In the presence of high wind speeds over snow covered land or sea ice, the ATLAS blowing snow algorithm compares the measured ground and atmospheric signals. If a ground return has been detected, the atmospheric signal in the bin directly above the ground exceeds a fixed multiple of the molecular backscatter, and the observed (backscatter) feature does not exceed 500m in depth, blowing snow is detected. Meteorology from the high-resolution NASA GEOS-5 Forward Processing products – wind speed, temperature, and humidity – is also sampled along the ICESat-2 orbit, allowing for predictions of blowing snow and SSA generation to be co-located in space and time with ATLAS observations.
For an initial investigation of November 2018 through December 2019 we find ATLAS detects blowing snow for 10% of unattenuated sea ice overpasses. The detected frequency maximizes during winter, with observed blowing snow frequencies of 15-25% over the central Arctic. The observed frequency is largest (20-25%) over the Fram Strait, consistent with high wind speeds as cyclones enter the Arctic from lower latitudes. Blowing snow sublimation is predicted by the model to occur for 18% of unattenuated sea ice overpasses, an overestimate of 80% relative to the observed frequency. Despite this, the temporal evolution of the observed frequency is captured well (r = 0.81). Spatially, the largest model overestimates occur over Hudson Bay and the Beaufort Sea. The model overestimates result in a large false alarm ratio (0.65-0.78) and a probability of detection that ranges from 0.50 (winter) to < 0.10 (summer).
The large majority of false positive cases occur at high wind speeds (> 7 m/s) but do not pass the ATLAS backscatter threshold for blowing snow. We will examine the snow characteristics for these cases to determine if snow depth or age influences the detected occurrence of blowing snow. Blowing snow does not occur in the model for over two-thirds (70%) of false negative cases for wind speeds in the ~ 4-7 m/s range: below the parameterized temperature dependent threshold used in the model, but above the fixed 4 m/s windspeed threshold used in the ATLAS blowing snow detection algorithm.
The optical depth of ATLAS blowing snow features increases rapidly above wind speeds of 6 m/s, consistent with larger numbers of snow particles being lifted higher into the atmosphere. Indeed, the height of detected features also increases with wind speed. However, the scattering ratio (ratio of measured to molecular backscatter) does not show a clear dependence on wind speed.
We will estimate the blowing snow sublimation rate from ATLAS measured backscatter and contrast it with the predicted sublimation rate, examining its dependence on wind speed and snow age and depth. Finally, we will use these relationships to refine Arctic wide predictions of SSA generation from blowing snow. In cases where SSA generation occurs, we will also examine observed aerosol extinction from ATLAS and the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) sensor, contrasting it to aerosol extinctions predicted by the GEOS-Chem chemical transport model.

