根据福建省沙溪口流域雨量的时空分布,以降雨量作为聚类指标,根据前期降雨影响因子按照时间序列取16个指标,在空间上根据流域测点距离水库入口的远近分成远、中、近3个层次。利用MATLAB统计工具将近年来的23场洪水过程分成了4类,继而对待预报的样本用对应的类所率定的参数进行预报。实践证明,将“系统聚类分析”方法引入该流域的洪水预报,提高了预报精度。 According to the spatial and temporal distribution of rainfall in Shaxikou area of ​​Fujian Province, rainfall was taken as the cluster index. According to the rainfall factors in the previous period, 16 indicators were taken according to time series, which were divided into far, middle and near 3 levels. Using MATLAB statistical tools, the 23 flood processes in recent years are divided into 4 categories, and then the samples for forecasting are predicted with the parameters determined by the corresponding classes. Practice has proved that the “cluster analysis ” method is introduced into the flood forecast of the basin to improve the prediction accuracy.