Evaluation and Error decomposition of multisource precipitation data in an alpine and endorheic river watershed
The quality of precipitation data are critical factor influencing the accuracy of runoff simulation in high-cold mountainous districts as it plays an important role in the ecological environmental protection and water resource management.The spatiotemporal characteristics of precipitation are analyzed in the headwater catch-ment of the Yarkant River Basin on the basis of GPM(Global Precipitation Measurement),AIMERG(the Asian precipitation dataset by calibrating the GPM-era IMERG),CMFD(China Meteorological Forcing Dataset)and ERA5(The fifth-generation atmospheric reanalysis of the European Center for Medium-Range Weather Fore-casts).Subsequently,the accuracy of the multisource precipitation data are evaluated against the observed precipi-tation.The error characteristics of various precipitation products was analyzed by means of the error decomposi-tion model.The main findings were as follows:(1)The spatial pattern for CMFD and AIMERG was character-ized by the increase from the north to south,which was consistent with the spatial pattern for the grid observation data set CN05.1(the National Climate Center of China Meteorological Administration precipitation dataset).An opposite pattern was detected for ERA5 and GPM.Additionally,AIMERG and CMFD displayed higher precipita-tion in the glacier area.(2)The inter-annual variation characteristics of various precipitation products were signifi-cantly different,and the ratio of summer and autumn precipitation to annual precipitation for most precipitation products was more than 60%.Among all the precipitation products,only AIMERG reproduced the seasonal pat-terns,such as the time when the maximum monthly precipitation occurred and the peak shape for the monthly pre-cipitation at all stations.AIMERG had the greatest ability to reproduce gauged monthly precipitation,with a high-er correlation coefficient(>0.6)and lower root mean square error(8.45-11.57 mm),whereas ERA5 show the poorest ability.(3)All precipitation products showed a higher performance in reproducing daily precipitation dur-ing the wet period(from May to October)than during the dry period(from November to April).AIMERG had a greater critical success index in both wet period and dry period than for other precipitation products.(4)The dom-inant error of the various precipitation products in summer was the hit error,whereas the dominant error in winter varied with the precipitation product.These findings provide some reference for the runoff simulation and algo-rithm improvement of precipitation products in the high-cold region,where meteorological data are limited.
reanalysis datasatellite precipitationAIMERGerror decomposition modelupper Yarkant River
万方数据
维普
