Study on the use of passive microwave observations for snowfall detection and retrieval over the Antarctic region

Advances in the knowledge of the processes that control the Antarctic cryosphere mass balance are essential in order to be able to predict its future changes and its contribution to the sea level rise. Among others the problem of measuring precipitation in the large Antarctic continent, is particularly challenging due to the scarcity of ground-based measurements of precipitation, and of precipitation events themselves. Remote sensing observations can contribute significantly to improve precipitation detection, but satellite snowfall retrieval over iced surface regions is currently one of the most problematic open issues of the precipitation science community. The goal of this study is to understand if it is possible to detect and quantify snowfall in Antarctica using Passive Microwave (PMW) observations, and to build a database to be used for PMW retrieval algorithm training, calibration, and validation activities over Antarctica. Reflectivity profiles available from the Cloud Profiling Radar (CPR) on board CloudSat are used along with multi-channel measurements from the AMSU/MHS radiometers on board NOAA and MetOp polar orbiting satellites, which provide a full spatial coverage of the Antarctic continent. We have considered a two year period (2009-2010) of CloudSat overpasses over Antarctica and we have used the dry snowfall retrieval algorithm developed by Kulie and Bennartz (2009), taking into account different assumptions for ice crystal shape and size distribution, to retrieve the snowfall rate from the CPR reflectivity profiles to be used as reference value in our study. First, a preliminary study on the Antarctic snowfall climatology has been carried out, based on the mean precipitation rate, precipitation frequency, and cloud cover frequency. Then, we have created a database built from the spatial and temporal matching of CPR and AMSU/MHS overpasses, where the CPR snowfall estimates are coupled with PMW measurements (Brightness Temperatures, TBs). Impact of CPR data representativeness within the radiometer Instant Field Of View, and beam filling effect, in relation to the different spatial resolution and scanning geometry of the two instruments, has been analyzed. We have studied the sensitivity of the TBs to precipitation taking into account the high signal variability due to snow and ice at ground and we have applied statistical procedures to construct linear or non linear combinations of these channels in order to reduce the surface background signal. The result of this study are presented and future perspectives of using the optimal set of channel combinations as input in neural network passive microwave snowfall retrieval algorithm over Antarctica are discussed.