Bias Correction of Long-Term Satellite Monthly Precipitation Product (TRMM 3B43) over the Conterminous United States

Abstract

Abstract The Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA) has provided a valuable precipitation dataset for hydrometeorological studies (1998–2015). However, TMPA shows some differences when compared to the ground-based estimates. In this study, a correction model is developed to improve the accuracy of the TRMM precipitation monthly product by reducing the bias compared to the ground-based estimates. The TRMM 3B43 precipitation product is compared with the Parameter-Elevation Regressions on Independent Slopes Model (PRISM) and with gridded precipitation estimates acquired from the CPC Unified Precipitation Project, two ground-based precipitation estimates, in the conterminous United States. The bias between the satellite and ground-based estimates is compared with mean surface temperature and elevation, respectively. A weak linear relationship is observed between the bias and temperature but a moderate inverse linear relationship is observed between the bias and elevation. Based on these observations, a linear model is developed for the TRMM 3B43–PRISM bias and elevation. The developed model is calibrated and validated using Monte Carlo cross validation with 25% of the available data as a calibration set and the remaining 75% of the data as a validation set. The estimated model parameters are then used in a correction formula for the TRMM 3B43 dataset for elevations above 1500 m above mean sea level. The corrected TRMM 3B43 product is verified for the high-elevation regions over the entire United States as well as in two high-elevation local regions in the western United States. The results show a significant improvement in the accuracy of the monthly satellite product in the high elevations of the United States.