Evaluation Of MSWEP Precipitation Data And Their Utility In Hydrological Simulation Over Yuanjiang River Basin

Precipitation is an important part of the water cycle.Accurate and reliable precipitation data is not only the key to study the spatial and temporal changes of precipitation,but also an important factor to improve the accuracy of hydrological simulation.Traditional rain gauges provide relatively accurate precipitation data.However,due to the high spatial difference of precipitation,the site observation data with uneven distribution and sparse distribution are difficult to meet the needs of hydrological models and other related researches.Satellite precipitation data have the characteristics of wide coverage,high spatial and temporal resolution,but there is often an overestimation or underestimation of the actual precipitation,especially in areas with complex terrain and large topographic changes.Multi-Source Weighted Ensemble Precipitation(MSWEP),which integrates rain gauges precipitation data,satellite precipitation data and reanalysis data,fully combines the advantases of site data and satellite data,and has the characteristics of long time series,high precision and high space-time resolution,providing relatively reliable Precipitation data on the global scale.In this study,The Yuanjiang River Basin in Yunnan province was selected as the research area.The accuracy test and hydrological simulation effect of MSWEP precipitation data were evaluated to explore the applicability of MSWEP precipitation data in this region,so as to provide a basis for hydrological process simulation,hydrological disaster monitoring and basin water resources planning.Firstly,precipitation data from meteorological stations were used to test the accuracy of MSWEP precipitation data from 1979 to 2015 on different spatial and temporal scales.On this basis,based on GR4J and GR2M hydrological models,the hydrological simulation effect of MSWEP precipitation data is evaluated on daily and monthly scales The main conclusions are as follows:(1)The accuracy test results of the MSWEP precipitation data show that the MSWEP precipitation data have a high consistency with the measured precipitation data at the meteorological station on the annual,seasonal and monthly scales(annual scale R=0.81,monthly scale R=0.89,spring R=0.88,summer R=0.77,autumn R=0.75,and winter R=0.81).On the daily scale,the performance of MSWEP precipitation data is poor,and the correlation coefficient R between the MSWEP precipitation data and the site precipitation data is 0.43.On the whole.MSWEP precipitation data was somewhat underestimated in the Yuanjiang river basin(Bias=-8.5%).In terms of space,the gravity center movement trajectory of MSWEP precipitation data is basically,consistent with that of the site,which can reflect the spatial distribution and evolution process of precipitation.On site scale,the correlation coefficient R of MSWEP precipitation data and site precipitation data is greater than 0.88,and the |Bias|of 80%sites is within 20% In terms of the detection ability of precipitation/non-precipitation events,the probability of detection of 88%sites is above 0.7.and the false alarm ratio of 93%sites is below 0.4.,indicating that MSWEP precipitation data has a strong ability to detect precipitation/non-precipitation events.(2)On the basis of GR4J and GR2M hydrological models,the hydrological simulation effect of MSWEP precipitation data and the precipitation data of the site was evaluated on the daily and monthly scales.The results show that the hydrological model driven by the precipitation data observed at the site performs well in simulating the flow process(daily scale:NSE = 0.82,Bias = 0.6%;Monthly scale:NSE = 0.87,Bias =1.8%).indicating that GR4J and GR2M hydrological model have good applicability in Yuanjiang River Basin.In scenario one(calibration model by using precipitation data observed at the site),the NSE and Bias of the hydrological model driven by MSWEP precipitation data were 0.73 and 0.84,respectively,at the daily scale and the monthly scale.The overall simulation effect was good and the flow process could be basically reproduced.In scenario two(calibration model by using MSWEP precipitation data),the flow simulation effect of the hydrological model driven by MSWEP data is better than that in scenario one.On the daily and monthly scales,the NSE was increased to 0.78 and 0.85 respectively,and the Bias was 2.4%and 1.3%,respectively.In extreme hydrological events,the ability of MSWEP data to simulate extreme flow is poor,the simulation of high flow(Q90)is underestimated(-17.6%),and the simulation of low flow(Q50)is overestimated(33.5%).To sum up,MSWEP precipitation data has a good application prospect in areas where data are scarce in the Yuanjiang River Basin,especially in long-term water resource planning and hydrological drought monitoring in the basin.