Multi-model assessment of black carbon in Arctic snow and sea ice, an AeroCom analysis

Climate forcing from black carbon (BC) in Arctic snow has only been simulated with a few models, though many aerosol models exist. Here, we apply aerosol deposition fields from 25 models contributing to the two Phases of the Aerosol Comparisons between Observations and Models (AeroCom) project to simulate and evaluate within-snow BC concentrations and radiative forcing. We accomplish this by driving the offline land (CLM) and sea-ice (CICE) components of the NCAR Community Earth System Model (CESM) with different deposition fields and meteorological conditions from 2004-2009, during which an extensive Arctic field campaign of BC-in-snow measurements occurred. We find that models generally overestimate BC-in-snow concentrations in the Arctic compared with the observations, though there is substantial model diversity. The model bias of BC-in-snow concentration ranges from -9.7 ng/g to +24.8 ng/g for Phase I models, and from -9.3 ng/g to +36.4 ng/g for Phase II models. There is a high correlation between BC-in-snow concentrations averaged over grid cells with observations and averaged over all Arctic surface snow, revealing that the field campaign sampling sites provide a good representation of the Arctic as a whole. Factors determining model BC concentrations in Arctic snow include Arctic BC emissions, transport of extra-Arctic aerosols, deposition efficiency of aerosols within the Arctic atmosphere, and meltwater removal of particles in snow. We find that model comparison with samples collected during the peak melt season (May-July, depending on location) is sensitive to the prescribed melt water scavenging coefficients in CLM and CICE. The Arctic BC residence time in Phase II models ranges from 4 to 25 days, implying that intermodel variation of BC deposition processes is large. We find that at most 40% of the BC which deposits in the Arctic (defined here as 60N-90N) originates from the Arctic in Phase I models, and 25% for the Phase II models. The mean Arctic BC-in-snow radiative forcing is 0.15 W/m² and 0.19 W/m² for Phase I and Phase II models respectively. We correct model radiative forcing for the biases in BC concentration by dividing the Arctic into 6 regions and applying the ratio between observed BC concentration and simulated BC concentration to the BC in snow radiative forcing in each region. The observationally corrected Arctic BC-in-snow forcing is 0.13 W/m² for Phase I models and 0.11 W/m² for Phase II models.