OPTIMIZATION OF MODEL PARAMETERS FOR GROUNDWATER FLOW USING PARALLELIZED PEST METHOD

During the last decades, a number of methods have been developed for the optimization of hydrologic model parameters. One frequently used method is the PEST method which is a LevenbergMarquardtbased optimization algorithm. As a nonlinear parameter estimator, the PEST method can exist independently of any particular models, can estimate parameters and/or excitations, and can carry out various predictive analysis tasks. Its model types cover a wide range. It is preferable to calibrate a MODFLOW model using the PEST method in many situations. However, for a largescale, highresolution or highcomplexity groundwater flow modeling, the parameter optimization requires massive computing time. In this study, the OpenMP programming paradigm is used to achieve the modest parallelism on a sharedmemory computer for the PEST method. Application of the parallel PEST method to Beishan area, Gansu province shows that the computing speed can be increased up to 3.7 times. The computing speed is defined as the ratio of the serial computing time and the parallel computing time. The parallel PEST parameter optimization method can be an useful and effective tool for large models.