Artificial intelligence in sand and dust storm research: A systematic and bibliometric review (1990–2025)

Sand and dust storms (SDS) are major atmospheric hazards with substantial impacts on environmental quality, human health, and socioeconomic systems. In recent years, artificial intelligence (AI) has become an important tool for improving SDS detection, monitoring, and forecasting. However, a comprehensive synthesis of publication trends, methodological development, and dominant data–algorithm relationships remains lacking. This study presents a systematic review and bibliometric analysis of AI applications in SDS research from 1990 to 2025. A total of 822 articles retrieved from the Web of Science Core Collection were analyzed using VOSviewer and CiteSpace, and 84 studies were selected for qualitative synthesis following PRISMA guidelines. Three developmental stages were identified based on publication trends, data sources, and AI methods: an emergence stage (1990–2005) characterized by early machine learning applications, a growth stage (2005–2015) marked by expanded use of satellite observations and diversified algorithms, and a maturity stage (2015–2025) dominated by deep learning and hybrid AI–physical modeling approaches. Publications rose sharply after 2018, reaching approximately 180 records in 2025. Reported performance reached R 2 = 0.97 for aerosol retrieval and >97.5% classification accuracy, but these study-specific results are not directly comparable across datasets, tasks, and validation methods. These advances have been enabled by multi-source data fusion integrating satellite products, ground-based observations, and atmospheric reanalysis datasets. Overall, AI applications in SDS research are evolving from exploratory analyses toward operational monitoring and early-warning systems. Future progress will depend on explainable AI, physics-informed modeling, standardized benchmarks, and improved data coverage in underrepresented regions.