Passive Microwave Precipitation Detection Biases: Relationship to Cloud Properties

Abstract

Accurate measurement of the Earth’s hydrologic cycle requires a more precise understanding of precipitation accumulation and intensity on a global scale. While there is a long record of passive microwave satellite measurements, passive microwave rainfall retrievals often fail to detect light precipitation or have light rain intensity biases because they cannot differentiate between emission from cloud and rain water. Previous studies have shown that AMSR-E significantly underestimates rainfall occurrence and volume compared to CloudSat. This underestimation totals just below 0.6 mm/day quasi-globally (60S-60N), but there are larger regional variations related to the dominant cloud regime. This study aims to use Moderate Resolution Imaging Spectroradiometer (MODIS) and the 94-GHz CloudSat Cloud Profiling Radar (CPR), which has a high sensitivity to light rain, with the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) observations, to help better characterize the properties of clouds that lead to passive microwave rainfall detection biases. CPR cloud and precipitation retrievals, AMSR-E Level-2B Goddard Profiling 2010 Algorithm (GPROF 2010) rainfall retrievals, and MODIS cloud properties were collocated and analyzed for 2007-2009. MODIS cloud microphysical and macrophysical properties, such as optical thickness, particle effective radius, and liquid water path were analyzed when precipitation is detected by CloudSat and missed by AMSR-E. Results are consistent with past studies and show large passive microwave precipitation detection biases compared to CloudSat in stratocumulus and shallow cumulus regimes. An examination of cases where AMSR-E failed to detect precipitation detected by CloudSat shows that warm rain detection biases occur more frequently within lower LWP, τ , and CTH bins, but biases at higher LWP, τ_ , and CTH contribute more to the total frequency of missed precipitation. Warm rain detection biases occur more frequently and biases contribute to more of the total frequency of missed precipitation for rve > 16 µ-m. Cloud property-dependent thresholds were calculated and compared against Advanced Microwave Scanning Radiometer (Earth Observing System) (AMSR-E) Goddard Profiling Algorithm (GPROF). All cloud property-dependent brightness temperature (TB) thresholds showed improvements in hit rate and volumetric hit rates. Cloud property-dependent TB thresholds were investigated to determine if thresholds can be improved by separately constraining data to environmental and cloud regimes. Descent and stratocumulus regimes, which generally consist of warm clouds, showed further improvements of warm rain detection. Results suggest that aprori knowledge of cloud property information and environmental information could significantly improve the detection of warm precipitation in GPROF retrievals.