Abstract
The 85-GHz polarization corrected temperature (PCT85) algorithm, using the V85 and H85 channels of the SSM/I sensor, is evaluated for estimation of midlatitude rainfall. The algorithm θ parameter and rain/no-rain thresholds are examined and found to be highly variable. Methods for automatic calibration, to amount for variable atmospheric and surface conditions, are presented. Derivation of θ and thresholds for each individual scene provides a marked improvement in rainfall identification accuracy over the equivalent monthly values. The algorithm is calibrated by comparison with radar data for the estimation of instantaneous rain rates. Detailed evaluation of a number of case studies suggest the relationship of PCT85 and rain rate is substantially different for frontal and mesoscale convective system rainfall. For most frontal conditions the PCT85 provides useful estimates of rain rates with sensitivity to rain intensities as low as 0.1 mm h−1. Overall, the PCT85 estimates of instantaneous rain rate at the footprint scale are to within ±75% of the radar quantity only 50% of the time. Systematic errors result from both the calibration process and from the inability of microwave scattering methods to identify warm rain processes, including orographically enhanced rainfall over land. The results show the need for improved empirical calibration of passive microwave algorithms to provide sensitivity to subsynoptic-scale surface and atmospheric conditions and rainfall processes.
