Currently, significant divergences in simulated BC burden and optical properties have been shown among global climate models (GCMs), and hence BC radiative forcing remains largely uncertain. BC aging is one of the key processes that contribute to uncertainties in BC simulations. Recently, Prof. Minghuai Wang's team improves BC aging in GCMs with machine learning (ML) methods and recent laboratory/in-situ measurements. The research papers were recently published in Journal of Geophysical Research: Atmospheres and Journal of Advances in Modeling Earth Systems.
The uncertainty of BC aging in GCMs is mainly related to BC aging threshold and mixing state assumptions. The research team first uses recent measurements over China to evaluate the modeled BC aging process in the Community Atmosphere Model version 6 (CAM6) (Fig. 1). The modeled condensation aging timescale is estimated to be about 0.8 h (17%) faster than the chamber measurement, and the modeled BC aging degree also increases faster than the SP2 observations, indicating the faster BC aging in the model. Further analysis shows that SOA condensation aging dominates (64%) BC aging across China. Increasing the aging threshold of monolayers (i.e., slowing down BC aging) increases the modeled surface BC concentration over remote Western China and BC burden, but hardly changes the systematical underestimation of surface BC concentration over China (with MEIC emission inventory). The research team further improves BC mixing state representation with machine learning in a GCM. Current GCMs usually assume aerosol to be fully internal mixtures with uniform composition within a certain size range (e.g., an aerosol bin in sectional models or an aerosol mode in modal models), resulting in high degree of internal mixing of BC with non-BC species and large relative coating thickness of BC (RBC, the mass ratio of non-BC species to BC in BC-containing particles). To improve BC mixing state representation, the research team coupled a ML model of BC mixing state index (χ) trained on particle-resolved simulations into the chemistry-climate model CAM6 with MAM4 aerosol module (MAM4-ML, Fig. 2). MAM4-ML uses parameter RBC to partition accumulation mode particles into two new modes, BC-free particles and BC-containing particles, and MAM4-ML adjusts RBC to make the χ simulated by MAM-ML match the χ predicted by the ML model. On a global scale, the degree of internal mixing of BC decreases by 19%. The RBC decreases by 52% and better agrees with observations (Fig. 3). The hygroscopicity drops by 9% for BC-containing particles, leading to a 20% reduction in the BC activation fraction. By affecting BC CCN activity, MAM-ML enhances the transport of BC from source regions to the North Pacific and polar regions, resulting in a 4% increase in global BC burden. This study highlights the application of ML model in the online GCMs to improve aerosol component prediction and key aerosol process treatments. The work about the effects of mixing state representation on BC absorption is also in preparation.
The first author of the papers is Wenxiang Shen, a PhD student under the supervision of Prof. Minghuai Wang. The corresponding author is Prof. Minghuai Wang. Other co-authors include Prof. Nicole Riemer from the University of Illinois Urbana-Champaign (UIUC), Associate Prof. Zhonghua Zheng from the University of Manchester (UoM), and other scholars from Nanjing University and Beijing Weather Modification Center. The study is jointly supported by the National Science Fund for Distinguished Young Scholars, General Program and Key Project of National Natural Science Foundation of China, the National Key Research and Development Program of China, Jiangsu Collaborative Innovation Center for Climate Change, and Frontiers Science Center for Critical Earth Material Cycling of Nanjing University.
(1) Evaluating BC aging process in CAM6 with measurements. (2) Improving BC mixing state representation by coupling machine learning model into the online CAM6. (3) The effects of BC mixing state representation on BC relative coating thickness (RBC), CCN activation, and mass concentration.
References:
Shen, W., Wang*, M., Riemer, N., Zheng, Z., Liu, Y., & Dong, X. (2024). Improving BC mixing state and CCN activity representation with machine learning in the community atmosphere model version 6 (CAM6). Journal of Advances in Modeling Earth Systems, 16, e2023MS003889. https://doi.org/10.1029/2023MS003889
Shen, W., Wang*, M., Liu, Y., Dong, X., Zhao, D., Yue, M., et al. (2023). Evaluating BC aging processes in the Community Atmosphere Model version 6 (CAM6). Journal of Geophysical Research: Atmospheres, 128, e2022JD037427. https://doi.org/10.1029/2022JD037427