Abstract
A precipitation retrieval algorithm based on the application of a 3D radiative transfer model to a hybrid physical-stochastic 3D cloud model is described. The cloud model uses a statistical rainfall clustering scheme to generate 3D cloud structure while ensuring that the stochastically generated quantities remain physically plausible. The radiative transfer model is applied to the cloud structures to simulate satellite remotely sensed upwelling microwave brightness temperatures TB's. Regression-derived relationships between model TB's and surface rainfall rates for Special Sensor Microwave/Imager (SSM/I) frequencies are used as the foundation of the retrieval algorithm, which is valid over oceans. A case study calibrates the retrieval algorithm to the European Centre for Medium-Range Weather Forecasts (ECMWF) numerical weather prediction model and applies the algorithm to SSM/I data obtained during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment. Comparisons between the satellite-derived precipitation amounts and radar-derived amounts, at a spatial resolution of approximately 55 km, give correlations of about 0.7 for instantaneous rain rates and 0.634 for monthly accumulations. Although the satellite-derived totals are reasonably well correlated with the radar totals, they also appear to contain a relatively large positive bias, which may in part be due to the ECMWF tuning. However, optical rain gauge measurements are lager than both the satellite- and radar-derived amounts, casting uncertainty into the level of bias of the satellite algorithm. Finally, an important aspect of 3D radiative transfer in precipitating systems is illustrated by demonstrating that satellite viewing angle effects realized in the simulation framework also appear to be present in empirical relations between SSM/I TB's and radar-derived surface rainfall rates.
