A Bayesian Runoff Simulation Method Considering the Influence of Flash Flood Disaster

Abstract:

The amplification of discharge caused by flash flood disasters has affected the performance of Bayesian runoff simulation in hilly watersheds. This paper， taking the Baisha River Basin in Dujiangyan with frequent flash flood disasters after the Wenchuan earthquake as the research area， explores the critical rainfall causing flash flood disaster in the basin， introduces the linear amplification factor to correct the amplification effect of flash flood disaster on flood flow， and constructs six Bayesian flood simulation schemes by coupling the first-order autoregressive residual error model with different heteroscedastic transformation functions. The outlet daily runoff process of the basin is then simulated using the GR4J model. The results show that the critical rainfall and linear amplification factor can effectively identify and correct the impact of flash flood disasters on discharge. After considering the influence of flash flood disasters， the accuracy， reliability， and precision metrics of the daily runoff simulation results in the Baisha River Basin are increased by 19%， 32%， and 62%， respectively. This method can be widely used in the basin with frequent flash flood disasters and provide a more accurate scientific basis for runoff forecasting in hilly watersheds.