The Evolution of Remotely Sensed Precipitation Products for Hydrological Applications with a Focus on the Tropical Rainfall Measurement Mission (TRMM)

This study examines the evolution of how remotely sensed precipitation products have impacted hydrologic modeling from six basins across the continental United States. Precipitation products include both ground-based (Multisensor Precipitation Estimator - MPE) and space-based products. Two space-based products are from the Tropical Rainfall Measurement Mission (TRMM) and include the real-time TRMM Multi-Satellite Precipitation Analysis (TMPA-RT) and TRMM 3B42 Research product. Precipitation products are compared between early (2004-2008) and late (2008-2010) periods. Additionally, version 6 and the new version 7 of these TRMM products are examined. Watersheds examined were moderately large (1000 to 1,000 square kilometers) and included the San Pedro (Arizona), Cimarron (Oklahoma); Alapaha (Georgia), mid-Nueces (Texas), San Casimiro (Texas), and the mid-Rio Grande basins, which is a bi-national basin that spans the Texas-Mexico border. Precipitation products are used to drive streamflow simulations using the Soil Water Assessment Tool (SWAT). The main results of this study concludes that MPE is a mature remote sensing product that generally supports superior hydrologic simulations based on standard performance metrics such as mass balance error, Nash-Sutcliffe efficiency coefficient, and coefficient of persistence. TRMM products support acceptable simulations and have improved in performance between early and late periods for TMPA-RT (both versions) and version 6 of TRMM 3B42 Research in five out of the six basins examined. This improvement is related to modification of TRMM in January 2009 with the addition of more satellite data and a climatologic bias correction, which greatly improves the real-time TMPA-RT product. Conversely, version 7 of the TRMM 3B42 Research has a positive bias compared to version 6, which is translated into poorer hydrological simulations of streamflow. Future research is urgently needed to determine if the issues observed in this study are indicative of a broader problem associated with the most recent version of TRMM.