Integrating Traditional and Artificial Intelligence Methods in Dust Aerosol Research: A Comprehensive Review

Abstract

Purpose of Review

Dust aerosols constitute an important component of atmospheric aerosols. Accurately characterizing their full life-cycle processes is a scientific prerequisite for conducting research and management related to climate, air quality, and dust disasters. Traditional methodologies based on multi-source observations and numerical simulations have significant limitations in data fusion, physicochemical process characterization, and simulation and prediction. In recent years, artificial intelligence (AI) technology, with its powerful feature extraction, nonlinear modelling, and multi-source data fusion capabilities, has provided new avenues for atmospheric environmental research. Therefore, this study systematically reviews the technological evolution from traditional methods to AI-based approaches, which is of great significance for breaking through the limitations of dust research and enhancing the understanding of dust aerosols.

Recent Findings

AI algorithms have demonstrated considerable application potential in atmospheric pollutant monitoring, simulation, and source apportionment. However, research specifically focusing on the application of AI algorithms in dust aerosols started relatively late, resulting in a comparatively limited number of relevant publications. This review provides an in-depth assessment of 64 publications up to October 31, 2025. The related research mainly focuses on the following three directions: intelligent monitoring and high-accuracy retrievals, dust numerical model optimization, and preliminary exploration of dust health effects. Nevertheless, significant gaps remain in the application of current AI technologies to dust aerosol research. For dust source apportionment, it still faces challenges in accurately quantifying the relative contributions from natural and anthropogenic dust emissions, as well as robust attribution of long-distance transport. The characterization of key processes in AI-based models, including the interaction between dust and air pollution, remains insufficient. Additionally, dust health effect assessment is confronted with challenges such as multiple confounding factors, unclear mechanisms, and data scarcity.

Summary

This review systematically summarizes the progress of AI algorithms in dust aerosol research, emphasizing the importance of integrating AI algorithms with traditional methods to deepen our understanding of dust aerosols and improve the performance of simulations and predictions. Future research should focus on promoting the construction of multi-source high-quality datasets, developing AI models constrained by physical and chemical mechanisms, and enhancing the interpretability and reliability of AI models in characterizing dust processes and evaluating their environmental and health effects. Such efforts will more effectively support the formulation of policies related to dust risk management, air quality governance, and climate adaptation.

Graphical Abstract