Hydrological assessment of the latest multi-satellite precipitation analysis IMERG over Nanliu River Basin in Southern China

ABSTRACT: In this study, early run and final run Integrated Multi-satellite Retrivals for GPM (IMERG) products are hydrologically evaluated with gauge analysis product as reference over Nanliu River basin in Southern China since March 2014 to December 2016 through Xin'anjiang hydrologic model. The IMERG products include the early run calibrated IMERG product (IMERG_ERCal) and final run satellite-only and gauge-corrected IMERG products (IMERG_FRUncal and IMERG_FRCal, respectively). Direct comparison with gauge indicates that both early run and final run IMERG products have a good performance in capturing the precipitation at spatial and temporal scales. IMERG_FRUncal and IMERG_ERCal shows compatible capabilities to detect rainfall in daily scale with highly correlative coefficient (CC) about 0.66, relative bias (RB) about -23.21%, and root-mean-square error (RMSE) about 10.68mm. IMERG_FRCal performed a litter better than IMERG_FRUncal and IMERG_ERCal with higher CC (0.7) and lower magnitude of RB (6.33%). Simulated stream flows with daily accumulated IMERG products as forcing data show large deviation from the observed stream flows with low Nash-Sutcliffe index (NSCE) less than 0.4, indicating that all of these IMERG products have limited potentials of hydrological utilization in this basin. Particularly, IMERG_FRCal shows negative NSCE (-0.05) close to zero but have lowest smallest magnitude of RB (-2.92%) when compared with gauged stream flows. IMERG_FRUncal and IMERG_ERCal exhibits a little better performance in hydrological simulation with higher NSCE (0.35 and 0.39, respectively) and larger deviations with RB about (-51.05% and -49.52%, respectively). At monthly scale, IMERG_FRCal demonstrates better performances with higher NSCE (0.49) and higher CC (0.86). On the contrary, IMERG_FRUncal and IMERG_ERCal shows dramatically decreased NSCE from 0.35 to -0.15, 0.39 to -0.23, respectively. This result manifest that the IMERG products users need to be cautious when use IMERG products for hydrological applications in this basin. This study is expected to offer insights into IMERG' potentials in hydrological utility and thus provide useful feedback to the IMERG algorithm developers and the users.

Index Terms—Hydrology, quantitative precipitation estimation, remote sensing, satellite.