Physical retrievals of over-ocean rain rate from multichannel microwave imagery. Part II: Algorithm implementation

A new physical inversion-based algorithm for retrieving rain rate over the ocean with the Special Sensor Microwave Imager (SSM/I) is described. In a departure from other rain rate retrieval algorithms, the satellite observables inverted in the present algorithms are not the raw brightness temperatures but rather normalized polarizations for 19.35, 37.0, and 85.5 GHz, plus an 85.5 GHz scattering index which is sensitive primarily to ice particles aloft. The normalized polarizations are interpreted as footprint-averages of theoretically derived analytic functions of the column optical depth associated primarily with liquid water. The effective vertical depth of the rain layer is specified as a function of the SSM/I estimated column water vapor. The retrieval algorithm performs an iterative search for a high resolution (12.5 km) rain field which is simultaneouly consistent with the 19.35 and 37.0 GHz normalized polarizations. The first-guess rain rate field is supplied by the 85.5 GHz scattering index. At gridpoints for which the rain column optical depth exceeds the dynamic range of the attenuation-based indices, the first-guess field is left essentially unmodified; elsewhere, the required consistency with the 19 and 37 GHz indices usually results in significant modification of the scattering-based rain rate estimates. The algorithm as described here is a prototype implementation which was developed with reference only to idealized theoretical models; empirical improvements to the numerical scheme and the model coefficients will be made in the future as results from the first Precipitation [algorithm] Intercomparison Project (PIP-1) and the second phase of the GPCP (Global Precipitation Climatology Project) Algorithm Intercomparison Project (AIP/2) are analyzed, as well as data from individual validation efforts. Although the algorithm is physically based and uses all SSM/I channels, it is computationally much less demanding than cloud/radiative transfer model-based inversion algorithms published elsewhere.