Abstract
The storage function (SF) models have been extensively used for the rainfall-runoff modeling in which the Kimura's model with lag time is widely used as a fundamental flow model, especially in Japan, due to its simple model structure. In this study, therefore, we aim to analyze the effect of lag time in the conventional Kimura's SF model on hydrograph reproducibility and compared with Prasad's SF model for an urban watershed in terms of error functions, storage hysteresis loop, and Akaike information criterion (AIC) perspective. The analysis of the effect of lag time on hydrograph reproducibility revealed that the use of optimum lag time in Kimura's model can greatly improve the performance. Further, the Kimura's SF model with optimum lag time exhibited higher hydrograph reproducibility associated with lowest error evaluation criteria and lowest AIC values in the single-peak events which makes it the superior model for single-peak events. Concurrently, Prasad's model depicted better performance in terms of reproducibility and AIC aspect during the multi-peak events, which indicates that it is the parsimonious model for multi-peak events.
