基于随机过程理论的长江巨洪发生概率预测模式与应用

Yangtze Large Flood Probability Random Process Model Prediction Research Based on the theory for stationary random processes

根据长江流域1849—1998年发生的巨洪资料序列,通过正态性、独立性等统计检验,确定序列的性质;然后,利用平稳独立随机过程理论建立概率预测模式,对长江流域巨洪发生概率作出研究性预测。结果表明,根据1849年以来长江巨洪资料样本,建立的巨洪发生间隔时间资料序列经对数变换后使原序列的线性和平稳性得以改善,有利于预报信息的提取;根据巨洪资料序列的性质,用平稳独立随机过程理论建立概率预测模式是合理的,制作长江巨洪发生概率预报是可行的;预计下次巨洪可能在2019年前后发生,2018年发生概率为59%,2019年发生概率为61%。

According to the data series of 150 a（1849—1998） for large flood in Yangtze River basin,and through the statistical analysis of normality and independency,the properties of the data series are determined.Based on the theory for stationary random processes,research prediction of large flood occurrence probability in Yangtze River basin is obtained.Results show that the interval time data series established according to the data samples of large flood in Yangtze River basin from 1849 are improved in linearity and stationarity via logarithmic transformation,which is benefit to the extraction of forecast information.According to the properties of the data series for large flood,it is reasonable to establish possibility prediction modes based on the theory for stationary random processes,and the probability forecast of large flood in Yangtze River basin is feasible.The next large flood will perhaps happen around 2019,and the occurrence possibility of large flood is 59% in 2018 while 61% in 2019.