Analysis Of Influencing Factors Of Precipitation In Xinjiang Based On GTWR Model

In recent years,the research on the spatial pattern change of precipitation has attracted extensive attention.Its change is closely related to the safety of ecological environment.There are many factors that affect precipitation,so it is very important to explore how precipitation and its influencing factors change in time and space.At present,many achievements have been made in the study of precipitation,which are mainly reflected in three aspects.First,the precipitation and its influencing factors are analyzed through mathematical modeling.Second,the spatial distribution and change trend of precipitation are analyzed by interpolation method.Third,the prediction of precipitation is studied.However,when analyzing the influencing factors of precipitation in China,the correlation coefficient between precipitation and various factors is calculated,and only the influence of a single factor on precipitation can be analyzed.On the basis of previous studies,this paper will consider the temporal and spatial characteristics of precipitation data at the same time,and use the spatio-temporal geographic weighted regression model(GTWR)to detect the spatio-temporal differences of the effects of multiple influencing factors on precipitation under the temporal and spatial effects.Based on the annual meteorological data of 41 monitoring stations in Xinjiang from 2011 to 2018,this paper selects wind speed,temperature,humidity,elevation,longitude and latitude,and time as the influencing factors of precipitation in Xinjiang from the three angles of meteorological factors,terrain factors,and time factors to analyze the influence of these factors on precipitation.Firstly,the correlation test and multicollinearity diagnosis are carried out on the proposed variables,and finally wind speed,air temperature,humidity and elevation are selected as the factors affecting precipitation.Secondly,using statistical analysis method,descriptive analysis of variables and geographic weighted regression model(GWR)are applied to spatial analysis of precipitation and various factors.Finally,a spatio-temporal geographic weighted regression model of precipitation is constructed to reveal the non-stationary relationship between precipitation and various factors in spatio-temporal.Experimental analysis results:(1)from 2011 to 2018,the station with the highest precipitation was Tianchi,with the precipitation reaching 893.50 mm,and the station with the lowest precipitation was Turpan,with the precipitation only 7.70 mm,and the standard deviation of precipitation was 130.60,indicating that the precipitation of each station was quite different;The standard deviation of wind speed,air temperature and humidity are 0.80,3.51 and 8.82 respectively,which indicates that the difference of wind speed,air temperature and humidity at each station is relatively small.However,the standard difference of elevation 482.57 shows that there is a great disparity in elevation at each station.(2)Stations with higher precipitation,such as Tianchi,Urumqi Animal Husbandry Test Station,Tacheng and Tuoli,are mostly distributed in the north and east of Xinjiang,while stations with lower precipitation,such as Brakes,Hotan,Pishan and Yutian,are mostly distributed in the south of Xinjiang.Therefore,the precipitation in the north of Xinjiang is higher than that in the south of Xinjiang,and the precipitation in Xinjiang is mainly distributed in the north and the south.The global spatial autocorrelation of precipitation shows that the Moran values from 2011 to 2018 are all greater than 0,which indicates that the precipitation has a positive correlation in space,and in 2016,the Moran value is the highest 0.3248,which indicates that the precipitation of each station has the strongest positive correlation in space in 2012.(3)Considering the spatial effect,the influence of wind speed,air temperature,humidity and elevation on precipitation is analyzed.It can be seen that the regression coefficient of wind speed is between-0.42 and-0.14,and the regression coefficient of wind speed at different stations is different,that is,the influence of wind speed on precipitation at different spatial positions is non-stationary;The regression coefficient of air temperature is between-1.10 and-0.77,and the regression coefficient of humidity is between 1.34 and 2.40.The regression coefficients of air temperature and humidity at different stations are different and have certain volatility.The regression coefficient value of elevation is between 0.16 and0.32,with a large difference,the influence of elevation on precipitation at different stations shows more non-stationarity.From this,it can be seen that different geographical locations and various factors have different impacts on precipitation,and the relationship between various factors and precipitation is not stationary in space.(4)Considering the time effect and the space effect,analyzing the influence of wind speed,temperature,humidity and elevation on precipitation,we can find that the influence of various factors on precipitation is non-stationary in time and space.Among them,the stations with positive effect of wind speed on precipitation are mostly distributed in the east of Xinjiang,and the stations with negative effect on precipitation are mostly distributed in the east and south of Xinjiang.Stations with positive effect of temperature on precipitation are mostly distributed in the south of Xinjiang and stations with negative effect on precipitation are mostly distributed in the east of Xinjiang.On the whole,humidity has a positive effect on precipitation.Only a few stations near the Kunlun Mountains show a negative effect on precipitation.On the whole,the elevation has a positive effect on precipitation,only a few stations are distributed in the stations near Tarim basin,and elevation has a negative effect on precipitation.With the change of time,the influence of various factors on precipitation at different stations will also change.(5)Comparing the results of GWR model and GTWR,it is found that the AIC value of GTWR model is87.59,and the AIC value of GWR model is 102.32.The smaller the AIC value,the better the fitting data of the model,so GTWR model is better than GWR model.The goodness of fit of GTWR model is 0.93,and that of GWR model is 0.78,which also shows that GTWR model has the best fitting effect.Therefore,GTWR model can be used to better analyze the temporal and spatial differences of the influence of various factors on precipitation.Studying the temporal and spatial differences of wind speed,temperature,humidity and elevation on precipitation in Xinjiang can summarize the laws of these factors on precipitation in different regions,thus providing the basis for making flood prevention and flood fighting decisions in different regions of Xinjiang.