A conceptual rainfall-runoff model considering seasonal variation

Among various deterministic rainfall-runoff models, the tank model, which is a typical conceptual rainfall-runoff model, is often preferred for its simple concepts. On the other hand, it requires much time and effort to obtain better results owing to the need to calibrate a large number of parameters in the model. Therefore, the demand for an automatic calibration method has been increasing. In this study, three optimization algorithms were tested for automatic calibration: one nonlinear programming algorithm (Powell's method) and two meta-heuristic algorithms, i.e. a genetic algorithm and harmony search. The success of the powerful heuristic optimization algorithms enables researchers to focus on other aspects of the tank model rather than parameter calibration. The seasonal tank model is devised from the concept that s easonally different watershed responses could be reflected by seasonally different parameter values. The powerful optimization tool used in this study enabled parameter calibration of a seasonal tank model with 40 parameters, which is a considerable increase compared with the 16 parameters of the non-seasonal tank model. In comparison, the seasonal tank model showed smaller sum of square errors than those of the non-seasonal tank model. The seasonal tank model could, therefore, be a successful alternative rainfall-runoff simulation model with its increased accuracy and convenience.