Environ Monit Assess. 2026 Jun 5;198(7):687. doi: 10.1007/s10661-026-15544-2.
ABSTRACT
Aeolian dust is one of the most pervasive natural air pollutants, strongly influencing atmospheric chemistry, visibility, climate feedback, and human health. This review critically evaluates the integration of geochemical fingerprinting techniques and YOLO-based deep learning frameworks for real-time aeolian dust monitoring, with the overarching objective of advancing source attribution, health risk assessment, and evidence-based policy formulation in dust-prone regions. The Middle East, particularly the Qatar Peninsula, experiences recurrent dust storms that transport mineral particles and trace metals, contributing to degraded air quality, reduced solar energy efficiency, and elevated risks of respiratory and cardiovascular diseases. Traditional monitoring approaches including satellite observations, ground-based measurements, and geochemical analyses such as inductively coupled plasma mass spectrometry (ICP-MS), scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS), and X-ray diffraction (XRD) have improved source apportionment and compositional understanding, yet they often lack real-time detection and high-resolution spatiotemporal coverage. Emerging artificial intelligence (AI) tools, particularly the You Only Look Once (YOLO) deep learning framework, offer transformative opportunities for atmospheric dust monitoring by enabling rapid, automated plume detection across multispectral and hyperspectral datasets. This review demonstrates that ICP-MS, SEM-EDS, and XRD, when applied in combination, provide robust multi-dimensional source fingerprinting: ICP-MS isotopic ratios discriminate crystal from anthropogenic contributions, SEM-EDS resolves particle-level morphology and elemental composition, and XRD quantifies mineralogical assemblages that vary systematically with source region and transport pathway. Concurrently, successive YOLO generations (YOLOv5 through YOLOv10) have achieved real-time dust plume detection with mAP values exceeding 88% on satellite and UAV imagery. The proposed multimodal YOLO-geochemical integration framework, in which YOLO-guided plume detection triggers targeted field sampling and geochemical labels iteratively retrain detection models, enables compositionally resolved, real-time exposure mapping that directly supports inhalation risk assessment against WHO PM2.5 and PM10 thresholds and metal-specific reference concentrations. Collectively, these findings establish a scalable, evidence-based monitoring architecture that strengthens early-warning systems, informs air quality regulation, and advances global strategies to mitigate the environmental and public health burdens of dust storms.
PMID:42247025 | DOI:10.1007/s10661-026-15544-2