y coordinate Hence well coordinates can be measured directly from a map or given in easting and northing etc Note GWSDAT cannot plot monitoring data in the absence of well coordi nates e WellName Name or identifier of
38. layed in a red font non detect in a black font The date interval for the displayed data is indicated above the spatial plot In this example a monthly model output interval has been selected and the displayed actual solute concentration values were sampled between the 4th April 2009 and the 3rd May 2009 If a GIS shapefile has been supplied then the major site features roads tanks etc are overlaid on the spatial plot as light blue lines To visualise the spatial plot for a different solute simply select from the Solute listbox from the Visualisation Options portion of GWSDAT user interface see Figure 6 The Solute Conc Units radiogroup controls whether to display solute concentrations in ug l or mg l To increment forward and backwards between time steps press the and Time Steps buttons again located in the Visualisation Options portion of GWSDAT user interface As the user increments through different time 14 GWSDAT User Manual v2 1 steps the small red horizontal bar at the top of the plot see Figure 5 changes position to indicate where the time step lies in the monitoring period The far left position indicates the beginning of the monitoring period and the far right indicates the end of the monitoring period By holding down the button the plot will update continuously to generate an animated series of images If the final time step is reached it will restart at the initial time step For more
39. le The third GIS Shapefiles table may be populated with links to the location of GIS shapefiles for use as basemaps or site plans if required Please ensure that there are no empty rows in these completed data input tables Two example data sets are provided with the software for training purposes the use of which is explained in section 4 GWSDAT User Manual v2 1 Historical Monitoring Data WellName Constituent SampleDate Result Units Flags Filenames shp SGS5 P1 Nitrate 05 11 2009 54 59 mgll E ace 551689 43 C SpatioTempiGls Files GWSDATex2 shp ISGS5P2__ Nitrate 05 11 2009 67 93 551679 43 224426 03 SGS5 P1 Sulphate 05 11 2009 99 mg ll E acc 551661 93 224430 99 SGS5 P2__ Sulphate 05 11 2009 61 mg 551674 93 224439 03 GDBH102 _ Ethylbenzene 03 11 2009 ND lt 1 ju 551678 76 224469 31 GDBH104 _ Ethylbenzene J3 11 2009 ND lt 1 _ jul 551696 31 22447470 GDBH104 _ Toluene 03 11 2009 ND lt 1 ju 55169631 224474 90 GDBH104 TPH 03 11 2009 36 ug 551664 76 22445911 Mw10 Ethylbenzene 3 11 2009 ND lt 1 fu 551696 93 224459 03 MW10 Toluene 03 11 2009 ND lt 1 _ lu 551695 93 224452 53 MW10 TPH 03 11 2009 ND lt 5 ug l 551636 84 22444036 Mw103 Ethylbenzene 03 11 2009 Su 551693 78 224446 68 MW103 Toluene 03 11 2009 Glug 5
40. methods to generate animations see Section 7 4 By left clicking the plot an identical but expanded plot is generated in a separate window This is particularly useful for plots which contain a lot of information This window can be resized and importantly the aspect ratio is maintained The user can save this plot in a variety of different graphical formats by selecting File gt Save as from the menu located at the top left of the window Alternatively the user can simply right click on the plot and choose to copy the image to the clipboard as either a metafile or bitmap and paste directly into another application Note that once the image has been saved to a graphical format the ability to maintain an aspect ratio of one is lost 15 GWSDAT User Manual v2 1 The spatiotemporal solute concentration smoother is a function which simultaneously estimates both the spatial and time series trend in site solute concentrations By smoothing the data in both space and time it provides a clearer interpretation of site solute concentration dynam ics than would otherwise be gleaned from the raw data However it is important to note that it is a smoother function and as such the predictions do not necessarily lie on the observed data points In the event that a sampled concentration value is significantly larger than the predictions of the spatiotemporal smoother the well label is coloured red and surrounded by braces e g lt MW 1
41. n be noticed that it yields a rather wiggly regression function In the right plot a suitable choice for A constrains the optimisation method to find values for the coefficients which result in a smoother regression curve Prior to fitting the regression coefficients a the observed solute concentration values are natural log transformed This avoids the possibility of predicting 40 GWSDAT User Manual v2 1 Response L Response 0 0 0 2 04 0 6 0 8 1 0 0 0 02 04 0 6 08 1 0 Figure 14 Curve based on 20 nodes in the basis without penalisation left with penalisation right negative concentration values and also helps the model cope with data which often spans several orders of magnitude Furthermore the uncertainty in the measured concentrations can reasonably be expected to be proportional to the magnitude of the value e g the uncertainty around a measured value of 10ug 1 would be expected to be very much less than the uncertainty sur rounding a measured value of 10000ug 1 The natural log transformation stabilises the variance The choice of the penalisation parameter A is a crucial matter as a too small value would result in overfitting i e capturing also the noise whereas an extremely large value would lead to underfitting flat predictive function as a result of loss of signal Several criteria have been traditionally proposed see 9 14 but we tackled the issue by means of a probab
42. nctions by Giovanni Petris and goodness of fit by Stephen Eglen splancs Spatial and Space Time Point Pattern Analysis R package http www maths lancs ac uk rowlings Splancs 36 GWSDAT User Manual v2 1 22 23 24 25 26 Deepayan Sarkar lattice Lattice Graphics 2008 R package http cran r project org web packages lattice index html Luke Tierney tkrplot TK Rplot R package http cran r project org web packages tkrplot index html Rolf Turner deldir Delaunay Triangulation and Dirichlet Voronoi Tesselation R package http www math unb ca rolf Yihui Xie animation Demonstrate Animations in Statistics R package http animation yihui name Achim Zeileis and Gabor Grothendieck zoo S3 infrastructure for regu lar and irregular time series Journal of Statistical Software 14 6 1 27 2005 37 GWSDAT User Manual v2 1 A Appendices A 1 GWSDAT Software Architecture The user entry point and data input platform to GWSDAT is Microsoft Excel by means of a custom built Excel Add in application The statistical engine used to perform geostatistical modelling and display graphical output is the open source statistical programming language called R 19 Members of the R community contribute statistical routines and functionality to this collaborative project by means of packages which can be downloaded from http cran r project org web packages The greatest majority of
43. nd time series trend in solute concentrations at a particular model output interval see Figure 9 The rows correspond to each well and the columns correspond to the different solutes The user can select between the options of displaying Trend Threshold Absolute or Threshold Statistical by using the radiogroup control called Display Table positioned to the right of this plot note that the example depicted in Figure 9 is set to display Trend The Trend and Threshold Statistical options use the fitted nonparametric time series trend smoother described in Section 7 2 e Trend This reports the concentration trend for each solute in every well within the selected model output interval The Trend Thresh old Indicator Matrix looks at the instantaneous gradient of the trend smoother solid blue line where it crosses the end of the current model output interval vertical grey line in the Well Trend Plot see Figure 8 The cells of the Trend Threshold Indicator Matrix are coloured to Trend 01 Feb 2006 Display table Trend Threshold Absolute Threshold Statistical Edit Thresholds Subset Plot Colour Key Show fan Reds Greens White Non Detects gt a lll Trend Trend Trend Figure 9 GWSDAT Trend amp Threshold Indicator Matrix 23 GWSDAT User Manual v2 1 indicate the strength and direction of the current trend White cells indicate a generally flat tr
44. ng the solute concentration as Yi gt bj x aj i 1 j 1 where the b j 1 m are m functions known as basis functions gener ally second or third order polynomials The first term in equation 1 is a linear combination of the basis functions b each evaluated at and aims at capturing the deterministic part of the i th observation generally known as signal the second term e accounts for the variability in the measure ment due to randomness and is usually termed as noise The behaviour of e is described in terms of a probabilistic model such a model guarantees that the value of e fluctuates around zero conveying the idea that we do not expect to make any systematic error in the measurement This model also comprises the notion that the expected spread of e is given by o7 the variance of the random component By using the matrix notation bi x1 bj x1 din 1 b1 x2 b x2 Dn 2 B x Wad s Ba e eas bas TY TA vee E equation 1 can be written in a more compact fashion as y B a a e Because as mentioned earlier we expect the e s to oscillate around zero a sensible choice for the regression parameters is the one that minimises the norm of the vector e defined as S a le ly B x al A large value of basis functions is generally chosen to allow the model to capture most of the signal The downside of this approach is that it tends also to overfit that is to fit the nois
45. nostic quantities For a given model time step a fine spatial mesh grid of predictions is generated The plume boundary region D for a given plume threshold concentration value is calculated using the R function con tourLines which is included in the base distribution of the R programming language 19 The plume area A is defined as A I f dady 3 where where x and y are the spatial coordinates and is calculated numerically using the areapl function from the R package splancs 21 The average plume concentration is defined as g f eed 4 where 1 y represents the predictions of the spatiotemporal solute con centration smoother evaluated at time t This integral and all subsequent integrals in this section is calculated numerically using a method described n 16 A Delaunay triangulation is performed using the R package deldir 42 GWSDAT User Manual v2 1 24 on the spatial mesh grid of predictions within the plume boundary D The integral is numerically approximated by summing up the individual vol umes under each prism formed Plume mass is calculated from the scaled product of plume area and average concentration The scaling factor encompasses the user specified value of ground porosity see Section 7 7 and appropriate scaling values for mapping together the volumetric concentration units e g ug l with the length scale see CoordUnits in Section 3 2 of the Well coordinates e g metres or feet
46. otemporal model and plotting minimum and maximum solute concentrations at the edge of the area bounded by the monitoring data see Figure 12 The grey shaded area between the minimum and maximum lines defines the region where plume diagnos tics can be calculated The user defined value of the plume threshold 31 GWSDAT User Manual v2 1 Estimated Plume Delineation Concentration Region for XYLENE at GWSDAT Basic Example o o D o a 5 2 C O o D o i 5 o o E e O o Yo 2003 2004 2005 2006 Date Figure 12 Example Estimate Plume Boundary plot The grey shaded area defines a region where plume analytics can be calculated concentration value is displayed as a horizontal red dashed line The user should select a solute concentration value that is just above the minimum line for the time period within which calculation of plume parameter is required It is clear from Figure 12 that a value of 5ug 1 works well in this example The selection of a higher solute concentra tion value within the grey area while still producing a closed plume would exclude part of the plume from the plume mass area analysis The selection of a lower solute concentration value e g 2 ug l would 32 GWSDAT User Manual v2 1 Plume Mass kg m in this example greatly reduce the time period over which the plume was closed e Plume Animation Example Site Plume Mass XYLENE Example Site Plume Area XYLE
47. r There is no warranty for the Program GWSDAT to the extent permitted by applicable law SHELL Affiliates of SHELL the copyright holders and or any other party provide the Program as is without warranty of any kind either expressed or implied including but not limited to the implied warranties of merchantability and fitness for a particular purpose The entire risk as to the quality and performance of the Program is with the LICENSEE Should the Program prove de fective the LICENSEE assumes the cost of all necessary servicing repair or correction 2 Installation Instructions For up to date advice on how to install GWSDAT additional training ma terials and software updates please refer to the GWSDAT Livelink site To access this site please contact your Shell COE Technical Assurance Regional Focal Point A minimum screen resolution of 1024 x 768 or better is required for correct display of the GWSDAT graphical user interface Lower screen resolutions may result in only part of the user interface being visible 3 Data Input Groundwater monitoring data is entered into GWSDAT by means of a stan dardised Excel input sheet Figure 1 GWSDAT will use the name of the input data sheet as the name of the analysis so please change accordingly The template has been designed with simplicity in mind There are only two input tables that must be completed namely the Historical Monitoring Data table and the Well Coordinates tab
48. s of fit of the 29 GWSDAT User Manual v2 1 spatiotemporal solute trend smoother e Output Predictions to Table This generates a csv comma seper ated value Excel spreadsheet with the predictions of spatiotemporal solute smoother together with the actual observed values e Spatiotemporal Diagnostic Plot This generates a matrix of graphs displaying the time series of solute concentrations on a well by well basis together with the predictions of the spatiotemporal solute smoother overlaid as a solid light grey line e Spatiotemporal Diagnostic Plot gt PPT This selection is identi cal but redirects output directly to Microsoft PowerPoint 7 7 Plume Diagnostics In GWSDAT v2 1 the user can select to automatically delineate a plume based on a user defined concentration threshold This functionality which is accessed via the Plume Diagnostics menu top right in Figure 6 allows the user to evaluate plume stability based on changes in the plume area centre of mass and plume mass through time The plume diagnostics analysis is performed on whichever solute is selected in the solute listbox see Section 7 1 and Figure 4 30 GWSDAT User Manual v2 1 CAUTION The plume diagnostics functions provide insight into rel ative changes in plume mass area and average concentration for a given site and monitoring well network Plume mass estimates are provided in units of kg m or kg ft to reflect this uncertainty Absolute
49. sing Options GWSDAT Options Select Model Output Interval Monthly GW Level Aggregation Method Mean Non Detect Handling Method Half of ND Value l Modelling Resolution Defaut y Figure 3 GWSDAT options form On initiation of a GWSDAT analysis data processing options are displayed as shown in Figure 3 The data processing options influence how the data is displayed and how non detects are handled GWSDAT User Manual v2 1 5 1 Model Output Interval The spatiotemporal model can generate predictions at a user specified inter val The three different options are as follows e All Dates Concentration and groundwater elevation contour plots are generated for every date represented in the input dataset This is a good option to choose if each monitoring event comprises samples measurements collected within one 24 hour period e Monthly Concentration and groundwater elevation contours are gen erated at monthly intervals working backwards in time from the latest date in the input dataset Choosing this option aggregates groundwa ter elevation data within each monthly interval so that a larger dataset is available for the plotting of elevation contours by local linear regres sion e Quarterly Concentration and groundwater elevation contours are generated at quarterly 3 month intervals working backwards in time from the latest date in the input dataset Choosing this option aggre gates
50. ss if required Spatial data is input in the form of well coordinates and wells can be grouped to separate data from different aquifer units The software also allows the import of a site basemap in GIS shapefile format Trend and contour plots generated using GWSDAT can be exported directly to Microsoft PowerPoint and Word to expedite reporting GWSDAT uses Microsoft Excel as the primary user interface and data entry platform The underlying statistical calculations and graphical output are generated using the open source statistical program R 19 More details on the software architecture and statistical routines can be found in Appendix A 1 and Appendix A 2 Potential applications where GWSDAT can add value cost savings and re duction in environmental liabilities through improved risk based decision making and response include e Early identification of increasing trends or off site migration e Evaluation of groundwater monitoring trends over time and space i e holistic plume evaluation e Nonparametric statistical and uncertainty analyses to assess highly variable groundwater data e Reduction in the number of sites in long term monitoring or active remediation through simple visual demonstrations of groundwater data and trends e More efficient evaluation and reporting of groundwater monitoring trends via simple standardised plots and tables created at the click of a mouse GWSDAT User Manual v2 1 Disclaime
51. these packages are licensed under the GNU General Public License http www gnu org copyleft gp1 html The authors of GWSDAT would like to gratefully acknowledge and thank the authors of the following packages which GWSDAT makes use of e animation 25 lattice 22 rpanel 4 and tkrplot 23 used for graphical output in GWSDAT e deldir 24 sp 18 splancs 21 and maptools 11 to provide spatial statistics routines e Kendall 13 sm 5 and zoo 26 to provide time series and trend detection routines e RDCOMClient 10 to provide report generation functionality to Mi crosoft Word and PowerPoint A 2 The Mathematics behind GWSDAT A 2 1 Spatiotemporal Solute Concentration Smoother The spatiotemporal solute concentration smoother is fitted using a non para metric regression technique known as Penalised Splines P Splines It is beyond the scope of this text to give a full and detailed explanation of this technique here However the following outlines some of the most important aspects for the purposes of GWSDAT For a more detailed explanation the reader is referred to 17 8 38 GWSDAT User Manual v2 1 Let y be the solute concentration at 2 1 Ziz Ziz where x and ziz stand for the spatial coordinates of the well and x 3 represents the corre sponding time point for the i th observation with i 1 n We start by modelli
52. trols whether to display a legend in the top right hand side of the plot giving a key of the plotting symbols e Scale to Conc Data By default the well trend plot x axis is scaled such that it spans the sampling dates of all data The y axis is scaled to span the current data concentrations and the user specified trend threshold limit see Section 7 3 By checking this control the x and y axes are scaled to the span the current combination of well and solute concentration data only e Log Scale Controls whether to use a logarithmic or linear scale for the y axis i e solute concentration values e Overlay GW levels Allows the user to overlay the corresponding groundwater level measurements on the well trend plot The scale is read from the right hand axis This function is useful for assessing correlations between groundwater levels and solute concentrations e Overlay NAPL Thickness Allows the user to overlay the corre sponding NAPL thickness level measurements on the well trend plot The scale is read from the right hand axis This function is useful for assessing correlations between NAPL thickness and groundwater levels As with the GWSDAT spatial plot left clicking the plot produces an ex panded plot generated in a separate window which can be saved as explained in Section 7 1 22 GWSDAT User Manual v2 1 7 3 Trend z Threshold Indicator Matrix The Trend and Threshold Indicator Matrix plot is a summary of the level a
53. user to investigate time series historical trends 20 GWSDAT User Manual v2 1 of solute concentrations in individual wells Figure 8 displays an example GWSDAT Well Trend plot of Benzene in well MW 01 using an illustra tive example dataset The actual sampled concentration values are plotted against sampling date and are represented as black solid points Orange points represent the substituted non detect values according to the selec tion chosen in Section 5 Red points represent the NAPL substitued solute concentration values The Visualisation Options see Figure 6 portion of the GWSDAT user in terface controls well trend plot display options To switch between different solutes and monitoring wells simply select from the Solute and Select Mon itoring Well listboxes The Time Series Plot Options checkbox control includes e Conc Trend Smoother This displays the estimated time series trend in solute concentration using a nonparametric smoother see Fig ure 8 The solid blue line displays the estimate of the mean trend level at a particular point in time The upper and lower dashed blue lines depict a 95 confidence interval around this estimate This is inter preted as one is 95 confident that the actual mean trend level lies within this region The smaller the 95 confidence interval the more confidence one has in the estimated time series trend Areas of the trend smoother fit in which the 95 confidence
54. values of plume mass cannot be calculated because the model does not define the vertical concentration distribution inside the plume The assumptions made in plume mass estimation are that e Vertical variation in plume concentration is averaged across mon itoring well screen intervals e The monitoring well network samples the full depth and extent of the plume and is also representative of concentration variation across the site The assumptions are similar to those presented in the paper by Ricker 20 e Edit Plume Thresholds This opens a data editor where the user can manually change the plume boundary threshold concentration limits in ug l for the different solutes There is no need to save the changes simply close down the data editor to automatically update the values e Set Ground Porosity This opens a data editor panel where the user can enter the effective interconnected porosity of the subsurface default 0 25 This value is used in the calculation of plume mass see Appendix A 2 2 e Estimate Plume Boundary This function provides a graphic to assist the user in selecting a solute concentration value that will yield a closed plume which is an essential requirement for the calculation of plume parameters The plume boundary contour is determined to be closed i e forms a loop when it does not extend outside the area bounded by monitoring data The function works by iterating through each time step of the spati
55. xample F ES Browse for ShapeFile _ HIDE About GWSDAT Figure 2 GWSDAT Main Menu The main menu options as shown in Figure 2 are as follows e User Manual Displays this pdf GWSDAT user manual e Insert Data File Blank Data Input Template Inserts a blank GWSDAT Data input template worksheet into the active Excel workbook Basic Example Inserts an example GWSDAT worksheet data set into the active Excel workbook Comprehensive Example Inserts a more detailed example GWSDAT worksheet data set which includes a site plan NAPL thickness data Electron Acceptor flagged solutes and multiple aquifers into the active Excel workbook e GWSDAT Analysis Begin GWSDAT analysis on the Excel active worksheet data GWSDAT User Manual v2 1 e Load GWSDAT Session Loads a previously saved GWSDAT ses sion see Section 7 8 for more information e Browse for Shapefile Interactively browse for a shapefile and add location to GIS Shapefiles table see Section 3 3 for more information e About GWSDAT Displays version information and Terms amp Con ditions for GWSDAT The two example data files are provided for training and demonstration purposes Select the basic or comprehensive dataset and then GWS DAT Analysis from the main menu Accept the default data processing parameters Section 5 and maximise the GWSDAT panel to practise using the data visualization tools Section 7 5 Data Proces
56. y of Glas gow for their invaluable contributions to the statistical aspects of GWSDAT Thanks also to Ewan Mercer from the University of Glasgow for his assistance in the development of the GWSDAT user interface We acknowledge and thank the R project for Statistical Computing and all its contributors without which this project would not have been possible A big thank you to Shell s worldwide environmental consultants for assistance in evaluating and testing the earlier versions of GWSDAT Thanks also to the Shell Year in Industry students Tess Brina Rosemary Archard Emma Toms Stephanie Marrs and Rachel Stroud who spent a great deal of time using GWSDAT and making suggestions for improvements We thank our colleagues Matthew Lahvis George Devaull Dan Walsh and Curt Stanley from Shell Projects amp Technology HSE Technology Soil amp Groundwater and Philip Jonathan from Shell Projects amp Technology An alytical Services Statistics amp Chemometrics for their support vision and advocacy of GWSDAT The original idea of GWSDAT was inspired by Marco Giannitrapani GWSDAT User Manual v2 1 1 Introduction The GroundWater Spatio Temporal Data Analysis Tool GWSDAT has been developed by Shell Global Solutions to help visualise trends in ground water monitoring data It is designed to work with simple time series data for solute concentration and ground water elevation but can also plot non aqueous phase liquid NAPL thickne
Download Pdf Manuals
Related Search