A Machine Learning-Based Long-Term Dataset of Blowing Snow Properties over Antarctica

A long-term, consistent dataset of blowing snow (BLSN), a common phenomenon over Antarctica, is essential for ice sheet mass balance analysis. While space-borne lidar missions such as Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) and Ice, Cloud, and Land Elevation Satellite (ICESat-2) have provided valuable continental-scale BLSN observations, their temporal resolutions and spatial reaches present notable limitations. Using CALIPSOs for training, a machine learning approach has been developed to generate hourly Antarctic BLSN data on the Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) grid, extending back to the 1980s. ICESat-2 observations are used to assess the post-CALIPSO period; the results demonstrate greater bias with BLSN layer heights and better agreement with BLSN fraction and optical depth. When compared with ground-based observations, MERRA-2 exhibits a comparable and lower BLSN fraction and lower layer heights than those recorded by ceilometers. This data record represents the longest continuous BLSN data record to date. This study provides an overview of key BLSN properties, including occurrence, height, and optical depth. The results highlight strong seasonal patterns: BLSN occurrence peaks during Antarctic winter, while both height and optical depth are higher in summer with no statistically significant trend. Spatially, East Antarctica exhibits higher BLSN occurrence than West Antarctica.