Over-ocean Evaluation of a GPM Radar-Radiometer Combined Algorithm Using TRMM-based Synthetic Data

In this study, the performance of a combined radar-radiometer retrieval algorithm for the Global Precipitation Measurement (GPM) Mission is investigated using synthetic data derived from Tropical Rainfall Measuring Mission (TRMM) observations. The combined retrieval algorithm is based on the generation of an ensemble of dual-frequency radar-only precipitation profile estimates and the calculation of associated microwave brightness temperatures. A statistical analysis of the agreement between calculated and observed brightness temperatures identifies the radar-only estimates that are most consistent with the observed brightness temperatures. Variables that are necessary for the calculation of brightness temperatures but which are not essential in the derivation of the radar-only retrievals are stochastically prescribed and refined based on the agreement between simulated and observed radiometer brightness temperatures. To evaluate this algorithm over oceans, TRMM observations are used. That is, vertical profiles particles size distribution (PSD) intercepts are assumed (one profile per Precipitation Radar (PR) footprint) for a large number of TRMM orbits. A single frequency radar-retrieval algorithm is applied to TRMM PR observations to derive vertical PSD profiles consistent with both the assumed PSD intercepts and the TRMM PR observations. TRMM Microwave Imager (TMI) observations are used to derive surface conditions (i.e. sea surface temperatures and sea surface winds) and parameterized information regarding the vertical distribution of water vapor and clouds in rain-free fields of view (FOVs). The surface conditions in rain FOVs are determined by interpolation of rain-free FOVs. The vertical profiles of water vapor, water and ice clouds are statistically prescribed in rain FOVs. The TMI derived surface conditions, the PR derived PSDs and the statistically prescribed water vapor and cloud profiles are used to synthesize Ka-band radar observations and associated (PR resolution) brightness temperatures at the GMI frequencies. The Ku-band PR observations, the associated Ka-band simulated radar observations, and the simulated GMI observations convolved at the GMI resolution constitute the synthetic observations used to test the GPM combined retrieval algorithm. The precipitation estimates derived from the synthetic observations are compared to reference precipitation assumed in the observation synthesis. Results indicate notably superior performance of the combined algorithm relative to that of a dual-frequency radar-only algorithm.