A nonlinear rainfall–runoff model embedded with an automated calibration method – Part 2: The automated calibration method

Abstract

The purpose of this paper is to develop an automated calibration method for the nonlinear computational units cascaded (NCUC) model. The simple genetic algorithm (SGA), a popular and robust optimization technique, is introduced in this paper as the basis of the automated calibration method. Therefore, the way to transform the model calibration problem into the optimization problem is first proposed. The general scheme to appropriately arrange the parameters of the NCUC model is then developed, so that the chromosomes of the SGA can be properly constructed. Two performance criterion functions, which are frequently used to evaluate the performance of the rainfall–runoff modeling, are adopted in this paper as the objective function to calibrate the NCUC model. Since the SGA imposes two restrictions on the fitness values, the key of the proposed automated calibration method is the evaluation of the fitness values. The methods to evaluate the fitness values according to the two objective functions are both given in this paper. With the proposed automated calibration method, high-quality parameters of the NCUC model can be obtained without modelers’ subjective interventions.