基于贝叶斯估计算法的降水变点研究  

Change-point Analysis of Precipitation Based on Bayesian Estimation Algorithm

作　　者：雷鑫昊 杨霖 LEI Xinhao;YANG Lin(School of Mathematics and Statistics,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]福州大学数学与统计学院,福建福州350108

出　　处：《福建商学院学报》2024年第5期45-50,共6页Journal of Fujian Business University

基　　金：福建省自然科学基金“判别检验理论及其在泊松过程参数模型的应用”(2021J05018)。

摘　　要：降水分布情况影响农业、畜牧业等产业的发展。近年的研究侧重使用Mann-Kendall(M-K)突变检验分析各地区降水数据的突变情况。本文旨在提供一种新的突变分析手段,利用贝叶斯估计算法分析福建省、海南省、贵州省、河南省、黑龙江省和内蒙古自治区1960—2016年年均降水事件的突变规律。将降水数据转化为年均降水量高于其均值的降水事件,假定累积事件数服从泊松过程,通过判断其强度函数是否存在变点,分析该地区降水的突变情况。结果表明,除河南省外,其他五个省和自治区年均降水均存在变点;贵州、内蒙古和黑龙江在突变年前后有明显变化规律,存在降水量偏多和偏少交替出现的风险。应考虑加强水利建设、完善水资源智能管理平台和布局地下储水系统等以应对风险。The precipitation affects the development of the industries,such as the agriculture and animal husbandry.In recent years,the Mann-Kendall(M-K)mutation test became the major technique to analyze the mutations in the precipitation data.In this paper,we aim to propose a new technique of the mutation detection for the precipitation data.The Bayesian estimation algorithm is used to analyze the mutation of the annual mean precipitation events in Fujian,Hainan,Guizhou,Henan,Heilongjiang Province,and Inner Mongolia Autonomous Region from 1960 to 2016.The accumulative precipitation event in this paper represents the number of year when the annual mean precipitation is higher than its mean in some specified period.Assuming that the accumulative precipitation event sequence follows a Poisson process,we analyze the mutation by determining whether the intensity function contains a change-point.The results show that,a mutation exists in the provinces analyzed except the Henan province.Guizhou,Inner Mongolia,and Heilongjiang provinces have obvious mutation,accompanied with the risk of alternating drought and water-logging.Based on the results,we propose the corresponding suggestions,such as strengthening water conservancy project,improving the intelligent management platform for water resources and the underground water-storing system,for the previous provinces.

关 键 词：降水事件 变点 泊松过程 贝叶斯估计算法 年均降水量数据 

分 类 号：P333[天文地球—水文科学]