Arctic Aerosol Model Validation and Evidence for Deposition Ice Nucleation over Siberia

Aerosol distributions have a potentially large influence on Arctic climate-relevant cloud properties but can be difficult to observe given the pervasive cloudiness, bright surfaces, generally low aerosol amounts, long polar nights, and data paucity over remote regions. As a result, Arctic aerosol-cloud interaction studies often rely on poorly validated model-based simulations of aerosols such as dust, which is an important source of ice nucleating particles in climate models. To better understand these uncertainties, we compared Arctic 2008-2016 dust and combustion aerosol distributions from the CALIPSO satellite with MERRA-2 reanalysis products and the FLEXPART model. In general, both MERRA2 and FLEXPART correctly predicted the presence and absence of strong dust and combustion aerosol layers; false positive and negative rates were <9% over oceanic regions, and <14% over terrestrial regions. However, the models performed less well for dust near the surface, with FLEXPART overestimating strong dust events, and MERRA-2 underestimating local dust sources. There is also evidence that CALIPSO misattributes diamond dust, which occurs up to 60% of the time in certain locations during winter, to dust aerosol. Interestingly, MERRA-2 and FLEXPART predict wintertime dust and combustion aerosols to occur most frequently over the same Siberian regions where diamond dust is most common, suggesting that aerosol impacts on deposition ice nucleation may occur most often over this region as well.