Toward a learnable Artificial Intelligence Model for Aerosol Chemistry and Interactions (AIMACI) based on the Multi-Head Self-Attention algorithm

Xia, Zihan; Zhao, Chun; Yang, Zining; Du, Qiuyan; Feng, Jiawang; Jin, Chen; Shi, Jun; An, Hong

doi:https://doi.org/10.5194/acp-25-6197-2025

Articles | Volume 25, issue 12

https://doi.org/10.5194/acp-25-6197-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/acp-25-6197-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 25, issue 12

Research article

|

25 Jun 2025

Research article |

| 25 Jun 2025

Toward a learnable Artificial Intelligence Model for Aerosol Chemistry and Interactions (AIMACI) based on the Multi-Head Self-Attention algorithm

Zihan Xia, Chun Zhao, Zining Yang, Qiuyan Du, Jiawang Feng, Chen Jin, Jun Shi, and Hong An

Supplement

https://doi.org/10.5194/acp-25-6197-2025-supplement

Data sets

Efficacy of China’s clean air actions to tackle PM2.5 pollution between 2013 and 2020 (http://meicmodel.org.cn/?page_id=560) G. Glen et al. 1038/s41561-024-01540-z

NCEP FNL operational model global tropospheric analyses, continuing from July 1999 NCEP https://doi.org/10.5065/D6M043C6

Model code and software

Source code for Hybrid WRFCHEM with Physics and AI Schemes (Aerosol Chemistry and Interactions) Zihan Xia et al. https://doi.org/10.5281/zenodo.15702248

Download

Article (10038 KB)
Full-text XML

Short summary

Traditional numerical schemes of aerosol chemistry and interactions (ACI) in atmospheric models are computationally costly and are often simplified or omitted, introducing uncertainties. We use an AI scheme to achieve fast, accurate, and stable end-to-end simulation for full ACI within an atmospheric model, replacing numerical schemes. This innovation is expected to enhance the accuracy and efficiency of ACI simulations in climate models that would otherwise neglect or simplify ACI processes.