Six archetypal profiles selected from APR2 datasets have been selected to test the radar algorithm under a variety of conditions. For each profile an ensemble of solutions with different algorithm assumptions and surface wind speeds (in order to vary the surface emissivity) is generated, and brightness temperatures at the GMI frequencies are computed for each retrieved profile. These ensembles are used to quantify the information content added by the GMI brightness temperatures in the following way: For each retrieved profile i in the ensemble of n solutions, the subset of m retrieved profiles with similar brightness temperatures (within 3K at all frequencies) is selected. The variance of profile parameters (DSD, cloud water, water vapor, surface wind) is calculated for each subset, and the mean variance is compared to the variance in the ensemble.
The results suggest that the microwave radiances strongly constrain the rain DSD, especially in the light rain cases where there can be dual solutions for a given set of reflectivities at Ka and Ku band. The low-frequency brightness temperatures are sensitive to wind speed even in the deep convective profiles with rain rates up to 30 mm/hr, suggesting that this could be a retrieved parameter in the GPM combined product over ocean in rain, and emphasizing the need to properly characterize multi-channel emissivities and their covariances over land to avoid introducing a source of bias. Cloud water and water vapor are poorly constrained by the radiances, yet have a strong impact on the retrieved vertical DSD profile. Obtaining good a priori estimates of these parameters should therefore be a focus of pre-launch field campaigns.