Merging Method Of Precipitation Based On Multi-source Remote Sensing Data And Ground Observation In Xinjiang

Precipitation is related to the latent heat released in an atmospheric column above the surface.Precipitation processes redistributed the differential solar energy received at the Earth's surface into the atmospheric interior to drive the large-scale circulation.By coupling with cloud,water vapor,atmospheric and oceanic circulation,soil moisture and surface albedo,atmospheric precipitation profoundly changes the energy balance of the earth.Ground observation is the most direct precipitation observation data,but most stations are located in low altitude areas,especially in Xinjiang,where the observation stations are sparse and unevenly distributed,which makes it difficult to express the actual precipitation information of the whole basin.With the rapid development of remote sensing technology,the retrieved satellite precipitation has become an important way to systematically understand regional and global precipitation and its changes.However,due to the limitation of the physical principle and algorithm of the precipitation retrievel,the quality of satellite precipitation is not high.How to fully integrate the respective advantages of ground and satellite observation precipitation data is the focus of current research.This paper takes Xinjiang as the research area to carry out the merging method of satellite precipitation and ground observation.The main work and conclusions are as follows:(1)Evaluation of multi-satellite precipitation products in Xinjiang,ChinaSatellite precipitation retrieval is a critical approach to understanding the spatial distribution of precipitation in Xinjiang,an arid area located in Northwest China,where weather stations are sparsely distributed.However,satellite precipitation retrieval in arid areas is a challenging task.The goal of this part is to evaluate the estimates of four satellite precipitation products,namely,the Global Satellite Mapping of Precipitation(GSMa P),Integrated Multi-satellit E Retrievals for Global Precipitation Measurement(IMERG),Tropical Rainfall Measuring Mission(TMPA)Multi-satellite Precipitation Analysis3B42/3B43(3B42/3B43)and Climate Prediction Center Morphing Technique(CMORPH),on half-hourly,hourly,3-hourly and daily scales based on rain gauge data.The findings of this study are as follows.1)The four products generally display a declining trend from northwest to southeast.IMERG exhibits a higher accuracy than CMORPH for all indexes at the half-hourly scale,while GSMa P performs better than other products based on most indexes at hourly and daily scales.2)In three sub-regions,i.e.,Southern Xinjiang(SX),Northern Xinjiang(NX)and the Tianshan Mountains(Tianshan),these products exhibit significant regional characteristics.The precipitation in SX,where rainfall observations are scarce,is overestimated by all products;in contrast,all products underestimate precipitation in Tianshan in NX,except for the underestimation by 3B42,precipitation was overestimated by the studied products.3)All satellite products performed better in the warm season than in the cold season at each time scale.During the warm season,apart from the apparent overestimation by CMORPH,the relative bias values of the other products are all within ±10%.During the cold season,all products underestimate precipitation mainly composed of snowfall,especially3B42,which yields the most underestimated values.4)IMERG performs well in the retrieval of the distribution of the probability density function(PDF)of the occurrence(PDFc)of gauge observations,especially at low precipitation intensities,and the difference between the estimated and observed precipitation volumes at the hourly scale is the smallest.However,GSMa P performed better at the daily scale according to the PDF for the volume of precipitation(PDFv).This study is the first to evaluate IMERG and CMORPH products at the half-hourly scale and is one of the few sub-daily evaluations of various satellite precipitation products in arid areas of China.Thus,our results provide a significant reference for the satellite retrieval of precipitation in arid areas.(2)Correction of the GPM 3IMERGM satellite precipitation product based on stepwise regression over Xinjiang,ChinaSatellite retrieval of precipitation is a great challenge in arid regions where light precipitation prevails.To improve the accuracy of satellite precipitation products in Xinjiang,which is the driest region in China,this study based on the evaluation of the performance of two widely used monthly satellite products,which are the Tropical Rainfall Measuring Mission(TMPA)3B43 version 7(hereafter 3B43V7)and the Integrated Multi-Satellite Retrievals for the Global Precipitation Measurement(GPM)mission(IMERG)(hereafter3IMERGM)respectively,corrected the 3IMERGM product by a stepwise regression model using topographic variables derived from digital elevation model(DEM).To comprehensively evaluate the performance of the corrected model,a tenfold cross validation method was used.The results showed that the corrected 3IMERGM(C-3IMERGM)performed much better than3 IMERGM.Specifically,CC was increased from 0.68 to 0.73,and RB is decreased from 7.76%to-1.65%.Furthermore,C-3IMERGM achieves a better precipitation distribution than the uncorrected satellite product and even than scarce gauge measurements.The actual spatial pattern of precipitation represented that the precipitation bands in the Kunlun Mountains located in southern Xinjiang were captured by C-3IMERGM,but missed by the other products.(3)Research on bias correction of TMPA 3B43 precipitation data based on BP Neural NetworkUsing 1998-2013 monthly TMPA Precipitation data and the observation precipitation data from 105 national basic weather stations in Xinjiang region,a stepwise regression model and a back-propagation(BP)neural network were established to correct TMPA precipitation.The results show that the models added geographical factors can increase the accuracy of TMPA precipitation effectively.Corrected by the two models,the overall correlation coefficients were 0.75 and 0.8,and the relative error were 4.88% and 3.19%;on the monthly-scale,the relative error range of TMPA monthly precipitation is-5.68%—54.44%,and stepwise regression for-4.26%—32.57% and neural network for-5.33%—24.48%respectively;In addition,the results showed that the qualities of the satellite precipitation products were improved in different degrees from the ST indices,with 0.01-0.49 for stepwise regression and 0.03-0.7 for neural network,respectively.Compared with the TMPA data before correction,the stepwise regression and BP neural network model can accurately and quantitatively reproduce the actual distribution of precipitation,and provide a more practical method for the area lack of the precipitation data.(4)Study on the bias correction of GPM IMERG precipitation data based on Geographically Weighted Regression over the Tianshan Mountains in ChinaPoint-scale gauge observations have inherent limitations,making it difficult to study the spatial and temporal distributions of precipitation in alpine regions due to gauge undercatch and complex terrains.The Global Precipitation Measurement(GPM)mission provides the new-generation satellite precipitation products which are promising alternative data sources in mountainous areas.However,quality evaluations and bias corrections should be conducted prior to the application of satellite data.In this study,over 1000 automatic weather stations(AWSs)compose an unprecedentedly dense ground station network over the Tianshan Mountains in China for bias correction.Firstly,the universal kriging interpolation is used to downscale the Integrated Multisatellite Retrievals for GPM(IMERG)product to 500 m to make sure a fair comparison with gauge observations.Then,two correction methods,i.e.,the stepwise regression(STEP)and geographically weighted regression(GWR),were applied to improve the accuracy of the IMERG products over this region.Both methods are established on various terrain factors and vegetation indexes that have strong relations with precipitation.The results show that(1)the corrected results using GWR outperform the conventional STEP method as well as the original IMERG data;(2)for the elevation assessment,the original IMERG performs best over the plain region(less than 1,000 m),while the best correction effect was found in middle and low-altitude areas(1,000-1,500 m);and(3)the performance of the GWR model is largely dependent on the number of available training stations in mountainous areas with complex terrains.Overall,the methods and results presented in this study provide insight into the correction of satellite precipitation data in mountainous areas without the need of ground observations.