Light absorption of aerosols remains a limiting factor for the capability of Earth System models in simulating global and regional hydroclimate through the aerosol direct and semi-direct effects. A recent analysis shows that the CMIP6 inter-model spread in precipitation inhibition due to uncertainty in aerosol absorption is around 35mm year⁻¹, which is seven times larger than the inter-model spread in historical precipitation change (±5mm year⁻¹). Carbonaceous aerosols including black carbon (BC) and brown carbon (BrC) dominate aerosol absorption in the shortwave, especially over source regions of anthropogenic and biomass burning emissions. However, their optical properties remain poorly constrained. A major challenge is the model representation of aerosol removal and absorption efficiency.

In this presentation, we will discuss the improvement of aerosol absorption in the latest development of DOE’s Earth System Model (E3SMv3), comparing to long-term observations and field measurements. A notable advancement in E3SMv3 aerosol model includes a modified volatility basis set (VBS) approach implemented for secondary organic aerosol (SOA) that contains multi-generational production and photolytic loss. We find that including SOA photolysis helps capture the decreasing trend in the mean aerosol size and OA:BC mass ratio during the aging of biomass burning aerosol, as indicated by the aircraft measurements. This inclusion reduces the overestimation of the simulated OA:BC ratio and aerosol absorption optical depth (AAOD) of biomass burning aerosols that were aged by several days. Additionally, we implemented a machine-learning (ML) model that is trained on particle-resolved simulations to distinguish BC-containing particles from BC-free particles in the accumulation mode. Compared to the default assumption of internal mixing, this change leads to significant reduction in the BC coating thickness, further reducing the overestimated OA:BC ratio and thus total absorption of aged aerosols. Together with other modifications of aerosol processes such as a shorter mixing timescale and enhanced wet removal of OA, E3SMv3 coupled with the ML mixing model shows significant improvement in simulating light absorption of aerosols.