Applicability evaluation of GPM satellite precipitation data in upper Huaihe River Basin

The new generation of global satellite precipitation product, GPM (Global Precipitation Measurement), offers an innovative approach to rainfall estimation.Taking the gauge′s observations as the reference, the accuracy of GPM satellite precipitation data is comprehensively evaluated from point to subbasin scale spatially and from daily to monthly scale temporally in the Wangjiaba Basin, located in the upper Huaihe River Basin.Furthermore, by integrating the distributed SWAT (Soil And Water Assessment Tool) model, the applicability of GPM for hydrological simulation is also evaluated.The results show that: ① From the aggregation of daily scale to the monthly scale, from the average of point scale to the subbasin scale, the rainfall accuracy of GPM is gradually improved.On an area-based scale, the statistical monthly GPM rainfall achieves the highest accuracy (correlation coefficient CC=0.95, root mean square error RMSE=1.10 mm, average absolute error MAE=0.79 mm).② GPM overestimates generally against the measured rainfall (relative error BIAS=17.50%~24.14%), particularly during non-flood seasons with low rainfall intensity and frequency.However, GPM demonstrates high accuracy in estimating high-intensity rainfall and rainfall during flood seasons, and the accuracy of GPM is particularly high in mountain regions and humid areas located in the south part of the Huaihe River′s mainstream due to abundant rainfall.③ When the static parameter method (i.e.the SWAT model parameters determined by the gauge observation) is adopted, the daily simulation accuracy which is driven by GPM is significantly degraded, but the monthly SWAT model can still maintain reliable accuracy.④ By adopting the dynamic parameter method (re-calibration of SWAT model parameters based on GPM rainfall), the daily and monthly simulation accuracies based on GPM rainfall can be guaranteed.This study indicates that although GPM may not accurately capture the daily rainfall processes, it exhibits high accuracy in estimating the total monthly rainfall.Therefore, GPM can basically substitute for gauge-based rainfall data in monthly flow simulation.In addition, for areas lacking rainfall data, the calibration of hydrological model parameters based on GPM historical data demonstrates good applicability for hydrological simulation.