Artificial Intelligence Applications in Aerosol Research

Information Aerosol science is entering a data-intensive era. Observations from satellites, ground-based networks, mobile platforms, low-cost sensors, laboratory systems, and chemical transport models are expanding rapidly, but these datasets are difficult to interpret with conventional approaches alone because aerosol processes are nonlinear, multiscale, and strongly coupled with emissions, meteorology, chemistry, exposure, and climate feedback. Artificial intelligence, including machine learning, deep learning, neural networks, Bayesian learning, explainable AI, and hybrid physics-AI models, offers new ways to retrieve, fuse, interpret, and predict aerosol information. These approaches can support aerosol optical depth retrieval, PM2.5 and PM10 forecasting, source apportionment, secondary aerosol analysis, uncertainty quantification, sensor calibration, exposure mapping, health-risk assessment, and air-quality decision support. This Topic Issue aims to provide a focused forum for state-of-the-art AI-driven aerosol research. We especially encourage contributions that combine rigorous aerosol science with transparent, validated, and interpretable AI workflows. Submissions should move beyond black-box predictions where possible by clarifying data quality, model generalizability, uncertainty, physical consistency, and practical relevance for cleaner air, climate understanding, and public-health protection. Original research articles, comprehensive reviews, methodological papers, benchmark datasets, and application-oriented case studies are welcome. Contributions connected to IAC 2026 themes and to the broader international aerosol community are particularly encouraged. Suggested Topics Topic Indicative Scope AI-enhanced aerosol monitoring and retrieval Machine learning for satellite AOD retrieval, lidar and ground-based inversion, low-cost sensor calibration, and multi-sensor data fusion. Air quality forecasting and early warning Deep learning and hybrid physics-AI prediction of PM2.5, PM10, ultrafine particles, haze episodes, dust, wildfire smoke, and regional pollution transport. Source apportionment and emission characterization AI-assisted receptor modeling, chemical fingerprint recognition, real-time source tracking, emission inventory improvement, and inverse modeling. Atmospheric chemistry and aerosol formation Data-driven discovery of secondary aerosol pathways, new particle formation, multiphase chemistry, aging, and reaction kinetics. Aerosol-cloud-climate interactions AI parameterization and emulation for aerosol optical properties, cloud condensation nuclei, ice-nucleating particles, radiative forcing, and climate feedback. Exposure, health, and risk assessment Exposure mapping, epidemiological modeling, oxidative potential prediction, risk stratification, and integration with health and mobility data. Data assimilation, uncertainty, and explainability AI-enhanced chemical transport modeling, Bayesian and ensemble approaches, uncertainty quantification, model interpretability, and physically constrained AI. Intelligent air quality management AI-optimized control strategies, smart sensor networks, digital twins, decision-support systems, and policy evaluation. Novel AI architectures, benchmarks, and open datasets Aerosol-specific machine learning frameworks, foundation models, graph neural networks, transfer learning, benchmark datasets, and reproducible workflows. Keywords: artificial intelligence; aerosol science; machine learning; air quality forecasting; aerosol retrieval; source apportionment; aerosol–cloud–climate interactions; exposure and health Submission deadline: 1 May 2027 Academic Editors Junji Cao , Institute of Atmospheric Physics, Chinese Academy of Sciences Frank Lee , Hong Kong University of Science & Technology Zhenxing Shen , Xi’an Jiaotong University Dongxu Yang , Institute of Atmospheric Physics, Chinese Academy of Sciences Meng Wang , Hong Kong Polytechnic University