基于雨型分类的山洪小流域洪水预报研究

Research on Flood Forecasting in Small Mountainous Watersheds Based on Rainfall Patterns

摘要: 洪水的形成受到多重因素共同作用，其中降雨作为主要驱动因子，其时间分布特征对洪水过程具有重要影响。对降雨进行雨型分类，并针对不同雨型进行差异化的预报，是提高水文模型预测精度的一种可行且有效的技术路径。本文以四川省山洪小流域清溪河为研究对象，引入无量纲曲线刻画降雨事件的时间分布，基于聚类分析对流域的不同降雨类型进行辨识，并进一步研究了降雨的时间分布特征对洪水响应及水文模型模拟能力的影响。结果表明，清溪河流域的主要降雨类型有三种，其中时程分布相对均匀的雨型最为常见。不同雨型在洪水响应方面表现出显著差异：降雨集中且峰值出现较早的雨型更易诱发单峰洪水，而分布均匀的雨型则倾向于形成多峰洪水。基于新安江模型对三类雨型对应的洪水过程进行模拟，所得平均纳什效率系数分别为0.75、0.77和0.81，表明模型对降雨过程均匀的雨型模拟效果最佳，而对短时强降雨类型的模拟精度相对较低。本研究揭示了流域降雨时间分布对洪水形成的影响规律，可为山洪小流域的洪水预报提供技术参考。

Abstract: The formation of floods is influenced by multiple factors, among which rainfall events, as the primary driving force, play a decisive role, with their temporal distribution being a particularly critical factor. Classifying rainfall into different patterns and adopting differentiated forecasting strategies is a feasible and effective approach to improving the accuracy of hydrological model predictions. This study took the Qingxi River Basin, a small mountainous watershed in Sichuan Province, as the research area. Dimensionless mass curves were introduced to characterize the temporal distribution of rainfall events, and cluster analysis was employed to identify different rainfall types within the basin. Furthermore, the impacts of rainfall temporal distribution on flood response and hydrological model performance were investigated. The results indicate that three main rainfall types are identified in the Qingxi River Basin, among which the relatively uniform temporal distribution is the most common. Significant differences are observed in flood responses under different rainfall types: rainfall with concentrated distribution and earlier peak occurrence tends to generate single-peak floods, while uniformly distributed rainfall is more likely to result in multi-peak floods. Flood simulations of the three rainfall types using the Xin’anjiang Model yield average Nash-Sutcliffe efficiency coefficients of 0.75, 0.77, and 0.81, respectively, suggesting that simulations perform best for uniformly distributed rainfall, but less accurately for short-duration and high-intensity rainfall events. This study reveals the crucial role of the temporal distribution of rainfall in shaping flood processes and provides valuable technical references for flood forecasting in small mountainous watersheds.