Research on dynamic early warning methods for flash flood disasters: a case study of Yuhang District, Hangzhou City

Abstract

Flash flood disasters are characterized by their sudden onset, rapid formation, high destructive power, and significant casualties, making them widely recognized as one of the most hazardous natural disasters globally. To improve the accuracy of flash flood forecasting and warning, this study comprehensively considers various factors including short- to medium-term weather forecasts, nowcasting of heavy rainfall, and real-time rainfall (water level) monitoring information for small watersheds. It also integrates the influences of rainstorm-flood characteristics, antecedent rainfall or soil moisture state changes, and channel flood routing on flash flood processes. The SCS runoff curve number method and the triangular unit hydrograph method are employed to determine dynamic early warning indicators for flash flood disasters during the risk warning stage and the real-time dynamic warning stage, respectively. Taking Yuhang District, Hangzhou City as a case study, this research addresses the issue of "missed alarms" in traditional flash flood warnings, which arises from using hourly rainfall data that artificially segments naturally continuous rainfall and easily misses peak rainfall intensity. By utilizing minimum-resolution data—specifically, 5-minute rolling rainfall data—for rolling analysis and warning, the scientific validity and rationality of the warnings are effectively enhanced.