Assessment of the Effects of Drop Size Distribution Variations Retrieved from UHF Radar on Passive Microwave Remote Sensing of Precipitation

Abstract

Vertical profiles of drop size distribution (DSD) parameters are produced from data collected with the National Oceanographic and Atmospheric Administration 915- and 50-MHz Doppler radars at Darwin, Australia, for the 1993?94 monsoon season. An existing algorithm is used to retrieve gamma size distribution parameters from the VHF and UHF Doppler radar spectra. The clear-air mean velocities and spectral widths obtained from the VHF radar are used to fit DSDs accurately to UHF spectra. Uncertainties in retrieved precipitation parameters are estimated from errors in both VHF and UHF spectra. The statistics of the retrieved profiles of DSD parameters are summarized and compared with surface disdrometer data from a site near Darwin. Retrieved vertical profiles of gamma DSDs are input to a microwave radiative transfer model to determine realistic variations in upwelling 10- and 19-GHz brightness temperatures due to uncertainties in drop size distribution. These brightness temperature variations are then used to estimate the error in simple emission-based passive microwave remote sensing algorithms for tropical rainfall due to the Marshall?Palmer assumption. For a viewing angle of 53.1° and for vertical polarization, the two-sigma scatter in brightness temperature is estimated to be ±7.0 K at 10 GHz and ±6.8 K at 19 GHz. The rms difference in brightness temperatures from the Marshall?Palmer brightness temperature, rain rate curve is estimated to be 6.7 K at 10 GHz and 4.9 K at 19 GHz. It is concluded that the scatter in the modeled brightness temperatures is primarily due to variations in the retrieved DSDs, though some scatter can be attributed to vertical inhomogeneity within the DSD profiles. The rms difference in the brightness temperature?rain rate relationship from Marshall?Palmer is consistent with a systematic shift in the retrieved DSDs toward smaller raindrops for a given rain rate than is predicted with the Marshall?Palmer DSD.