DESIGN OF AN INVERSION-BASED PRECIPITATION PROFILE RETRIEVAL ALGORITHM USING AN EXPLICIT CLOUD MODEL FOR INITIAL GUESS MICROPHYSICS

This paper describes the design and validation of the FSU precipitation profile retrieval algorithm for applications with SSM/I passive microwave measurements. The algorithm employs the principles of multifrequency inversion based on forward radiative transfer modeling. A Sobolev 2-stream solution to the radiative transfer equation (RTE) is used as the forward RTE model and is described herein. The method is shown to be very accurate, retaining the same degree of computational efficiency inherent to simpler 2-stream flux models. Tests of the model against more detailed multistream, adding-doubling models demonstrate that the Sobolev solution produces radiance accuracies of approximately 1%. An advantage of the Sobolev approach is that the intensity field can be expanded in a mathematically consistent fashion, an essential feature in applications with the off-nadir SSM/I microwave measurements. A 4-dimensional non-hydrostatic cloud model provides the microphysical underpinnings of the algorithm, and is used to generate the initial guess profiles for the inversion procedure. The various stages of the algorithm are described, as well as two different methods of computational implementation for storm-scale and global-scale applications. The paper also summarizes a number of different rainrate validation analyses that have been carried out at the two scales, as well as examining the capabilities of the algorithm in diagnosing the vertical latent heating structure. The validation results represent a mixture of quantitative comparisons to radar and raingage datasets, and more qualitative comparisons to the numerical modeling results of other investigators. Because of known uncertainties in the validation data in terms of their accuracy and representativeness, and the underlying problems with time-space matching of the comparisons, it is not yet possible to place absolute confidence limits on the retrievals. However, taken as a whole, the rainrate validation analyses and the estimated latent heating profiles present solid evidence that the profile approach is returning credible rainfall estimates whose uncertainties are commensurate with those of current validation data.