Advancing operational global aerosol forecasting with machine learning

Ke Gui; Xutao Zhang; Huizheng Che; Lei Li; Yu Zheng; Linchang An; Yucong Miao; Hujia Zhao; Oleg Dubovik; Brent Holben; Jun Wang; Pawan Gupta; Elena S Lind; Carlos Toledano; Hong Wang; Zhili Wang; Yaqiang Wang; Xiaomeng Huang; Kan Dai; Xiangao Xia; Xiaofeng Xu; Xiaoye Zhang

doi:10.1038/s41586-026-10234-y

Back

Advancing operational global aerosol forecasting with machine learning

Journal article

Open access

Peer reviewed

Advancing operational global aerosol forecasting with machine learning

Ke Gui, Xutao Zhang, Huizheng Che, Lei Li, Yu Zheng, Linchang An, Yucong Miao, Hujia Zhao, Oleg Dubovik, Brent Holben, …

Nature (London)

03/04/2026

DOI: 10.1038/s41586-026-10234-y

PMID: 41781617

Files and links (1)

url

https://doi.org/10.1038/s41586-026-10234-yView

Published (Version of record) Open Access

Abstract

Aerosol forecasting is important for air-quality management, health risk assessment and climate change mitigation . However, it is more complex than weather forecasting, owing to the interactions between aerosol physicochemical processes and atmospheric dynamics, resulting in high uncertainty and computational costs . Here we develop a machine-learning-driven Global Aerosol-Meteorology Forecasting System (AI-GAMFS), which provides reliable 5-day, 3-hourly forecasts of aerosol optical components and surface concentrations. AI-GAMFS combines a vision transformer and U-Net in a backbone network, robustly capturing the complex aerosol-meteorology interactions via global attention and spatiotemporal encoding. Trained on 42 years of aerosol reanalysis data and initialized with Global Earth Observing System Forward Processing (GEOS-FP) analyses, AI-GAMFS delivers operational 5-day forecasts in 1 minute. Evaluation with independent ground-based observations suggests improved performance compared with the Copernicus Atmosphere Monitoring Service and regional dust models in forecasting aerosol optical depth and dust components. Compared with GEOS-FP , it has a lower root-mean-square error for global aerosol optical depth, with comparable dust forecasting skill and improved surface aerosol component forecasts over the USA and China. Our results provide a step forward in leveraging machine learning to refine aerosol forecasting and may help warn against aerosol pollution events such as dust storms and wildfires.

Details

Title: Subtitle: Advancing operational global aerosol forecasting with machine learning
Creators: Ke Gui - Chinese Academy of Meteorological Sciences
Xutao Zhang - Chinese Academy of Meteorological Sciences
Huizheng Che - Chinese Academy of Meteorological Sciences
Lei Li - Chinese Academy of Meteorological Sciences
Yu Zheng - Chinese Academy of Meteorological Sciences
Linchang An - China Meteorological Administration
Yucong Miao - Chinese Academy of Meteorological Sciences
Hujia Zhao - China Meteorological Administration
Oleg Dubovik - Centre National de la Recherche Scientifique
Brent Holben - Goddard Space Flight Center
Jun Wang - University of Iowa
Pawan Gupta - Goddard Space Flight Center
Elena S Lind - Goddard Space Flight Center
Carlos Toledano - Universidad de Valladolid
Hong Wang - Chinese Academy of Meteorological Sciences
Zhili Wang - Chinese Academy of Meteorological Sciences
Yaqiang Wang - Chinese Academy of Meteorological Sciences
Xiaomeng Huang - Tsinghua University
Kan Dai - China Meteorological Administration
Xiangao Xia - Chinese Academy of Sciences
Xiaofeng Xu - China Meteorological Administration
Xiaoye Zhang - Chinese Academy of Meteorological Sciences
Resource Type: Journal article
Publication Details: Nature (London)
DOI: 10.1038/s41586-026-10234-y
PMID: 41781617
NLM abbreviation: Nature
ISSN: 1476-4687
eISSN: 1476-4687
Publisher: Springer Nature
Grant note: NASANational Natural Science Foundation of China: 42588301, 42522509, 42090033, 42175153, 42375188, 42275195 National Key Research and Development Program of China: 2023YFC3706305 Youth Innovation Team of China Meteorological Administration: CMA2024QN13 Basic Research Fund of CAMS: 2023Z021, 2025XM002 Lichtenberger Family Chair professorship in the University of Iowa
We thank NASA for providing the MERRA-2 reanalysis, GEOS-FP analysis and forecast data, and AERONET observational datasets; the European Centre for Medium-Range Weather Forecasts for supplying the CAMS aerosol forecast products; the SDS-WAS for the regional dust forecast products; and the IMPROVE and EPA-CSN networks for the surface aerosol component data. This research was supported by the National Natural Science Foundation of China (42588301, 42522509, 42090033, 42175153, 42375188 and 42275195), the National Key Research and Development Program of China (2023YFC3706305), the Youth Innovation Team of China Meteorological Administration (CMA2024QN13) and the Basic Research Fund of CAMS (2023Z021 and 2025XM002). J.W.'s participation is supported by the Lichtenberger Family Chair professorship in the University of Iowa.
Language: English
Electronic publication date: 03/04/2026
Academic Unit: Electrical and Computer Engineering; Civil and Environmental Engineering; Physics and Astronomy; Chemical and Biochemical Engineering
Record Identifier: 9985142951602771

Metrics

1 Record Views