Research On Precision Evaluation And Error Attribution Of Various Satellite Precipitation Products In China's Mainland

This paper uses the rain gauge data provided by the National Meteorological Information Center of the China Meteorological Administration as the benchmark data to evaluate the accuracy of TRMM 3B43,global precipitation measurement program(GPM)multi-satellite integrated precipitation product(IMERG)and the global satellite precipitation mapping product(GSMaP).In addition,the error source of IMERG and GSMaP was analyzed using the error component method.The main conclusions are as follows:(1)The precipitation estimated by 3B43 V7 has a good correlation with the rain gauge data at various time scales.3B43 V7 has the worst performance in winter and has obvious overestimation every month.In terms of space,3B43V7 performs better in southeast China than in northwest.Amount of precipitation has a great influence on the performance of 3B43V7.In areas with low amount of precipitation,the error is large.3B43 V7 has better product performance under 3500m.(2)All products of IMERG and GSMaP perform well on the monthly scale,with similar time trends.Compared with other products,GSMaP_G has the best performance on all time scales.GSMaP_G has better precipitation event detection capabilities than other satellite precipitation products,especially in winter.All products of IMERG and GSMaP have obvious spatial variability.All products have similar spatial trends in CC and RMSE,with higher values in southeast China and lower values in northwest China.After specification correction,GSMaP_G and IMERG F show better performance than other products in most regions,especially the former.All products can reproduce the probability density distribution of daily scale precipitation intensity.With the increase of precipitation intensity,the consistency of satellite precipitation observation products and rain gauge observations gradually decreases,RB gradually decreases,and precipitation detection capabilities gradually improve,especially GSMaP_G.All products perform better at low altitudes than at high altitudes.GSMaP N has a strong dependence on terrain,and its performance decreases rapidly with the increase of altitude.The precipitation detection capabilities of all products performed well in southern China,especially in the southeast region.The result of the ground data correction product is not very satisfactory in summer,but it shows strong precipitation detection ability in winter.(3)Hit bias is the most important factor causing the total bias.Various indicators present a spatially low south to north high distribution trend.In winter,satellite products mainly have large errors in the Yangtze River Basin of China,showing an underestimation.In the Yangtze River Basin,there are large errors in missed precipitation and false precipitation.Except for GSMaP_G,other products cannot reduce the errors caused by precipitation from various error sources.In summer,the products that have not been corrected by ground rain gauge data are obviously affected by the elevation.With the increase of altitude,the total bias and hit bias show an increasing trend,and the total bias is mainly affected by the hit bias.Gauge correction algorithm can effectively reduce the error caused by elevation.In winter,the error of IMERG products in high-altitude areas is smaller than that in low-altitude areas,and when the altitude is greater than 3000 meters,the missed precipitation and hit bias tend to zero,and the total bias is mainly due to the false precipitation.IMERG_F and GSMaP_G have reduced terrain errors well in both seasons,but there may be overcorrection.The change curve of satellite products in summer has more characteristics than winter.A common feature in both seasons is that most of the missed and false precipitation is concentrated in the range of precipitation intensity less than 16mm/d.This study will provide theoretical support for the application of satellite precipitation products to hydrological research and water resources management in China,and provide some reference for algorithm developers in the process of algorithm optimization.