Hydrological evaluation of the TRMM multi-satellite precipitation estimates over the Mishui basin

This study aims to evaluate the rainfall estimation accuracy and hydrological potential of the updated Version 7 Tropical Rainfall Measuring Mission (TRMM) multi-satellite precipitation product (TRMM 3B42V7, post-real-time estimates). First, the TRMM 3B42V7 precipitation estimates were compared and validated against the ground rain gauge observations from Jan. 2003 to Dec. 2008. Then, the TRMM 3B42V7 precipitation product was applied to drive the Gridded Xinanjiang model for streamflow simulation and the SCEM-UA algorithm was utilized to analyze and diagnose the model parameters uncertainty. At daily scale, 3B42V7 shows a better performance with lower mean error and bias (only about 6.68% underestimation). However, in terms of the absolute error sense, its data accuracy was not performed as expected (approximately 57.76% of absolute bias). At monthly scale, the TRMM 3B42V7 estimates have been improved dramatically by the month-to-month gauge adjustments during the data processing algorithm. The simulating accuracy of the daily streamflow driven by TRMM 3B42V7 is not satisfactory in the common sense for missing some flood events, but it can still capture the diurnal variation of the streamflow. By contrast, the TRMM-driven streamflow simulation performs better at monthly scale relative to the daily scale. Furthermore, it is capable to represent the seasonal and inter-annual variations of the streamflow. The daily and monthly 95% confidence intervals contain most of the measured streamflow process. This work was anticipated to provide the insights to the hydrologic user community when applying the TRMM satellite precipitation products for water resources simulation and management, especially over ungauged or data-scarce basins.