基于Bayes判别分析法的密云县山洪泥石流预报模型

针对密云水库上游密云县境内山洪泥石流灾害多发的特点,首次将Bayes判别分析法原理运用于自然灾害的预测预报.运用系统科学和山洪泥石流预报原理,对密云水库上游密云县境内影响山洪泥石流发生的系统内部要素和系统外部环境条件进行分析,确定用影响山洪泥石流发生的前15 d的实效雨量和当日雨量作为预报模型因子,并建立起一组山洪泥石流预报模型,经自身验证法和交互验证法对模型进行检验,其判断正确率均为82.4%,达到了Bayes判别分析要求的判别模型自身验证正确率>80%的要求,两种方法检验结果一致,模型表现稳定,表明所建立的模型可靠,可以投入使用.研究结果可为该地区的山洪泥石流预测预报提供一种较为简单易行的方法,为密云县山洪泥石流灾害防治提供一定的支撑,以减少山洪泥石流发生所导致的灾害损失.

Prediction Model of Flush Flood and Debris Flow in Miyun County Based on Bayes Discriminatory Analysis

In view of the fact that flush flood and debris flow disasters occur frequently in the upper stream of Miyun reservoir,the theory of Bayes discriminatory analysis is firstly applied in predicting the natural disasters.System science and debris flow prediction theory are adopted to analyze internal elements and external environmental conditions of the system which influences the occurrence of flush flood and debris flow in the area.Accumulated precipitation of the preceding 15 days and intraday precipitation are introduced as model indexes and a set of prediction models is then established.By checking the model with oneself validation and cross validation,the accuracy for judging is 82.4%,which meets the requirement of Bayes discriminatory analysis that the accuracy of judged model should be more than 80%.Two validation results are consistent with each other and the model is stable.Therefore,the established model is credible and can be put into practical use.The research offers a simple and quick way for predicting flush flood and debris flow in the area and provide a support for the prevention and control of flush flood and debris flow in Miyun County,so as to reduce the loss caused by the two types of disaster.