Accuracy Assessment And Hydrological Simulation Of Multi-source Satellite Precipitation Products In The Yanhe River Basin

As an important component of the global water cycle,precipitation plays a vital role in meteorology,hydrology and ecology.However,the traditional meteorological stations in the arid and semi-arid regions of northwest China are relatively sparse,and the lack of high-precision observed precipitation data restricts the development of related research.Satellite precipitation products are a kind of data product established by meteorological satellites to extrapolate ground precipitation from the detection of precipitation clouds,which has the advantages of large observation range,long observation time and continuous dynamic changes,etc.It has been widely used as a reference precipitation information for climate research in areas with sparse ground meteorological stations.In this paper,the accuracy and error characteristics of three satellite precipitation products,CMADS,CMFD and MSWEP,are evaluated from different temporal and spatial scales based on the measured precipitation data from 37 rainfall stations in the Yanhe River basin as the study area,and their detection capabilities for precipitation/non-precipitation events are explored.Finally,the SWAT hydrological model is driven by the measured precipitation data and the three satellite precipitation products,and the hydrological simulation effects of different precipitation data are compared and analysed to investigate the applicability of different satellite precipitation products in the Yanhe River basin.The main conclusions are as follows.(1)The results of the accuracy assessment of the satellite precipitation products show that the accuracy performance of each satellite precipitation product varies greatly under different time scales.On the daily scale,the MSWEP product performs best overall,while the CMFD product underestimates precipitation in winter and spring months when precipitation is low,and performs the worst;the MSWEP product also performs best in terms of precipitation event detection with low false alarm rate FAR,high detection rate POD and success rate CSI.The accuracy of the satellite precipitation products at the monthly scale has improved significantly compared to the daily scale,with the MSWEP product having the highest accuracy,followed by the CMFD product and the CMADS product being the worst.The accuracy of the annual scale precipitation products decreases compared to the monthly scale,with the CMFD product having the highest accuracy and the CMADS product the worst.The accuracy characteristics of the three satellite precipitation products in space are generally characterised by a "low north-west and high south-east" and "low upstream and high downstream" basin.The overall performance of the CMFD and MSWEP products is better than the CMADS products in terms of accuracy in different months of the year,and it is also found that the satellite precipitation products perform better in the summer and autumn months when there is more precipitation.(2)The inversion results of the satellite precipitation products for extreme precipitation events show that the Rx1 day,Rx5day and SDII indices of the three satellite precipitation products generally differ significantly from the rainfall stations on both time and space scales,significantly underestimating the precipitation intensity indices of the basins,with the CMADS product showing the most serious underestimation.The CMFD product is better at characterising the CDD index than the CMADS and MSWEP products,and the CMADS product is not suitable for characterising the CDD index.The three satellite precipitation products performed weaker on the CWD index than the CDD index,all slightly overestimating the CWD index for the basin,with the CMFD product performing relatively poorly.All three satellite precipitation products significantly underestimated the number of days of heavy rainfall(R25mm)and the amount of heavy precipitation(R95p)in the basin,with the CMFD product performing relatively well.(3)The results of the SWAT hydrological model runoff simulation show that the SWAT model has good applicability in the Yanhe River basin,the CMADS and CMFD precipitation products perform better in the month-by-month and day-by-day runoff simulations of the SWAT model and have higher application value,the Nash coefficient of the MSWEP precipitation product in the validation period does not meet the model evaluation criteria,and the runoff simulation in the Yanhe River basin The applicability is poor.In addition,the runoff simulation results of the four precipitation data at the monthly scale are significantly better than those at the daily scale.