Multi-Sensor Mapping of Aquatic Vegetation Using Sentinel-2, SAR–Optical Fusion and Derived Spectral and SAR Indices in South Florida

Abstract

Accurate mapping of submerged aquatic vegetation (SAV) and emergent aquatic vegetation (EAV) is essential for hydrologic assessment and wetland management. This study developed and evaluated a multi-sensor aquatic vegetation mapping framework using Sentinel-1 SAR, Sentinel-2 optical imagery, and derived spectral and SAR indices in the Everglades stormwater treatment areas (STAs) of South Florida. Three feature configurations were compared: Sentinel-2 only (S2), Sentinel-2 combined with Sentinel-1 (S2 + S1), and Sentinel-2 and Sentinel-1 augmented with derived spectral and SAR indices (S2 + S1 + Indices). Random forest (RF) and TabNet classifiers were trained to discriminate SAV and EAV. On the internal test set, both classifiers achieved similar overall accuracies (OA = 0.83–0.85), with no clear performance advantage between RF and TabNet. However, differences emerged in spatial generalization. The S2 + S1 + Indices configuration consistently produced the most robust performance in independent validation areas, achieving OA values of 0.86 and 0.90 for TabNet. The results indicate that the combination of Sentinel-1, Sentinel-2, and derived spectral and SAR indices generally provides the most robust and transferable feature representation, with the greatest benefits observed in the unseen spatial validation areas. These findings demonstrate the potential of multi-sensor data fusion for scalable wetland vegetation monitoring in cloud-prone and hydrologically dynamic environments.