CloudSat snowfall estimates over Antarctica and the Southern Ocean: An assessment of independent retrieval methodologies and multi-year snowfall analysis

CloudSat spaceborne radar snowfall retrievals using two different methodologies – the 2C-SNOW-PROFILE (2CSNOW) CloudSat product and the combined Kulie and Bennartz (2009) technique with the Hiley et al. (2011) reflectivity (Z) to snowfall rate (S) conversion (KBH) - are compared over Antarctica and surrounding Southern Ocean environments. KBH algorithm sensitivity tests are performed to demonstrate how retrievals are affected by algorithm assumptions (e.g., vertical reflectivity continuity test, the choice of near surface bin used to make surface snowfall rate retrievals, and temperature filters). These algorithm components are found to be detrimental to snowfall detection over this region by significantly reducing the snowfall population compared to 2CSNOW, especially over ocean regions prone to ERA-interim indicated convective snow. After accounting for key algorithm differences, 2C-SNOW mean annual snowfall rates are systematically higher than KBH due to the Z-S relationship adopted. Spatial annual mean snowfall accumulation differences between the two datasets are minimized over interior Antarctica, but large differences are observed over the ocean. 2C-SNOW Z-S relationships for all snowfall events (Z=10.9 S1.3), over ocean and sea ice (Z=8.2 S1.3), escarpments and Antarctic coastal areas - below 2000ma.s.l. (Z=6.7 S1.4), and Antarctic plateau - above 2000ma.s.l. (Z=5.5 S1.6) are also derived. A multi-year 2C-SNOW mean annual snowfall analysis is also provided, and comparisons with ERAInterim snowfall datasets show similar spatial patterns, but magnitude differences over oceans are observed. A monthly 2C-SNOW and acoustic depth gauge analysis is provided to demonstrate qualitative trends in snowfall accumulation between the respective datasets.