Large inter-model spread in simulated aerosol load over Arctic sea ice

Kurzfassung

Before the development of AEROSNOW aerosol optical depth (AOD) retrievals over Arctic sea ice, comprehensive observations across the central sea-ice region were limited, and understanding of aerosol variability relied largely on models. We evaluate how sixteen CMIP6 models simulate AOD over Arctic sea ice using AEROSNOW observations, focusing on spring Arctic Haze and summer clean-air conditions. Observations show localized spring AOD maxima (0.12-0.18) near marginal ice zones adjacent to northern Canada, Alaska, and Siberia, followed by a decline to 0.05-0.07 in summer. Most models (12 of 16) underestimate the spring enhancement by 40-75%, while four overestimate it by up to 340%, reflecting differences in aerosol composition, transport, and wet scavenging. Although the multi-model mean approximates observations due to compensating biases, IPSL-CM5A2-INCA, EC-Earth3-AerChem, and MRI-ESM2-0 produce seasonal mean AOD values closer to AEROSNOW. Among them, EC-Earth3-AerChem captures the observed seasonal amplitude and monthly variability more consistently.