Automatic Model Parameter Optimization Speedup Using Efficient Methods and Parallel Computing.

Field applications of agronomic models generally require determination of many parameters. This is a difficult process partly because of uncertainty in measured or estimated values and the often wide range of acceptable values for given field conditions. Automatic parameter adjustment methods such as PEST (Model-Independent Parameter Estimation) are available to assist users in determining values that accurately simulate field targets (e.g., runoff, streamflow, crop yield). However, these optimization methods can take many thousands of model runs to complete. Recently, PEST has been modified to enhance optimization efficiency and has parallel computing ability. Preliminary results suggest that using these efficiencies with a single 8 core computer reduces RZWQM optimization times by more than a factor of 10 (from several days to a few hours).