00231309

open access

Version of Record

Parameter Determination of a Tank Model for Probable Maximum Runoff Prediction

English (英語)

神戸大学都市安全研究センター研究報告

4

87-96

神戸大学都市安全研究センター

2000-03

2007-04-11

It is often realized in runoff analysis that model parameters identified from several observed runoff events in a certain watershed vary widely though each event has been well reproduced. Accordingly, when the runoff prediction with that model is required, it is hard to select proper numerical values of the model parameters. This study aims to propound a procedure of determining the parameters of a tank model that can be used to predict not only the peak discharge but also the hydrograph neighboring the peak due to the heavy rainfall of low occurrence probability even when data for parameter identification was limited to comparatively small runoff. The variation in the parameters is inferred to be due to variations in the runoff process and the contributing area when rainfall is not heavy enough to make the hydrological condition of runoff fully developed. Hence, multiple regression analysis was performed to examine the correlation of model parameters with rainfall characteristics and runoff characteristics which are expected to reflect the hydrological condition of each runoff event. It is found that the runoff coefficient and the centroid of rainfall distribution before the occurrence of peak discharge are significant quantities to estimate the proper model parameters. In addition, the regression expressions provide the information of concentration time. Accordingly, it becomes possible to determine the rainfall distribution from the intensity-duration curve for a prescribed return period and to predict the runoff.

紀要論文