论文部分内容阅读
针对不符合正态分布的水文时间序列,建立用于时间序列变点分析的贝叶斯数学模型.研究了某水文站1942—2008年共67年年径流最大值时间序列的突变,得到1989年是最大可能变异点.贝叶斯数学模型中仅给出了变点位置的后验概率,但是最大后验概率时间位置处是否确实发生变异,则有待进一步讨论.因此文中认为需要进一步检验1989年前后的变化是否在随机波动范围内或者是确实发生了变异.文中采用再抽样自助法对结果进行深入的可靠性分析,证实1989年前后确实发生了突变,变异程度远远超过了随机波动范围.考虑到时间序列的随机性,建议在今后工作中采用贝叶斯变点分析模型确定变异点时,采用自助再抽样方法对结果进行可靠性分析.
A Bayesian mathematic model for time-varying change-point analysis is established for hydrological time series that do not fit the normal distribution. A sudden change in the annual runoff maximum time series of a hydrological station from 1942 to 2008 was studied and obtained in 1989 Is the maximum possible variation.By the Bayesian model, only the posterior probability of the position of the change point is given, but whether the variation of the maximum posterior probability position is indeed to be further discussed.It is therefore believed that further examination of the 1989 Whether the changes before and after is within the range of random fluctuations or whether there is indeed a mutation occurs. In this paper, the re-sampling self-help method is used to conduct in-depth reliability analysis of the results, confirming that abrupt changes did occur before and after 1989 and the degree of variation far exceeded the range of random fluctuations. Considering the randomness of the time series, it is suggested that the Bayesian change point analysis model be used to determine the variation point in future work, and the self-help resampling method is used to analyze the reliability of the results.