Spatial Estimation And Impact Factors Of Summer Afternoon And Evening Precipitation In Zhejiang Province Under Complex Terrain

In terms of precipitation estimation,how to consider the complex spatial changes of precipitation and its impact factors has always been a difficulty.Zhejiang province is located in the southeast coast of China,with frequent meteorological disasters and complex terrain(mountains and hills account for about 75% of the total area of the province),and the spatial distribution of precipitation is not uniform,so it is important to accurately estimate the spatial distribution of precipitation and grasp the impact factors of precipitation for the rational use of water resources,disaster prevention and mitigation.In this paper,based on observation data from stations,machine learning models considering spatial heterogeneity are used to estimate the precipitation from afternoon to evening in summer in Zhejiang Province and compare them with the ordinary linear regression(OLR)model to explore the factors influencing the spatial distribution of precipitation during this period,and the following main conclusions are obtained:The summer afternoon precipitation in Zhejiang Province shows a decreasing trend from southwest to northeast,with two precipitation centers in the south and north,forming a precipitation "wet tongue" of southwest-northeast trend.Afternoon precipitation has a certain influence on the spatial distribution of evening precipitation,which is manifested as the similar spatial distribution patterns of both,with a southwest-northeast trending "wet tongue",and a positive correlation between afternoon precipitation and evening precipitation in most areas of Zhejiang Province.The summer afternoon and evening precipitation has obvious spatial heterogeneity in Zhejiang Province.The geographically weighted regression(GWR)model considering spatial heterogeneity,and the geographically neural network weighted regression(GNNWR)model,which uses neural network to consider spatial heterogeneity,are significantly better than the OLR model considering only global relationships,in terms of performance,distribution patterns of error and estimated precipitation.Moreover,the GNNWR model has the highest prediction accuracy for precipitation,showing that the model can accurately estimate the spatial distribution of summer afternoon and evening precipitation in Zhejiang Province.Based on the spatial weight of the impact factors given by the GNNWR model,the impact factors of the spatial distribution of summer afternoon and evening precipitation in Zhejiang Province are further analyzed,and the results show that geographical,topographic and meteorological factors have an impact on the spatial distribution of summer afternoon and evening precipitation in Zhejiang Province.In the afternoon,under the background of the East Asian summer monsoon,the sea breeze blowing from the East China Sea to Zhejiang Province converges with the southwest wind blowing from the highlands to the plains stimulated by the local topographical thermal action,resulting in the airflow convergence of in the north of Taizhou,which promotes the formation of northern precipitation center.There is also a convergence of southerly monsoon and easterly sea breeze along the southern coast areas,and under the dynamic lifting action of windward slope,precipitation is increased and a strong and stable southern precipitation center is formed.For the evening precipitation in Zhejiang Province,it has a certain positive correlation with the afternoon precipitation,showing the continuity of the latter’s impact on the former,but there are also differences: the change of the evening pressure gradient force causes the disturbance of the north wind component in northern and central Zhejiang Province and the westerly land wind disturbance in southern Zhejiang Province,which is favorable to the ground convergence area of afternoon precipitation moving to the southeast coast,and a certain altitude and slope are conducive to the generation and maintenance of the westerly land wind.As the northward and westward components weaken the transport of warm and humid air flow from the sea,the intensity of precipitation in the evening is weakened and distributed closer to the coast compared to the afternoon precipitation.This paper shows that the GNNWR model based on neural network has certain advantages in characterizing meteorological elements with spatial heterogeneity and nonlinear characteristics.It can reproduce the spatial distribution characteristics of summer semidiurnal precipitation in Zhejiang Province and accurately estimate precipitation under complex terrain,which proves the applicability of the method in the field of meteorology.