An Algorithm for Estimating Precipitation Using Combined Radar-Radiometer Observations from GPM

A method for combining observations from a spaceborne dual-wavelength radar and multi-channel passive microwave radiometer is developed, based upon an optimal estimation framework. At the core of the method is a “generalized Hitschfeld-Bordan” procedure that can be used to efficiently generate profiles of precipitation particle size distribution (PSD) parameters consistent with Ku-band radar observations, given candidate profiles of the intercepts of the size distributions. The candidate intercept profiles are varied, and the resulting PSD profiles are used to simulate reflectivities at Ka band, as well as the path-integrated attenuations (PIA’s) at both Ku and Ka band. The candidate intercept profiles are varied iteratively to obtain a best fit of the Ka-band reflectivities and Ku/Ka PIA’s. Although the Ku-Ka band radar method produces reasonable estimates of precipitation, these estimates can be improved with the addition of information from passive microwave observations at the GPM Microwave Imager (GMI) frequencies (10 - 183 GHz). In particular, cloud water and water vapor have a significant impact on Ku-Ka band precipitation estimates, and these parameters are therefore adjusted to yield Ku-Ka band profile solutions that are in optimal agreement with observed microwave radiances. The success of this method depends upon reasonable a priori estimates of all adjustable parameters, as well as the error covariances of those parameters. In initial tests these a priori data are provided by cloud system resolving model simulations. In the development of the combined algorithm, an overarching consideration is that it should be modular, in the sense that alternative precipitation particle scattering properties, radiative transfer methods, and surface emission/reflection models can be substituted and tested. Therefore, the modular structure of the algorithm will allow for improvements contributed by GPM combined algorithm team members and their collaborators. Applications of the combined radar-radiometer algorithm to synthetic data will be presented at the conference.