Mid-latitude rainfall biases between spaceborne radars and the role of drop size distribution assumptions

Abstract:

CloudSat-CPR and GPM-DPR rainfall retrievals are compared and statistically merged over the midlatitude oceans. The CPR produces significantly higher rainfall accumulation than the DPR. We place these results in meteorological context by compositing around midlatitude cyclones using a regime dependent framework defined by column water vapor and cloud liquid water path. Differences primarily emerge from biases ahead of extratropical warm fronts. We utilize two separate statistical distribution merging techniques to blend CloudSat and DPR rainfall products. The merged mean rain rate between 40 and 60 degrees latitude varies between 1.52 and 2.30 mm/day while the DPR alone gives 1.38 mm/day. Importantly, we find that a large part of the discrepancy between the merged distribution and the DPR does not result from missed detection of light rainfall, but rather from differences between the rain rates derived from CPR and the DPR algorithms. Differences in rates from the two platforms point towards algorithmic uncertainties related to drop size distributions (DSD) and sensor resolution.

To understand the impacts of DSD assumptions on these results we present our ongoing work to run an ensemble of rainfall retrievals using the CPR operational algorithms where the DSD is modified to match in-situ measurements from the extensive OceanRAIN dataset. The in-situ DSD’s have larger rain drops that the default CPR assumptions, which causes an increase in the CPR rates. These experiments further increase the discrepancies between CPR and DPR retrievals. The unresolved differences point to the need for future precipitation observing systems to include coincident measurement from multiple frequencies spanning Ku through W-band.