Evaluation On The Performance Of Various Precipitation Datasets Of Precipitation In The Upper Reach And Mid Reach-Down Reach Of Shule River Basin

Climate change is closely relative to the production and life of the human being.Precipitation which is a significant indicator of climate system,has been widely concerned by national governments and scholars.Water cycle is an important way of substance circulation and energy exchanges of the natural world,in which precipitation is a critical link between the air and the ground.From the simulation of hydrologic processes,it can be seen that the data on precipitation is the core data in hydrologic models.Therefore,how to obtain long-term and accurate information about space-time distribution of precipitation is an issue of observation and science in the study of surface processes and water cycles.Traditionally,relatively accurate data on precipitation are primarily obtained by observing weather stations or hydrometric stations directly,and the information on spatial distribution of precipitation is mainly gotten by spatial interpolation of these data.With the development of science and technology,precipitation space products now include three types: interpolation products of ground-based observed data,satellite remote sensing inversion of precipitation products as well as reanalysis data of data assimilation of ground observation.Though different types of precipitation products can elucidate the space-time characteristics of precipitation to a certain extent,their applicability varies considerably in different regions,and the performance of diverse products of the same type differs enormously in different regions as well.Thus,scientific analysis of the applicability of precipitation products in the local region has become the premise of the application of diverse precipitation products.The Shule River basin lying on the west of the Hexi Corridor,is a vital economic region of the "Silk Road economic belt" and is of great significance to the economic and social development of Northwest China.There are flat oases and deserts in the middle and lower reaches of the Shule River,while there is high and cold mountain area in the upper reaches.The national weather stations in the basin are very limited,especially in the mountainsupstream where no long-term observation data on precipitation are available.Therefore,conducting the research on the applicability of different precipitation products in the Shule River basin is essential to understand the hydrologic processes and manage the water resources in the basin.Based on the observed data on precipitation form nine national weather stations and four self-built stations of the Institute of Ecological Environment and Resources,Chinese Academy of Sciences,in and around the Shule River basin,this paper evaluated the applicability of nine precipitation products(CMFD,CMORPH,ERA-Interim,GPCC,GPCP V2.3,JRA-55,TRMM 3B43,HAR and CAPD)in the upper reaches and seven products(CMFD,CMORPH,ERA-Interim,GPCC,GPCP V2.3,JRA-55 and TRMM3B43)in the middle-lower reaches respectively on the monthly,seasonal and annual scales;compared the applicability of precipitation products between the upper and the middle-lower reaches;discussed the reasons that the performance of diverse products varied in the upper and middle-lower reaches.The main conclusions are as follows:(1)Every precipitation product can basically demonstrate the characteristics of annual distribution and the trend of interannual change of annual mean precipitation in the upper reaches of the Shule River.However,the products named CMORPH and GPCP V2.3performs poorly and underestimates the precipitation remarkably;the products called ERA-Interim,JRA-55,HAR and CAPD overestimates obviously.In terms of the spatial distribution of annual mean precipitation,the principle of spatial distribution of GPCC approaches that of observed data the most,followed by TRMM 3B43 and CMFD.The spatial distribution law of CMORPH differs greatly from that of observed data.(2)All precipitation products can relatively accurately reflect the characteristics of annual distribution and the trend of interannual change of annual mean precipitation in the middle and lower reaches of the Shule River.The trend of interannual change of CMFD is closest to that of the observed data,with TRMM 3B43 behind.The estimation of Jra-55 is much higher than the observed precipitation.Compared with the spatial distribution obtained through interpolation of ground-based observed data in the middle-lower reaches of the Shule River,the estimation of GPCC is the best,followed by TRMM 3B43 and CMFD.The characteristics of spatial distribution of CMORPH are far from those of the observed value.(3)On the monthly scale,CMFD can accurately reflect the monthly precipitation in the upper reaches of the Shule River.The effect of GPCC and TRMM 3B43 is well too,with CAPD,GPCP V2.3 and ERA-Interim closely behind.JRA-55 and HAR overestimates considerably,while CMORPH gets the worst simulation effect.On the seasonal scale,precipitation products have the highest accuracy in autumn,followed by summer,spring and winter.CMFD has the highest precision,closely followed by GPCC and TRMM 3B43.CMFD performs best in spring,CMFD and GPCC does well in summer,GPCC does best in autumn,and HAR performs best in winter.Relatively,CMORPH gets the worst effect in four seasons.On the annual scale,CMFD,GPCC and TRMM 3B43 accurately estimate annual precipitation,with CMORPH performing worst.Among the seven precipitation products,CMFD has the most precise evaluation of monthly precipitation in the middle and lower reaches of Shule River,with GPCC and TRMM 3B43 closely behind.ERA-Interim and GPCP V2.3 has the similar accuracy of the assessment of monthly precipitation.Jra-55 significantly overestimates the monthly precipitation and CMORPH performs worst among these seven precipitation products.On the seasonal scale,the sequence of the accuracy of precipitation products from good to bad is autumn,summer,winter and spring.CMFD has the best simulation effect on the precipitation in the middle-lower reaches on the seasonal scale,followed by GPCC,TRMM 3B43,GPCP V2.3and ERA-Interim,while the effect of CMORPH and JRA-55 is the worst.Relatively speaking,CMFD has higher accuracy in summer and autumn,GPCC performs better in autumn and TRMM 3B43 simulates well in autumn and winter.On an annual scale,the order of the accuracy of evaluation of precipitation products in the middle and lower reaches of the Shule River is: CMFD > GPCC > TRMM 3B43 > GPCP V2.3 >ERA-Interim > CMORPH > JRA-55.(4)From the difference of effect on evaluation among precipitation products at every station,it shows that the accuracy of all precipitation products at S4 Station in the upper reaches is lower than that at any other stations,and the best evaluation is found inYeniugou Station.GPCC,CMFD and TRMM 3B43 performs best at every station.JRA-55 significantly overestimates.ERA-Interim and CMORPH are unable to accurately estimate the precipitation at most stations.CAPD performs well at national weather stations.GPCP V2.3 does well at S4 Station which has low altitude and performs moderately at other stations.In the middle and lower reaches of Shule River,the precipitation products have the worst effect on evaluation of precipitation at Hongliuhe station and Lenghu station.CMFD,GPCC and TRMM 3B43 have notable advantages of evaluation of precipitation,followed by GPCP V2.3 and ERA-Interim.Jra-55 overestimates obviously still,while CMORPH performs poorly again.(5)The sequence of the gradient of precipitation among all products in the upper reaches of the Shule River is: JAR-55 > CMFD > GPCC > CPAD > ERA-Interim > TRMM3B43 > GPCP V2.3 > CMORPH.In the middle and lower reaches of Shule river,there is no obvious gradient of precipitation between the observed data and precipitation products.(6)From the comparison of the accuracy of monthly precipitation data between the upper and the middle-lower reaches of Shule River,the absolute accuracy of evaluation of seven precipitation products(CMFD,CMORPH,ERA-Interim,GPCC,GPCP V2.3,JAR-55 and TRMM 3B43)in the upper reaches is lower than that in the middle-lower reaches,but the relative accuracy is better than that in the middle-lower one.The primary reason is that the precipitation in the upper reaches is much higher than that in the middle-lower reaches.