Development and intercomparison of AMSU rain rate algorithms

Eleven months of collocated AMSU and NEXRAD WSR-88D ground benchmark data were collected and analyzed in order to evaluate existing and developing rain rate estimation algorithms. These algorithms are all based upon the scattering signature of the higher-frequency AMSU-B channels, and one of the algorithms is currently used operationally by the NESDIS Microwave Surface and Precipitation Products System team. This algorithm uses a cloud model-derived regression approach to determine rain rate based on computed ice water path; the ice water path is derived from the 89 and 150 GHz scattering signature according to the guidance of radiative transfer modeling studies.

Two alternative algorithms we have developed involve the compilation of monthly mean brightness temperature fields computed for suspected precipitation-free AMSU pixels. One algorithm relies upon the strong sensitivity of 150 GHz radiation to hydrometeor scattering, and significant reductions of brightness temperature from the monthly mean background are interpreted in terms of a rain rate. A related algorithm relies upon the combined response of the 89 and 150 GHz channels.

The Heidke skill score is applied to calibrate the algorithms using the NEXRAD data, and it was found that all three algorithms are reasonably adept at delineating the areal extent of precipitation, and that the NESDIS and 150 GHz algorithm provide rain rates that tend to be proportional to the NEXRAD rain rate product. Preliminary results suggest that the combined 89 and 150 GHz algorithm is not as robust at rain rate estimation over the full range of precipitation intensities.