Validation of Maritime Rainfall Retrievals from the TRMM Microwave Radiometer

Abstract

This study compares the surface rainfall retrieved from the Goddard profiling (GPROF; version 5) algorithm with Kwajalein ground-based radar (KR) observations at 0.1°, 0.25°, and 1° resolutions. Comparisons of the GPROF-retrieved rainfall with KR observations for 178 overpasses show that GPROF overestimated surface rainfall with respect to the KR by 16%. Power spectral density comparisons between GPROF and KR rain maps at 0.1° resolution show that GPROF-retrieved rain maps are less spatially variable at wavelengths less than 50 km in the mean, suggesting that GPROF rainy areas are smoother and more spatially extensive than those observed by the KR. Sensitivity of rainfall retrievals to the melting layer and the impact of 85-GHz channels were tested. This study introduced Klaassen's melting-layer parameterization into the GPROF algorithm that reduced the GPROF-retrieved rainfall amount by 7.5%. Considering the poor correlation between upper-level ice amounts and surface rainfall, this study only estimated the convective area fraction from 85-GHz brightness temperature and neglected 85-GHz brightness temperature in the rainfall retrieval. This modification reduced GPROF-retrieved surface rainfall amount by 3.5% and made the GPROF-retrieved rainfall retrievals more consistent with the KR observations where the GPROF algorithm overestimated surface rainfall with respect to the KR because of strong ice scattering. For the 178 overpasses used in this study the total rainfall amount retrieved by the revised GPROF algorithm overestimated rainfall by 1% with respect to that measured by the KR.