The Analysis Of Change-point In Meteorological Data

As an important topic of statistics,the analysis of change-point has been widely usedin finance,quality control and meteorology fields.In this dissertation,cross-section and panel data change-point models are developed to analysis structural breaks and we distinguish climate factor change-point and non-climate factor change-point that are estimated in meteorological data.In the first chapter,we introduce the research significance of the change-point,and describe the research of change-point in meteorological field.Then we briefly introduce the problem of single change-point,multiple change-point,common change-point and the current research of change-point method in climate.In the second chapter,we briefly introduce the theory and algorithm steps in Bayes method,Monte Carlo method,M-H algorithm and ASAMC algorithm which related to the analysis of change-point in meteorological data.In the third chapter,according to monthly mean temperature series the position of change-points were estimated by Annealing Stochastic Approximation Monte Carlo method in multiple meteorological stations of Anhui Province from 1957 to 2015.We also explore climate and non-climate factors that influenced structural at each station.In the fourth chapter,a panel data is made up by the monthly average temperature series of multiple stations in Chapter 3.We use the common change-point test statistic to estimate the common breaks of panel data for statistical analysis with the existing metadata records.In the fifth chapter,we applied to the theory of simple linear regression change-point model based on the three-parameter Weibull distribution to estimate the change-point of annual maximum wind speed from 1980 to 2016 in Anhui Province,and then study the results of change-point with historical wind speed information.In the chapter sixth,we summarize the research of statistics in this dissertation.