< Previous20 Figure 2-3: Default Parameters and Options menu. 21 Chapter 3 MODEL CREATION AND PARAMETER OPTIONS Typically, model creation begins with navigating to the area of interest and creating a domain of sufficient size to capture the regional dynamics (more on this below). Users may opt to generate purely synthetic models (in which none of the global framework are utilized, in which case a user background may be applied (see Chapter 7). This chapter will primarily deal the typical case of using all/some of the spatial framework data during modeling. 3.1 Delineating the Model Domain After navigating to the area of interest (see Section 2.3), use the red ‘DomainRect’ button to create a rectangular-shaped modeling domain within the MDWA, or use the ‘DomainCreate’ button to trace out a polygon of any shape as the model domain (e.g., to follow along a groundwater divide or watershed boundary). • When drawing a rectangle, indicate the position of the first vertex with a single LM click, and indicate the position of the vertex diagonal from the initial vertex with another single LM click2. • When drawing a polygon, single-click the LM button to define the vertices of the polygon that will define the model domain. Hover the cursor over the circles at vertex locations and line segment mid-points to enable node editing capabilities. Click-drag a node to adjust the size or shape of the model domain. After any change is made, an ‘un-do’ button () will appear next to the last node that was edited. Click on to return to the previous node position. Note changes can only go back one node-edit in the current version of MAGNET. An example of creating and adjusting the model domain is shown in Figure 3-1. 2 Allow for a brief period of time (i.e., at least 2-3 seconds) between LM clicks. Otherwise, the locations of the vertices may not be consistent with what is desired. 22 Figure 3-1: Creating and adjusting the model domain using the ‘DomainRect’ and ‘DomainZone’ button. The blue in-set map highlights the ‘un-do’ button available after altering any node position. Use the ‘SaveShape’ button to finalize the model domain. This will return the cursor to “navigation mode” (i.e., the cursor assumes the mouse navigation utilities described in Section 2.1). If ‘SaveShape’ is not utilized, a new rectangle will be started (when using ‘DomainRect’), or another vertex will be added to the polygon (when using ‘DomainZone’). 3.1.1 Note on Domain Size Be default, the conditions along the domain’s lateral and bottom boundaries are ‘no-flow’ (i.e. flux across the boundary is zero at all points). However, lateral boundary conditions can be derived from an existing model encompassing the study area. Thus, an “iterative” procedure is recommended for proper characterization of flow dynamics within the study area (see Section 3.6). 23 3.2 Model Grid and Solver Settings In the top-left portion of the Default Parameters and Options menu, the user can select/input a number of model grid options. This includes: the number of model nodes in the horizontal (west-east) direction (NX), the number of sub-layers in the vertical direction (i.e., into the earth), matrix solver settings, and multiplicative factors for spatially-explicit raster files of hydraulic conductivity and recharge (see Figure 3-2). Figure 3-2: The Model Grid and Solver settings available in the Default Parameters and Options menu. Default values/settings are shown. The number of nodes in the north-south direction is automatically computed based on NX and the shape of the model domain. MAGNET allows for vertical discretization of geological (conceptual) layers into desired number of computational layers. Subdividing can be done based on 1) the land surface as the top boundary surface (this is the default setting); or 2) with a water table solved by a previously simulation within the study area (check the box next to ‘SubLayers=’ to use this setting). In the latter case, the previous water table used for sub-dividing must cover the entire model domain, but need not be the same size. However, what is most common is to: • first simulate the model domain without sub-dividing the aquifer in the vertical direction; • then check the box next to ‘Water Table as Top’ in the Default Parameters and Options menu, save the changes (click ‘OK’), and re-run the simulation. In the Basic version of MAGNET, NX is limited to a maximum values of 40, and the number of geological (conceptual layers) is restricted to one, and the number of computational layers is restricted to three. No such restrictions apply for the Premium version of MAGNET. The Premium version does not limit grid size nor the number of sub-layers in the vertical direction. 3.2.1 Zone Attributes Factors 24 Click the ‘Zone Attr Factors’ link to open the Zone Attribute Factors sub-menu. This menu is used to apply multiplicative factors for spatially-explicit rasters of hydraulic conductivity and recharge. For example, if a factor of 1.5 is used, every value in every raster cell is multiplied by 1.5. The user must check the box(es) next to the parameter (Conductivity or Recharge) and then specific the multiplicative factor in the text-box to the right. Click ‘OK” to save changes. 3.2.2 Matrix Solver Settings Click the ‘Matrix Solver’ link to open the Matrix Solver Options sub-menu. This menu includes parameters and options for the numerical scheme(s) and algorithms used to solve mathematical groundwater problem formulated for the study area. The algorithms available for solving the discretized flow and/or transport equations include: • Hybrid Successive Over Relaxation • Successive Over Relaxation • Conjugate Gradient • Bi-Conjugate Gradient • Bi-Conjugate Gradient with Partial Pivoting The user may specify the convergence criterion for the numerical solutions, namely: the maximum number of iterations; the threshold of acceptable relative percent error (between successive iterations); the damping/relaxation factor; and the number of preconditions (or condition number) The user may also customize the ‘outer loop’ matrix solve scheme by specifying the number of maximum iterations, the threshold of acceptable relative percent error, the Relaxation Factor, and the frequency with which the dry/wet cells are scanned in the solution matrix. The reader should consult a reference on matrix solver algorithms described above (see sparse matrix solvers) The default grid and matrix solver settings are shown in Figure 3-2. 3.3 Aquifer Elevations The top and bottom elevations of the model can be assigned in a number of different ways using the Default Parameters and Options menu. Figure 3-3 shows the options/parameters that are available, which are discussed in the following subsections. 25 Figure 3-3: Options and parameters for assigning aquifer elevations in MAGNET. Default values/settings are shown. 3.3.1 Top Elevations By default, the top elevation of the model follows the DEM of a resolution consistent with the model cell size (e.g., if the cell size is ≥ 90m, the 90m resolution will be used, if the cell size is ≈30 m, the 30m resolution will be used etc.). The user may also choose from 90m, 30m, 10m, and 1m (for select regions) resolutions by accessing the drop-down menu next to ‘DEM Res.’ Choose the ‘Constant’ option and use the text box to enter the elevation to be used as the top boundary throughout the model domain. User raster files can also be used to represent the aquifer’s top surface. To do this, check the box next to the ‘Import’ link, then select the link to open a file browser. Browse to and select the appropriate file for importing. A prompt with file metadata will appear after the file has been uploaded, and the file path and name will appear in the text box next to ‘Import’. If the units of the DEM are not meters, use the text-box next to ‘multiplier to meter’ to enter the appropriate multiplier, e.g., to convert from ft. to m, enter 0.3048, as 1 ft.=0.3048m. 3.3.2 Bottom Elevations The bottom boundary of the model domain (aquifer bottom) can be assigned as: • a constant value prescribed by the user (see Figure 3-4a); • a surface following the aquifer top elevation surface such that the aquifer has a constant thickness (Figure 3-4b); 26 • a constant elevation that is a prescribed value below the minimum DEM elevation within the model domain (Figure 3-4c) (this is the default setting); • a surface following the bedrock top surface raster on the IGW Server (Figure 3-4d); and • a surface based on a user bedrock top surface raster file, which can be uploaded by checking the box next to ‘Inport’ and then selecting the link to open a file browser. Browse to and select the appropriate file for importing (Figure 3-4e). • a surface prescribed by a spatially-variable aquifer thickness from the IGW Server (Figure 3-4e) • a surface prescribed by a spatially-variable aquifer thickness from a user aquifer thickness raster file (Figure 3-4f). The value in the text box will be applied for the first three options in the appropriate manner, e.g., if 200 ft is in the text box, and the ‘Thickness option is selected, 200 ft will be used as the value to be subtracted from top elevation nodal values across the model domain. Figure 3-4: Different possible settings for the aquifer bottom elevations. 27 3.3.2 Note on Minimum Aquifer Thickness When there is significant variability in thickness of the aquifer layer (or computational layers in the case of vertical sub-dividing), issues may arise related to numerical instability of the solution process. To alleviate/avoid such issues, a minimum aquifer (or computational) layer thickness may be assigned (see Figure 3-3) as a percentage of the maximum aquifer thickness. For example, if the default value of 20% is applied for the minimum aquifer thickness, any areas of the model domain with a thickness less than 20% of the maximum thickness will adopt a thickness equal to the value representing 20% of the maximum thickness. Of course, the value assigned by the user which is largest enough to avoid numerical issues (but small enough so that key details are not lost) depends on the range of aquifer thickness variability and the numerical scheme used when solving the groundwater problem. The default values shown in Figure 3-3 are recommended (especially for beginners), as they yield relatively robust solutions regardless of location. 3.4 Hydraulic Conductivity A deterministic approach is used for parameterization of hydraulic conductivity (K) of the aquifer. Users assign/input horizontal (Kxx) values and implicitly assign vertical (Kzz) and north-south (Kyy) conductivities through prescribing vertical and horizontal anisotropy ratios (Kxx/Kzz and Kxx/Kyy, respectively). Users can assign a constant effective Kxx value for the entire model domain, or they can use the spatially-explicit raster from the IGW Server (see Figure 3-5). Figure 3-5: Options and settings for parametrizing the hydraulic conductivity of the aquifer. Default values/setting are shown. User raster files can also be used to assign Kxx values across the model domain. To do this, check the box next to the ‘Import’ link, then select the link to open a file browser. Browse to and select the appropriate file for importing. A prompt with file metadata will appear after the file has been uploaded, and the file path and name will appear in the text box next to ‘Import’. Note that zones may be added in the model domain to assign a value of Kxx that overrides the value(s) provided in the Default Parameters and Options menu (see Section 4.2.1) 28 3.4.1 Anisotropy Ratios By default, the aquifer system is assumed to be isotropic (i.e., Kxx/Kzz = Kxx/Kyy = 1). To apply vertical and/or anisotropy ratios, check the box(es) next ‘Kxx/Kyy’ and/or ‘Kxx/Kzz’ and enter the desired ratio(s) in the text box(es) (see Figure 3-5). 3.5 Recharge and Surface Seepage In MAGNET, water may be added to the aquifer through the top boundary as recharge (precipitation/surface water that infiltrates down to the water table) or can be removed through the top boundary as surface seepage (in the case where the hydraulic head is larger than the land surface elevation). Recharge to the aquifer is treated as specified flux [L T-1] that is constant in time3. Users can specify a single value to be applied through the model domain (default setting), or can use the spatially-variable recharge raster available from the IGW server. User raster files can also be used to assign recharge values across the model domain. To do this, check the box next to the ‘Import’ link, then select the link to open a file browser. Browse to and select the appropriate file for importing. A prompt with file metadata will appear after the file has been uploaded, and the file path and name will appear in the text box next to ‘Import’. If the units of the recharge raster are not meters per day, use the text-box next to ‘multiplier to m/day’ to enter the appropriate multiplier, e.g., to convert from in./yr to m/day, enter 0.0000696, as 1 in./yr. = 6.96 x 10-5 m/day. Note that zones may be added in the model domain to assign a recharge value that overrides the value(s) provided in the Default Parameters and Options menu (see Section 4.2) Figure 3-6: Options for applying recharge along the top of the model boundary. Default values/settings are shown. Surface seepage is computed as the amount of hydraulic head exceeding the land surface (computed as aquifer head minus land elevation) multiplied by the leakancy (a factor representing hydraulic conductivity per unit thickness of the land surface). The default value is 1 day-1. To remove surface seepage as a possible sink of groundwater for topmost model cells, set the leakancy to zero. 3 Interactive Groundwater (IGW) includes a module for process-based and non-process-based watershed modeling, which enables time-dependent recharge simulation. 29 Figure 3-7: Surface leakance parameter text box. 3.6 Boundary Conditions & Hierarchical Modeling In Section 3.1.1, it was mentioned that, by default, the conditions along the domain’s lateral and bottom boundaries are ‘no-flow’, and that lateral boundary conditions can be derived from an existing model encompassing the study area. The general idea is to: first create (see Section 3.1) and simulate (Section 5.1) a steady-state model that contains – and is a few times bigger than – the area of interest. This is referred to as the “Parent” model. Then, create a smaller, nested model (or series of models) that derives its/their boundary conditions from the parent model(s). The box next to ‘Implicit Parent Model’ must be checked in the Default Input Parameters and Options menu after drawing the “Child” model domain within the Parent Model domain. Interpolation of Parent model head values will be applied along the boundary of the (finer-grid) Child model. Figure 3-8: Options for applying boundary conditions for flow and/or solute transport modeling. 3.6.1 Hierarchical Modeling Example An example of the hierarchal modeling approach for applying boundary conditions is shown in Figure 3-9. The initial model (top-left of Figure 3-9) is chosen to be much larger than the area of interest so that the major regional “groundwater mounds” can be identified and used to guide submodel delineation. The idea is to incrementally zoom in to the area of interest – at each step using the location of groundwater mounds to place groundwater boundaries, in the assumption that flow patterns further beyond the groundwater mounds have little impact on the area of interest. Note that once ‘DomanRect’ or ‘DomainZone’ is selected Next >