Accurate mapping of benthic habitats is essential for effective marine resource management and conservation. This study presents a novel multi-sensor fusion approach combining Sentinel-2 multispectral imagery (January 30, 2024) with ICESat-2 ATL24 satellite laser bathymetry (January 24 and March 4, 2024) to classify five benthic habitat types (Coral/Algae, Seagrass, Sand, Rock, and Rubble) in Key Largo, Florida Keys. We employed Random Forest classification on 15,600 points with 12,646 labeled training samples, achieving overall accuracy of 89.25% (κ = 0.87, F1 = 0.891). ICESat-2's GT3R beam provided superior bathymetric data quality, yielding 8,794 high-confidence points. Analysis revealed significant depth stratification (Kruskal-Wallis H = 3149.24, p < 0.001), with seagrass confined to shallow waters (3.52±1.74m, 95% <5m) and sand occurring deepest (5.54±2.34m). Feature importance revealed Blue band (19.4%) as most influential, followed by Green (17.3%), NDWI (13.2%), NDVI (12.0%), and depth (11.6%). ICESat-2 integration improved accuracy by 12-15% over spectral-only approaches, demonstrating multi-sensor fusion value for operational benthic habitat mapping. Keywords: Benthic habitat mapping, ICESat-2, Sentinel-2, Random Forest, Satellite laser bathymetry, Multi-sensor fusion, Coral reefs, Seagrass, Florida Keys
Shobha Mourya Dumpati (Sat,) studied this question.