H41C-0813:
Real-Time Correction of TRMM Precipitation for Hydrological Monitoring over China

Thursday, 18 December 2014
Xuejun Zhang and Qiuhong Tang, IGSNRR Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing, China
Abstract:
An operational hydrological monitoring system will greatly benefit disaster mitigation and water resources management. The operational hydrological models are usually calibrated using retrospective meteorological station data but forced by real-time data from satellite estimates in the operational system. A main challenge is the inconsistency between the retrospective meteorological forcings and real-time forcing data which may introduce bias in real-time hydrological estimates. Here we attempted to correct the Tropical Rainfall Measuring Mission (TRMM) real-time precipitation by removing the systematic bias between the retrospective precipitation data and the satellite precipitation product. The bias correction was performed at each 0.25 degree grid cell to match the cumulative probability function (CDF) of satellite precipitation with an observation-based precipitation data (IGSNRR precipitation) in the period of March 2000 to December 2010. An independent validation was conducted in 2011-2013 to evaluate the performance of the adjusted satellite product over 10 large river basins in China. The results show that precipitation of the adjusted satellite data agreed well with IGSNRR precipitation while the unadjusted data overestimated precipitation at most basins. The daily precipitation distribution of the adjusted product is closer to the IGSNRR distribution compared to the unadjusted precipitation as expected. Using the adjusted precipitation data, the hydrological model could reproduce better hydrographs than the unadjusted data over all the river basins. In particular, the model forced by the adjusted precipitation can capture the high flow well while the model forced by the unadjusted precipitation significantly overestimated high flow in flooding seasons. Our experiments suggest that the real-time satellite precipitation, after bias-correction, can serve as forcing for hydrological monitoring to generate estimations consistent with the retrospective hydrological simulations.