Study On A Distributed Model Of Monthly Precipitation In China Over The Complex Terrain

China is situated at East Asia monsoon region, where precipitation is one of the most important factors which significantly affect the climate and environment in China. Gridding estimation of precipitation factors engage significant influence on meteorology, climatology and hydrology. Based on previous research on related topic, this study presents a physical experience-based statistic model for the gridding estimation of monthly precipitation at areas with complicated topography in China. In this model, observation from observation stations, data of DEM and TRMM, as well as processed results of CFSR and MERRA are utilized. Following the idea of separation and combination, we split the precipitation into two terms, i.e. global trend and local modification which depends on local topography. The stepwise regression algorithm with consideration of region and time (monthly) is employed in the model. Moreover, data of TRMM, CFSR and MERRA are also effectively modified in the present study. Comparisons with many other models suggest that the newly developed statistic model is generally effective to predict the spatial and temporal distribution of monthly precipitation in China as well as some local characteristics. In addition, the predictions of precipitation from TRMM, CFSR and MERRA are found to be closer to the actual results near surface. In short, our physical experience-based statistic model may provide essential solutions for further development of gridding estimation model of precipitation, and also demonstrate a new approach for modification of precipitation data.The main works in this study are as follows:(1) We performed evaluations of the accuracy of monthly precipitation from TRMM, CFSR and MERRA, which are obtained by linear interpolation according to the longitude and latitude of the stations. It is found that these three data are highly related to the surface precipitation and generally illustrate the spatial distribution of precipitation in China as well as its seasoned variation. However, the predictions in mountainous areas and complicated regions indicate low accuracy. Therefore, results from TRMM, CFSR and MERRA can be utilized as the term of global trend and modification can be supposed according to topography.(2) The concept of dominate precipitation direction and peak precipitation increase direction has been proposed, and the dominate precipitation direction and peak precipitation increase direction of each station can be calculated from surface monthly precipitation and the general slope direction of station. The influence of slope direction to precipitation has been quantitatively analyzed and the vertical increment rate of precipitation under different slope directions has been discussed. Various topography-based modification models of local precipitation have been established, with the consideration of different approaches in constructing the term of altitude. The results indicate that the monthly major and minor precipitation slope can be identified through the cosine of dominant precipitation direction and station slope direction, and the vertical increment rate of precipitation is related to the cosine of slope direction and the direction of peak increment rate is some angular function of the dominant precipitation direction.(3) The physical experience-based statistic model for the gridding estimation of monthly precipitation at areas with complicated topography has been proposed, where the coefficients in the model are calculated using a stepwise regression algorithm with consideration of region and time. Comparisons of our predictions against actual data show that the error of our model for estimation of monthly precipitation is retained within20%and effective modification can be done to basic precipitation data; local characters can also be identified.(4) Results, based on various terms of global trend, of our present model are compared with those from modified common models and modified GUAN model. It is found that estimation models with TRMM as the global trend term demonstrate highest accuracy among other models, and estimation models with consideration of the relation between increment rate and slope direction are quite stable and accurate.