一种基于马氏距离识别法的流域临界雨量拟定方法应用探讨

从水文学和统计学角度出发,通过分析流域前期雨量指数(下垫面条件)及相应某一时段最大平均雨量之间的关系,构建了一种马氏距离(Mahalanobis)判别模型,用于解决中小河流山洪预警临界雨量指标拟定关键技术问题。利用该方法,对鄱阳湖水系的潦河流域进行了应用试验,选取该流域近年来发生的29次洪水样本,拟定时间尺度为6 h的临界雨量线性判别模型。模型检验结果显示,率定期合格率约为85%,误判率较低;检验期模拟预测的合格率约为78%,模型预测效果较好,证明了该方法用于中小河流山洪预警临界雨量拟定的可行性。该方法简单、易行、实用性强,对开展山洪预警具有一定应用参考价值。

Discussion on application of a Mahalanobis Distance Identification Method-based drafting method for river catchment critical precipitation

From the aspects of hydrology and statistics，a Mahalanobis Distance Identification Model for solving the key technical problems from drafting the critical precipitation for pre-warning on torrential flood for medium and small rivers is established through the analysis made on the relationship between antecedent precipitation index（underlying surface condition）of river catchment and the maximum average precipitation within a certain period. Based on this method，an application experiment is made on Liaohe River Watershed in Poyang Lake Water System； in which 29 samples of the floods occurred in the watershed in recent years are selected for drafting the critical precipitation linear discriminant model with the temporal scale of 6 h. The model test shows that the qualified rate within the calibration period is about 85% with a lower misjudgment rate，while the qualified rate of the simulated prediction is about 78% within the testing period，thus the prediction effect of the model is better. Therefore，the feasibility of the method to be applied to draft the critical precipitation for pre-warning on torrential flood for medium and small rivers is verified. This method is simple and convenient to be used with strong practicality，and then has a certain application value for the pre-warning on torrential flood.