A novel dynamic flash flood early warning framework based on distributed hydrologic modeling

Flash flood disaster prevention urgently requires high-precision hydrological models. However, extreme weather increases the uncertainty of flash flood risks, making it difficult for existing hydrological models to accurately simulate the flash flood processes. This study focuses on the Daxi River watershed in Guangdong Province. Based on verifying the applicability of the China Flash Flood Hydrological Modeling System (CNFF), the Manning formula is introduced to invert the calculation of the critical rainfall threshold for flash floods. The flash flood-inducing factors were then qualitatively analyzed using multiple indicators to determine their risk levels, thereby proposing a dynamic early warning framework for flash flood disasters. Results indicate that: 1) the CNFF demonstrated relatively high simulation accuracy, with the average Nash-Sutcliffe Efficiency (NSE) during the calibration and validation periods being 0.79 and 0.89, respectively; 2) dynamic rainfall warning indicators for 1-hour intervals were obtained under different soil moisture conditions. The warning, danger, and extreme danger flow rates were determined to be 176.3 m³/s, 308 m³/s, and 483.6 m³/s, respectively; 3) utilizing flood disaster risk assessment indicators, the risk distribution of flash floods has been evaluated, with the relatively high-risk areas, high-risk areas, and low-risk areas accounting for 31.1 %, 7.26 %, and 6.42 %, respectively; 4) the delay times for issuing warnings in different risk zones were determined, achieving dynamic early warning for flash flood disasters. The above research results will provide theoretical references for improving the flash flood defense methods.