Spatial And Temporal Characteristics Of Precipitation And Its Influence Factors

Precipitation is an important research subject in meteorology,and its quantity greatly affects human production and life.Excessive precipitation can cause natural disasters such as floods and waterlogging.Therefore,the systematic study of the spatial distribution characteristics,meteorological influencing factors,and temporal variation characteristics of precipitation in China plays a crucial role in the reasonable allocation and utilization of water resources.In this study,we used the daily value dataset(V3.0)of basic meteorological elements from China’s national surface meteorological stations from 1951 to 2016 to explore the characteristics of precipitation change in terms of spatial distribution,meteorological influencing factors,and temporal distribution.The present study investigates the spatial distribution characteristics of precipita-tion by using the annual total precipitation vector data of each meteorological station as the research data.Firstly,the daily precipitation data of the meteorological station are preprocessed by supplementing the missing values.The total annual precipitation of each station is then calculated,forming a 65-dimensional vector data for each sta-tion with the total annual precipitation as the component.The K-means clustering algorithm,PAM clustering algorithm,and DBSCAN clustering algorithm are used to cluster and analyze each station,and the clustering results are basically similar,indicat-ing that the clustering is reasonable.The precipitation contour map of China is drawn,and a confusion matrix is established to compare the accuracy of the three clustering results,and the K-means clustering algorithm is found to be the most effective.Fi-nally,according to the K-means clustering results and the latitude and longitude data characteristics,China is divided into six categories of regions:central and southern regions,Hainan region,northwestern region,southeastern region,northeastern region,and central region.The results of the spatial clustering characteristics are used to study meteorological factors and time characteristics accurately.To study the meteorological factors affecting precipitation,the average annual to-tal amount of meteorological data for each type of station is taken as the research object.Firstly,the annual average value of meteorological data excluding precipitation is calculated,and the correlation coefficient between different meteorological data and precipitation is calculated.Then,the meteorological data of temperature,humidity,and evaporation,which are strongly correlated with precipitation data,are extracted according to the size of the correlation coefficient.Then,panel data regression analysis is carried out using the precipitation,temperature,humidity,and evaporation annual average data to analyze the panel data of various regions based on the stations in-cluded in each region,obtaining the regression model of meteorological factors that affect precipitation in each region.The panel data regression models of each region have an R~2value higher than that of the model established using all stations nation-wide,which can more clearly and accurately characterize the influence of meteorological factors on precipitation.The following conclusions are drawn:the overall precipitation in China is mainly affected by temperature,humidity,and evaporation,and precipita-tion is positively correlated with temperature and negatively correlated with humidity and evaporation;in terms of regional discussion,wind speed has no obvious effect on precipitation in various regions,and air pressure has the most significant impact on precipitation in the Hainan region,while evaporation has the most significant impact on precipitation in the central region.The panel data analysis of meteorological factors demonstrates the usefulness of spatial clustering in the study of precipitation and pro-vides a theoretical basis for the analysis of temporal changes in precipitation in various regions.To study the temporal variation characteristics of precipitation,the monthly pre-cipitation data of the central station of each type of region obtained by the K-means clustering algorithm are used as the research data.Based on the precipitation data of the clustering center stations of each region,the monthly average precipitation days are qualitatively calculated for seasonal division.The variance coefficient is calculated for the traversal division of monthly precipitation and the quantitative seasonal division is performed.Interval estimation is performed on the monthly precipitation variation to test the rainy season interval and confirm the rainy and dry seasons in each region.The stationarity test and periodicity judgment are performed on the precipitation series of each season of each region.The results show that there is an interannual periodicity between the seasons and each series type is a non-stationary periodic series suitable for using the STL decomposition model.The STL decomposition is performed on each series to extract the seasonal component S_t,the trend component T_t,and the residual component L_t.The Fourier least-squares estimation and linear least-squares estima-tion are performed on the seasonal component S_tand the trend component T_t,and the normality test and white noise test are performed on the residual component L_t.For the sequences that fail the test,the VAR model is used to further extract information.After confirming that the residual is white noise,Monte Carlo simulation is used to obtain the distribution and predicted value of the residual component L_t.The model is used to predict precipitation in 2017 and the prediction effect of the model is compared with that of the SARIMA model and the STL decomposition model without seasonal division,demonstrating the feasibility of the model prediction.Based on the results of the seasonal component S_tand the trend component T_tcombined with meteorological factors,the annual precipitation and interannual rainy season precipitation changes are analyzed.The results show that all regions of China show a trend of first increasing and then decreasing in the annual precipitation,and precipitation is mainly concentrated in the rainy season,with the highest precipitation occurring from June to September.The interannual variation of rainy season precipitation in all regions of China generally shows an upward trend.For different regions,the rainy season precipitation in the cen-tral and northwest regions is stable,while that in the central-southern,southeastern,northeastern,and Hainan regions is more active,with the rainy season precipitation in Hainan showing a decreasing trend.