Enhancing flood peak simulation in data-scarce mountain river basins: the CRFMODEL framework

Accurate flood peak prediction in data-scarce mountain river basins remains critically challenging due to limitations in existing models: excessive focus on complete hydrological processes, inadequate incorporation of rainfall time distributions, and poor interpretability of flood peak formation mechanisms. To overcome these constraints, this study develops the CRFMODEL framework—a novel approach integrating Comprehensive Rainfall Factors (CRF) that synergize runoff generation theory, routing mechanisms, and rainfall time distribution patterns. Requiring only 9 parameters and readily accessible rainfall/flood peak discharge data, CRFMODEL mechanistically quantifies impacts of antecedent moisture conditions (AMC) and peak-triggering rainfall. Applied across eight Chinese mountain catchments (102–12,624 km2) using 280 flood events, the model demonstrates superior performance compared to the benchmark Xin’anjiang (XAJ) model, showing significant improvements in key performance metricsand achieving a qualification rate of at least 90% in all eight study basins (ARE < 20%), which meets China’s Grade-I flood forecasting standards—the highest level of accuracy in China’s flood forecasting system. Although the model relies on point rainfall data and may not fully account for spatial variability, this parameter-parsimonious framework bridges empirical efficiency with physical interpretability, providing a valuable tool for enhancing flood resilience in vulnerable data-scarce mountain river basins.