APPLICATION OF DIFFERENT HYDROLOGIC MODELS IN FLASH FLOODS SIMULATION

In recent years, the flash flood occurs frequently and intensively. It has become a world-wide focus in the field of disaster prevention and mitigation. The flash flood critical rainfall determination method, flash floods forecasting model and experience forecasting method are the most commonly used methods. This study focused on the application of hydrological models for flash floods simulation.
The research has adopted 7 flood forecasting models with different types and structures: topography-based API model, Xinanjiang Model, Xinanjiang Model with excess infiltration, Sacramento Model, IHACRES, BP-KNN Model and TOPKAPI. Every model will be applied in semi-humid and semi-arid watersheds Banqiao, Maduwang, Zhidan in China, which have high risk of flash floods. According to the characteristic of flash floods, a set of evaluation for the simulation results isput forward. The purpose is to find out one or several hydrological models fit for research area and to provide reference for future related research.
Combining with topography and runoff characteristics of each watershed, the paper collected and compared various simulation results of different models. Results indicated that models performed varies in semi-humid and semi-arid basins because of the complicated runoff mechanism in these areas. Among the conceptual hydrological models, Xinanjiang Model with excess infiltration performed better than the models with single runoff mechanism. TOPKAPI has a better simulation results than the lumped models topography-based API model and IHACRES. However, it has higher data requirements. BP-KNN model contains no physical mechanism and performed best in calibration period, but the accuracy falls greatly in validation period.