Projected Aerosol Changes Driven by Emissions and Climate Change Using a Machine Learning Method

Projection of future aerosols and understanding the driver of the aerosol changes are of great importance in improving the atmospheric environment and climate change mitigation. The latest Coupled Model Intercomparison Project Phase 6 (CMIP6) provides various climate projections but limited aerosol output. In this study, future near-surface aerosol concentrations from 2015 to 2100 are predicted based on a machine learning method. The machine learning model is trained with global atmospheric chemistry model results and projects aerosols with CMIP6 multi-model simulations, creatively estimating future aerosols with all important species considered. PM_2.5 (particulate matter less than 2.5 μm in diameter) concentrations in 2095 (2091-2100 mean) are projected to decrease by 40% in East Asia, 20-35% in South Asia, and 15-25% in Europe and North America, compared to those in 2020 (2015-2024 mean), under low-emission scenarios (SSP1-2.6 and SSP2-4.5), which are mainly due to the presumed emission reductions. Driven by the climate change alone, PM_2.5 concentrations would increase by 10-25% in northern China and western U.S. and decrease by 0-25% in southern China, South Asia, and Europe under the high forcing scenario (SSP5-8.5). A warmer climate exerts a stronger modulation on global aerosols. Climate-driven global future aerosol changes are found to be comparable to those contributed by changes in anthropogenic emissions over many regions of the world in high forcing scenarios, highlighting the importance of climate change in regulating future air quality.