AUTOMATIC CENTERLINE EXTRACTION OF COVERD ROADS BY SURROUNDING OBJECTS FROM HIGH RESOLUTION SATELLITE IMAGES

Multi-layer aerosol optical depth (AOD) is retrieved using the geostationary Spinning Enhanced Visible and Infrared Imager (SEVIRI) and machine learning, trained on Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) data. The model provides AOD at a 3 km × 3 km spatial and 15 min temporal resolution over Europe. It accurately captured multi-layer AOD dynamics during Saharan dust transport and the Mount Etna eruption, demonstrating consistent physical accuracy.