Research on Regional Early Warning of Flash Flood Disasters in Small Watershed Based on Multi-source Early Warning Data Fusion

Abstract

The research on the early warning method of flash floods serves as an important groundwork for the prediction and early warning of flash flood disasters. Being restricted by various objective factors such as lack of data and complicated formation mechanism of flash flood disasters in small watersheds, accurate and targeted early warning has consistently been a weak link within the flash flood control system. In this regard, focusing on a series of problems, such as "short risk forecasting period", "frequent missing and false reports" and "poor early warning accuracy", this research explores a hierarchical partition management system that fulfills the demands of early warning based on the analysis of watershed lag time and risk situation. Meanwhile, based on the comprehensive consideration of the characteristics of various multi-source early warning data, such as meteorological forecast rainfall and monitoring rainfall water level, this research proposes corresponding early warning modes for different types of early warning objects. On these grounds, this research applies the proposed early warning system to Xiaguan Creek in Shangyu District, Zhejiang Province. Relevant findings indicate that the effective extension of the forecasting period, coupled with the reduction of the phenomenon of "missing and false reports", provides scientific support for the early warning and transfer decision related to flash floods.