Dynamic monitoring of dust transport effect on maritime visibility using multi source satellite data and advanced deep learning approach

Dust storms over the Red Sea pose serious challenges to maritime navigation, reducing visibility, increasing collision risk, and disrupting port operations. This study provides a comprehensive spatiotemporal analysis of dust transport variability from 2015 to 2023 using multi-source satellite datasets (MODIS, CALIPSO, MERRA-2) and advanced analytical methods, including GIS-based modeling and convolutional neural networks (CNNs). The integrated framework captured interannual and seasonal variability in aerosol optical depth (AOD), revealing peak dust activity in 2015, 2017, and 2022, largely driven by intensified Shamal winds (predominantly northwesterly). Seasonal analysis showed dust transport reaching its maximum in spring in the northern Red Sea, shifting toward the central and southern sectors during summer and autumn. A strong positive correlation (R² > 0.60) between wind speed and AOD confirms wind‑driven entrainment as the primary dust mobilization mechanism, while deviations point to the influence of synoptic-scale circulation and surface conditions. Mann-Kendall analysis identified a statistically significant decline in visibility over the past decade, reflecting the increasing impact of dust storms on maritime operations. The CNN model demonstrated improved prediction of dust related visibility reduction by integrating multisource inputs, underscoring its practical value for monitoring and maritime risk management.