Precipitation Data Fusion Between Imerg Satellite And Rainfall Station And Its Application In Hydrological Process Simulation

Precipitation is an irreplaceable key element in the process of global water cycle.As an important input parameter of watershed runoff and concentration process,data quality has a direct impact on the accuracy of watershed hydrological simulation.Therefore,it is necessary to obtain a set of precipitation data with high precision and uniform spatial and temporal characteristics.Based on this,this study takes Chengbi River Basin as the research object,discusses the methods to improve the precipitation quality and analyzes its spatial and temporal accuracy.The main contents and achievements of this study are as follows:(1)BIAS,root mean square error(RMSE)and correlation coefficient(CC)were used to evaluate the accuracy of IMERG precipitation data in different time scales(day,month and year).The results show that the accuracy of imerg precipitation data has some errors in daily,monthly and annual scales.CC is the best in the monthly scale,followed by the daily scale and the worst in the annual scale;RMSE is the lowest in the daily scale,followed by the monthly scale,and the largest in the annual scale;BIAS is 0.7%in the daily,monthly and annual scales.(2)Two methods of geographic weighted regression(GWR)and geographic difference analysis(GDA)are used to fuse the measured precipitation at rainfall stations in the basin and IMERG satellite precipitation.The accuracy of the fusion precipitation is verified based on the indicators CC,RMSE and BIAS.The results show that:with GWR,CC is 0.974,RMSE is30.030,BIAS is-0.021;when using GDA,CC is 0.969,RMSE is 32.528,bias is0.015,GWR fusion precipitation is better than GDA fusion precipitation.(3)SWAT hydrological model suitable for Chengbi river basin is constructed by single site method and multi site method.The results show that the hydrological simulation results of multi-station method are better than those of single station method in both daily and monthly scales.Taking the daily scale as an example,the multi site method was used with R~2of 0.887,NSE of 0.856,Re of-0.186;using single site method,R~2was 0.876,NSE was 0.822,Re was-0.215.In the validation period,R~2was 0.800,NSE was 0.743,Re was-0.187by multi site method;R~2was 0.697,NSE was 0.681,Re was-0.218 by single site method.Whether the rate is regular or the validation period,the model calibration based on multi site method can reduce the systematic error of hydrological model and improve the accuracy of the model.(4)On the basis of the model construction,the SWAT hydrological model driven by GWR fusion precipitation data,rainfall station measured precipitation and imerg satellite precipitation is used to simulate the daily and monthly hydrological process.The results show that the accuracy of hydrological simulation using GWR fusion precipitation is better than that of imerg satellite precipitation on both monthly and daily scales.In the monthly scale,R~2of imerg satellite precipitation is 0.809,NSE is 0.660,Re is-0.262;R~2of GWR fusion precipitation is 0.932,NSE is 0.854,Re is-0.206.On the daily scale,the R~2,NSE and Re of imerg satellite precipitation are 0.679,0.534 and-0.325respectively;R~2,NSE and Re of GWR fusion precipitation are 0.791,0.717 and -0.238 respectively.