Laserfiche WebLink
objective function. An objective function is used by parameter estimation programs such as <br />MODFLOW-2000 and PEST to quantify the model's goodness of fit when using either manual <br />techniques or a parameter estimation program. The objective function that will be used for the <br />SPDSS model will include a weighted combination of heads and fluxes, as discussed in Section <br />4. <br />The model parameters being modified during calibration are flnose flnat, based on the available <br />data, have $ne largest uncertainty and impact the objective function value as they are varied. <br />They >lnclude the follow>ing, listed iln the likely order of priority for implementation and <br />decreas>ing order of parameter uncerta>nnty: <br />• Aquifer hydraulic conductivity (K) <br />• Streambed conductance <br />• Pecharge <br />• Well pumping <br />• Lateral boundary inflows <br />• Specific yield <br />Numerous simulations us>ing fine parameter estimation program will be run by varying these <br />model parameters with>ln predefined ranges that bracket the probable range in these <br />characteristics. Tlne MODFLOW model results will be evaluated using the automated parameter <br />estimation package according to the calibration criteria discussed in Sections 3.1 and 3.2. These <br />criteria will be evaluated using an objective function that quantifies the difference between <br />model simulated and field observed values (residuals). For each sinnulation, a residual of the <br />weighted calibration targets will be produced using the objective function. Each simulation will <br />also be evaluated by its ability to meet the target water budget, head convergence and mass <br />balance criteria. Typically, head convergence will be acceptable us>ing a value of 0.1 ft for <br />transient simulations, and a mass balance of less flnan one percent for final time steps lln a stress <br />period. There are no generally accepted standards for head convergence criteria or mass <br />balance, however, the selected criteria are adequately conservative and will result in a valid <br />numerical solution. <br />The calibration will proceed using a combination of automated and manual methods to obtain <br />the best fit between model results and field data. The automated calibration process will <br />provide information on the residuals for the parameters used >ln the objective function. These <br />parameters unclude: <br />• grounndwater levels, <br />• stream flow, <br />• stream gain/loss. <br />However, the automated calibration process can lead to several combinations of model <br />parameters resulting in a sinnilar residual. To further evaluate calibration results other <br />parameters will be evaluated manually. These parameters are both numeric and non-numeric <br />and include the following: <br />• model water balance, <br />• groundwater budget, <br />• stream diversions, <br />P~ SPDSS T~8 2 Final TM 10-08-0t~.doc 4 <br />