< Previous80 bottom-left graphic in Figure 6-3). See subsection 4.2.1 for details on making sure a zone budget is created for a zone conceptual feature. Select the radio button next to ‘Cell’ to show the zone budget for an individual grid cell (see bottom-right graphic in Figure 6-3. Use the arrows next to ‘I=’, ‘J=’, and ‘K=’ to specify the model cell location of interest17 17 The origin for the I and J coordinates is the bottom-left vertex of the model domain. In MAGNET, ‘1’ corresponds to the bottommost layer in a MAGNET model, while the largest number corresponds to the topmost layer in a MAGNET model. 81 • Figure 6-3: Example Zone (water) Balance charts. Top-right:: zone budget for entire domain polygon; bottom-left: zone budget for a zone added to the model domain; bottom-right: zone budget for a single model cell. 82 6.3 Model Results vs. Observations (Manual Calibration) After model simulation is completed (in the case of steady state modeling) or paused (transient modeling), simulation head results (either for the steady-state flow field or the flow field of the current time step) can be compared to available observations by selecting the ‘Calibration’ button along the left-side of the MAGNET Modeling Environment (see Section 2.7). This will prompt the user to use a subregion of the model for comparison (choose ‘OK’ in the prompt) or the entire model domain (choose ‘Cancel’). 6.3.1 Loading Observational Data Whether a subregion or the entire model domain is used for comparison, the 45 Degree Line Based Calibration Chart will appear after a selection is made (see Figure 6-4). The next step is to select the data source for observational data. Access the drop-down menu next to ‘Obs Data Source’ and select from one of the following options: • Table – enter observational data directly into the interface • File – import observational data from a properly formatted .txt file • IGW Server – import observational data from the IGW server18 18 In the current version of MAGNET, this option is only available for the State of Michigan. 83 Figure 6-4: The 45 Degree Line Based Calibration Chart used for “manual calibration” (comparing current model outputs with observations). Table Option. If the Table option is selected from the drop-down menu, an empty Calibration Data window will appear (see Figure 6-5). The first (header) line must include the following text: WellID,Time,X,Y,Zf,Zt,LyrIndex,Obs,Color One observation per line may be added after the first line. • the 1st value is the well ID (WellID) • the 2nd is ‘0’ (to indicate that the comparison is for steady state, or the current time step in the case of transient modeling (Time) • the 3rd is the x-coordinate of the observation (X), e.g., longitude • the 4th is the y-coordinate of the observation (Y) , e.g., latitude • the 5th is the well screen top elevation (Zf), in units of meters • the 6th is the well screen bottom elevation (Zt), in units of meters • the 7th is the layer index (LyrIndex), which can be used if the well is to be placed in a specific geological/computational layer in the case that well screen top and bottom elevation information is not available. By default, this value is set to ‘0’ so that the Zf and Zt values are used • the 8th is the head observation/field measurement (obs), in units of meters • the 9th is used to specify long-type RGB (Red-Green-Blue) color of the data markers in the calibration plot 84 An example of the correct header line and a line of observation inputs is shown in Figure 6-6. After adding inputs for all observations, use the ‘Click to Add Data’ button, then click ‘OK’ to use the data points in the calibration chart. Click on the ‘ReDraw’ button to plot simulation results versus the observation data (see next subsection). Figure 6-5: Calibration Data window, ready for direct input of observational data. Figure 6-6: Example header line and line of observation inputs. File Option. If a .txt file with the appropriate format is available on user’s local machine or network drive, it can be loaded into the Calibration Data window. Select File from the drop-down menu next to ‘Obs Data Source’. When the Calibration Data window appears, click the “Browse…’ button to launch a file browser and navigate to and select the appropriate file for importing. An example Calibration Data window after importing observation data is shown in Figure 6-7. To use the imported data for comparison to model results, click the ‘Click to Add Data’ button, then click ‘OK’ to use the data points in the calibration chart. Click on the ‘ReDraw’ button to plot simulation results versus the observation data. Figure 6-7: Calibration Data window after importing observation data from a file. 85 IGW Server Option. In the current version of MAGNET, water wells with measurements of groundwater levels are available for comparison with simulated head. If the IGW option is selected from the drop-down menu, the Filter for Well Data window will open (see Figure 6-8). This window allows the users to choose which types of wells to from which to import water level observations import: • Drift(D) – include wells of any type completed in unconsolidated sediments associated with glacial drift and debris • Bedrock(R) – include wells of any type terminated in consolidated materials/fractured rock • Public Supply – water supply wells • Industrial – wells supporting industrial or commercial activities • Irrigation – wells used for irrigation needed for agriculture/landscaping • Residential – domestic-use water wells • Unknown(U) – water wells without use information • Other – water wells used for some other purpose than those listed above Check the radio box next to each option to include it in the well selection for importation, or uncheck to filter it out from the import. Figure 6-8: Filter for Well Data window for importing data from the IGW server The user may further refine the well selection by specifying a time interval from which to select wells on the basis of well construction date. To include a filter for the earliest date of construction, check the radio box next to ‘From:’ and enter the starting date (year, month, day) in the text boxes provided. To include a filter for the latest date, check the radio box next to ‘To:’ and enter the ending date (year, month, day) in the text boxes provided. Lastly, the user may restrict the well selection by a specified data range. Check the radio box next to ‘Limit:’, then enter the maximum and/or minimum head values for import into the text box next to ‘Max’ and/or Min (units: meters), respectively. After all filters have been selected/applied, click ‘OK’ to continue importing data from the IGW server. A prompt will appear asking if data should be downloaded for further analysis (select ‘OK’) or if it should just be used for display in the Calibration chart (select ‘Cancel’). Click on the ‘ReDraw’ button to plot simulation results versus the observation data. 86 6.3.2 Calibration Plot & Options After the observational data has been imported and ‘ReDraw’ has ben selected, the Modeling Samples vs. Calibration plot and the available chart options will appear. Additionally, any downloaded data will be displayed in the text window below the plot, which can be utilized for offline analysis if desired (see Figure 6-9). Figure 6-9: Modeling Samples vs. Calibration plot, extracted model values and residuals (relative to the observations), and chart options. 87 Each data point provides information on the agreement between the simulated head and the observed head. Points that fall directly on the 45 degree line represent perfect agreement between the model and measurement. Points above indicate a modeling result that is larger than the observed value, while points below indicate a modeling result that is smaller than the observed value. Statistical indicators are provided directly below the plot: • Npts – the number of observational data point available for the comparison • RMS Err – the root-mean-square error (units: meters) • Mean Err – the mean error (units: meters) • Nash-Sutcliffe E – the Nash-Sutcliffe coefficient (0 ≤ E ≤ 1) Interpolation Methods. Because water level measurements in wells never/rarely occur at the location of am model node, interpolation of simulated head values is needed to compare to observations. Three options are available in MAGNET: • 2D (traditional) bi-linear interpolation • 3D Trilinear interpolation • 3D Weighted interpolation for multi-layer models, where the weighting is determined by screen-transmissivity (see Figure 6-10). By default, 2D interpolation is used. Figure 6-10: Scheme used for 3D Weighted interpolation of simulated head for comparison with observations. 88 Chart Options. The user may adjust the axes units, the data marker size, and add statistical indicator overlays directly to the Calibration plot. To change the axes units between meters and feet, use the drop-down menu next to ‘ChartUnit:’. To change the size of the data markers (units: display pixels), click in the text box next to ‘Symb size’ and enter the desired value. To add confidence intervals to the Calibration Plot, check the radio box next to ‘Show std’ and enter the number of standard deviations used for computing the confidence interval in the text box provided. Additionally, users may add a moving-window, “Band-mean’ average to the Calibration plot by checking the radio box next to ‘Add Band-mean’. The value entered in the text box next to ‘Numb:’ is the number of equal-sized “bands” to divide the data range into for computing the average simulated and observed head within each band. Alternatively, users may input the size of the band by entering a value in the text box next to ‘Width’19. Figure 6-11 provides an example of adding Confidence Intervals and the Band-mean markers. 19 If a non-zero value is entered in the ‘Numb’ box, this option will override the ‘Width’ option. 89 Figure 6-11: Example Calibration plot with Confidence intervals (red dotted line) of one standard deviation and Band-mean indicators (yellow triangles). Residual Maps. In MAGNET, it is possible to display the locations of comparison points which yield residuals above/below a prescribed threshold. Use the drop-down mean next to ‘Show residual map’ to apply one of the following mapping options: • Distribution – the residual for all comparison points is indicated by colored data markers • Outside std – the location of comparison points that yielded residuals larger than the number of standard deviations used for computing confidence intervals CI). • Inside std – the location of comparison points that yielded residuals larger than the number of standard deviations used for computing confidence intervals. Each of the possible residual mapping options are shown in Figure 6-12. If the Distribution option is chosen, it is useful to check the radio box next to ‘Horizontal’ or ‘Vertical’ to display a legend in the horizontal or vertical direction, respectively. When the ‘Outside std’ option is selected, comparison points yielding residuals that are outside the CI and with a simulated head less than observed head are displayed as blue triangles in the MDWA. Conversely, comparison points y yielding residuals that are outside the CI and with a simulated head greater than observed head are displayed appear as red triangle (see Figure 6-12). Next >