Reconstruction and early warning of flash floods in Zhongdu River, Sichuan Province, China

Flash floods in small mountainous basins are characterized by sudden onset, high destructiveness, and significant early warning challenges, which are further exacerbated by scarce hydrological data and limited technical support. To address this gap, this study proposes a robust flash flood early warning framework tailored for data-scarce small mountainous basins, using the Zhongdu River Basin in Sichuan Province, China, as a case study. To enhance the accuracy of flash flood simulations and the reliability of warnings under data-limited conditions, this study couples the Xin'anjiang runoff model (XAJ) with the kinematic wave-based geomorphic instantaneous unit hydrograph (GIUH) for runoff simulation and routing and integrating the water level rise rate method to construct a comprehensive early warning system. Validation using the August 2018 flash flood event in the study area yields promising results: flood peak error (FPE) within 20%, peak time difference (PTD) less than 1 hour, and a Nash–Sutcliffe Efficiency (NSE) exceeding 0.8. Computed flood marks closely match observed values, with a deviation below 2%, and the water level rise rate method produces no false alarms. This research offers a practical technical reference for determining flash flood warning indicators and developing early warning systems in data-scarce mountainous basins.