Laserfiche WebLink
incorporated through weighting factors assigned to each target class. The calibration targets <br />and their weighting factors are discussed in Section 3. <br />The calibration process used in the SPDSS alluvial groundwater model involves the use of both <br />manual and automated parameter estimation techniques. The automated parameter estimation <br />technique is an alternative to manual methods in which a parameter is modified, a simulation <br />made, the results are evaluated, another parameter is modified and the process repeats. T11e <br />size and complexity of the SPDSS model warrants an automated process, once a stable model <br />producing a water budget within a reasonable range is obtained and initial sensitivity analyses <br />are completed. Automated parameter estimation techniques Have been employed for over a <br />decade and extensive guidance exists for their implementation (Doherty 2004). The automated <br />model calibration will be undertaken as described >11 Section 4. <br />2.0 Calibration Process <br />Model calibration is t11e process >IZ which model input parameters are varied wit11>11 <br />predetern~led ranges >11 an iterative manner until modeled results match observed data within <br />an acceptable range. The calibration process is a series of steps undertaken to calibrate a <br />model. This process is illustrated u1 Figure 2-1. <br />The overall model calibration process will be conducted in three steps. <br />1. Calibration to a representative steady-state period <br />2. Calibration to a representative transient period <br />3. Verification of calibration to t11e full study period <br />The time periods for these calibration steps are described in Section 3.1 below. The calibration <br />process starts with the development of model input files. This includes defining the model <br />configuration, initial model parameters, and stresses. These initial inputs are developed using <br />the data centered process using programs and tools developed specifically for this purpose. <br />These tools are discussed further >11 the Task 48 Alluvial Groundwater Modeling Report. This <br />uzitial model must be assessed to ensure that it is numerically stable over likely ranges of input <br />parameters, and that factors suc11 as dry cells are mininuzed. Model control parameters, <br />including selection of a solver and appropriate solution parameters must be configured to <br />enhance fllis stability. Initial sensitivity analyses are conducted in order to identify model <br />parameters that have the greatest control over the goodness of fit of the model to field data. <br />The calibration process will proceed by first approximating model parameters using asteady- <br />state calibration period (Figure 2-1). The model parameters from the steady-state calibration <br />will then be used as initial estimates for the transient calibration period to refine the model. <br />Finally, t11e calibrated model will be run over fl1e entire study period to verify that acceptable <br />agreement between the model and field data has been reached. Each of these steps is iterative in <br />nature. This is t11e process that will be implemented for the SPDSS and is described in more <br />detail below. <br />Automated parameter estimation techniques will be used during each stage of the calibration <br />process. Since the goodness of fit of the model is defined by comparing model results to field <br />data, a quantitative measure of this fit needs to be developed. This measure is defined as an <br />P~ SPDSS T~8 2 Final TM 10-08-0t~.doc 3 <br />