Comparison of optimization algorithms in parameter calibration of tank model

Among various deterministic rainfall-runoff models, tank model is often preferred for its simplicity. On the other hand, it requires much time and effort to obtain better results owing to the calibration of too many parameters in the model. Therefore, the demand for applying automatic calibration method has been increased. In this study, three optimization algorithms are tested for the automatic calibration: one nonlinear programming algorithm (Powell's method) and two meta-heuristic algorithms (Genetic Algorithm and Harmony Search). As a result, the two heuristic methods show a very good performance in the calibration compared to Powell's method. Attempts have been made to improve the performance of HS (Harmony Search). The improved HS shows better calibration results than GA does in a given CPU time. The Harmony Search algorithm of this study contributes in solving the biggest problem in using tank models, parameter calibration.