Study On Extreme Precipitation Charactristics And Precipitation Prediction Model Construction In Guangxi

In recent years,under the background of global warming,the researchs on extreme precipitation has been a hot topic;and Guangxi belongs to subtropical monsoon climate,which is affected by monsoon,land sea interaction and underlying surface,which makes extreme precipitation events frequent in Guangxi.Extreme precipitation leads to more uneven distribution of precipitation in Guangxi.At the same time,due to the karst development and poor water holding capacity of soil layer,natural disasters such as flood and drought are more frequent.Moreover,the occurrence of extreme precipitation has obvious regional,local and seasonal characteristics,and its disaster mechanism is not clear,and the cause of formation is complex.Based on this,in view of the shortcomings of the previous research on extreme precipitation in Guangxi,this paper uses the hourly and daily meteorological data of 82 meteorological stations in Guangxi from 1961 to 2016 to study the spatiotemporal variation characteristics,heterogeneity analysis and correlation analysis of extreme precipitation at annual and seasonal scales,and constructs a precipitation prediction model,which provides a theoretical basis for disaster prevention and mitigation in Guangxi in the future.The main conclusions are as follows:(1)The threshold of extreme precipitation in Guangxi is mainly characterized by decreasing from coastal areas to inland areas;the heavy precipitation(R95p),frequency and severity(EPSI)are all in an increasing state with the increase of time,while for the average intensity,except for zone IX(coastal climate zone),the other climatic regions are in a rising state;through mutation test,each index of each climate zone is obtained In space,R95 p decreases first and then increases from southeast to northwest,and frequency decreases from northeast to southwest.The average intensity mainly decreases from Central Guangxi,and the severity decreases from south to north.The first three modes of R95 p are obtained by EOF analysis According to the EOF analysis of EPSI,the first mode is the North-South reverse type.(2)Based on the analysis of the non-uniformity of extreme precipitation in Guangxi,it is concluded that the concentration degree of the nine climatic regions has an increasing trend,except for the concentration degree of zone I(Northwest Guangxi climate zone)and zone VII(Southern Guangxi climate zone),the concentration degree of other climatic regions shows a decreasing trend;the concentration periods of nine climatic regions are zone I(Northwest Guangxi climate zone),zone II(Northern Guangxi climate zone)and zone IV(The results show that there is a trend of slow delay in Southwest Guangxi,zone Ⅴ(Central Guangxi),zone VII(Southern Guangxi)and zone IX(coastal climate),and the concentration of gas in three(Northeast Guangxi climatic region)and VI(Southern climate area of Northeast Guangxi)regions is earlier;the concentration of nine climatic regions is abrupt,except that there is no mutation in areas IV,VI,VII,VIII and IX zones.In terms of space,the concentration degree showed a decreasing trend from west to east,and the concentration period showed a decreasing trend from southwest to northeast;the contribution rate of r95 p in nine climatic regions was basically the same,except for IX,the other climatic regions were in an increasing state,and the occurrence rates of nine climatic regions showed an increasing trend.(3)From coastal to inland,Guilin,Hechi,Mengshan,Laibin,Longzhou,Nanning,Yulin and Beihai are selected as typical areas to analyze the correlation between extreme precipitation,temperature and tropical cyclone in different areas of Guangxi.Copula analysis is conducted to obtain Guilin winter,Hechi summer,Mengshan spring,Mengshan summer,Mengshan autumn,Laibin autumn,Longzhou spring,Yulin summer and Beihai summer There is a certain correlation between the seasonal extreme precipitation and temperature,which can be characterized by Clayton copula function;there is a certain correlation between the extreme precipitation in different regions and the daily maximum wind speed of tropical cyclone;there is a certain correlation between the precipitation and the daily maximum wind speed in eight typical regions;the extreme precipitation and daily maximum wind speed in Laibin and Beihai can be represented by Clayton Copula function.(4)Through the establishment of PSO-SVM +BP neural network precipitation prediction model,in the absence of tropical cyclone,the MSE and average relative error of the combined model of temperature and precipitation are 0.013 and 0.098%,respectively;when there is a typhoon,the MSE and average relative error of the combined model are 0.18 and4.868%,respectively Compared with other single prediction models,the prediction accuracy of PSO-SVM + BP model is improved,which overcomes the blindness of SVM prediction model.