Fitting Methods for Conceptual Catchment Models

Abstract

To use conceptual (mathematical) models of catchment hydrology, it is first necessary to know the values of the parameters that control the behavior of the model. This has usually been done by fitting the model subjectively to actual data. Automatic optimization methods programmed for high-speed digital computers were used to fit a model objectively to given catchment data. A simple model incorporating the types of component used in more realistic models was used to test nine different optimization methods. A brief description of how each optimization method works has been given. By using synthetic error-free data, test problems with known solutions were constructed. Of the nine methods tested, a modified version of the deterministic rotating coordinate method of Rosenbrock was found to be the most effective for fitting the test model.