Sea salt aerosols (SSA) produced on sea ice surfaces by blowing snow events or the lifting of frost flower crystals have been suggested as important sources of SSA during winter over polar regions. The magnitude and relative contribution of blowing snow and frost flower SSA sources, however, remain uncertain. In this study, we use 2007–2009 aerosol extinction coefficients from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument onboard the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) satellite and the GEOS-Chem global chemical transport model to constrain sources of SSA over Arctic and Antarctic sea ice. CALIOP retrievals show elevated levels of aerosol extinction coefficients (10– 20 Mm−1 ) in the lower troposphere (0–2 km) over polar regions during cold months. The standard GEOS-Chem model underestimates the CALIOP extinction coefficients by 50 %– 70 %. Adding frost flower emissions of SSA fails to explain the CALIOP observations. With blowing snow SSA emissions, the model captures the overall spatial and seasonal variation of CALIOP aerosol extinction coefficients over the polar regions but underestimates aerosol extinction over Arctic sea ice in fall to early winter and overestimates winter-tospring extinction over Antarctic sea ice. We infer the monthly surface snow salinity on first-year sea ice required to minimize the discrepancy between CALIOP extinction coefficients and the GEOS-Chem simulation. The empirically derived snow salinity shows a decreasing trend between fall and spring. The optimized blowing snow model with inferred snow salinities generally agrees with CALIOP extinction coefficients to within 10 % over sea ice but underestimates them over the regions where frost flowers are expected to have a large influence. Frost flowers could thus contribute indirectly to SSA production by increasing the local surface snow salinity and, therefore, the SSA production from blowing snow. We carry out a case study of an Arctic blowing snow SSA feature predicted by GEOS-Chem and sampled by CALIOP. Using back trajectories, we link this feature to a blowing snow event that occurred 2 days earlier over first-year sea ice and was also detected by CALIOP.