Diachronic analysis of vegetation dynamics in the Stagnone di Marsala Lagoon through satellite images

The comprehensive analysis of underwater ecosystems is paramount for preserving marine biodiversity, particularly in regions subject to significant anthropogenic stress. This study focused on the multi-temporal analysis of satellite imagery to accurately map the submerged vegetation within the Stagnone di Marsala lagoon (Sicily, Mediterranean basin). This coastal lagoon is distinguished by unique submerged ecological habitats, including quasi-emergent Posidonia oceanica structures forming atolls and cordons, which are exposed to persistent anthropogenic and natural pressures. Satellite images utilized in the study were procured from commercial platforms such as WorldView, QuickBird, and Pleiades and cover a 21-year period (2003–2024); the images span visible and near infrared spectral bands and have a spatial resolution of 2 meters. Before processing, the images were calibrated to the Bottom of Atmosphere (BOA) reflectance and corrected for water column contributions. The diachronic analysis of vegetation dynamics was based on the Light Gradient Boosting Machine classification methodology. Validation efforts involved in-situ data collection from 79 surveyed points, yielding an overall accuracy of 0.84, which was adjusted to 0.80 when accounting for causal concordance probabilities. Results indicate a substantial decline in the extent of atolls and cordons, with satellite observations revealing a reduction of approximately 75% over the study period. The outcomes underscore the critical role of remote sensing in monitoring underwater vegetation dynamics and enhancing conservation strategies.