부산대학교 도서관

The radiative forcing of anthropogenic aerosols associated with aerosol‐cloud interactions (RFaci) remains the largest source of uncertainty in climate prediction. The calculation of particle number concentration (PNC), one of the critical parameters affecting RFaci, is generally simplified in climate models. Here we employ outputs from long‐term (30‐year) simulations of a global size‐resolved (sectional) aerosol microphysics model and a machine‐learning tool to develop a Random Forest Regression Model (RFRM) for PNC. We have implemented the PNC RFRM in GISS‐ModelE2.1 with a mass‐based One‐Moment Aerosol module, which is one of CMIP6 models. Compared to the default setting, the GISS‐ModelE2.1 simulation based on RFRM reduces the changes of cloud droplet number concentration associated with anthropogenic emissions, and decreases the RFaci from −1.46 to −1.11 W·m−2. This work highlights a promising approach based on machine learning to reduce uncertainties of climate models in predicting PNC and RFaci without compromising their computing efficiency. Plain Language Summary: The largest uncertainty in assessing climate change is due to the interaction of aerosols with clouds and its effect on the Earth's energy budget. To reduce this uncertainty, it is important to accurately quantify aerosols. This is possible by accounting for the physical processes and interactions happening at the scale of the aerosol sizes. However, such an approach would be computationally demanding on climate models, making them impractical to study historical changes. To address this dilemma, we use a machine learning/artificial intelligence (ML/AI) technique that learns aerosol microphysics. When coupled to a climate model, it speedily quantifies aerosols in strong agreement with atmospheric measurements. Compared to the climate model without this implicit physical treatment, the historical changes in cloud droplet numbers and the cooling effect of aerosols are now estimated lower and less uncertain. This highlights the potential of ML/AI in reducing climate model uncertainties without hindering their computational efficiency. Key Points: A physics‐guided machine learning (ML) model for particle number concentration (PNC) couples with a global climate model with low computational overheadML PNC are in better agreement with measurements and reduce cloud droplet number concentration changes caused by anthropogenic emissionsRadiative forcing due to aerosol‐cloud interactions indicates weaker cooling (−1.11 vs. −1.46 W·m−2) and is closer to IPCC median value [ABSTRACT FROM AUTHOR]