Nonlinear statistical precipitation retrievals using simulated passive microwave imagery

A nonlinear statistical retrieval algorithm used to estimate precipitation parameters from simulated passive microwave imagery is described. The microwave imagery is based on simulated 3-dimensional microphysical cloud data. The full resolution (1.5 km) upwelling brightness temperature maps are subsequently degraded to the resolution expected from a low-Earth orbiting satellite. The nonlinear statistical retrieval algorithm is composed of four stages: (1) a Karhunen-Loeve transform, (2) a projection operator, (3) a nonlinear inverse mapping, and (4) a linear minimum mean square error (LMMSE) estimator. Retrievals of rain rate and integrated ice content at both full and degraded resolutions were performed. The results show that rain rate retrieval accuracy suffered more than integrated ice content retrieval accuracy as the resolution was degraded. This is expected because the spectral signature of ice is found primarily in the higher frequency (hence, higher resolution) channels, while the spectral signature of rain is found in lower frequency channels