Spatiotemporal Analysis of Snow and Vegetation Dynamics Using Copernicus Sentinel-2 Imagery in the Maritime Alps

The research examines how vegetation and snow cover patterns change over time and space within the Maritime Alps (Western Italian Alps) which serves as a climate change and ecological transition hotspot. The research utilizes Sentinel-2 Level-2A multispectral imagery which underwent processing through two cloud-based Earth Observation (EO) platforms known as Google Earth Engine (GEE) and Sentinel Hub to track environmental changes from 2018 to 2024. Vegetation phenology and snow persistence were assessed using the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Snow Index (NDSI). A dual-threshold approach was applied: fixed thresholds (NDVI = 0.30, NDSI = 0.42) enabled temporal consistency, while adaptive Otsu thresholding addressed scene-specific spectral variability and topographic effects that typically reduce classification accuracy. The validation process used four reference datasets which consisted of CORINE Land Cover 2018 and LUCAS 2022 and Carta degli Habitat and Copernicus Fractional Snow Cover (FSC) with their corresponding spatial scales and thematic precision levels. The evaluation of map accuracy included Overall Accuracy (OA) and Producer’s and User’s Accuracy (PA, UA) and Cohen’s Kappa coefficient (¿) as statistical metrics. The results indicated that Otsu thresholds achieved superior results for detecting changes when the environment shifted between snowmelt and early vegetation growth but fixed thresholds delivered better results for monitoring long-term changes. The cross-platform consistency between GEE and Sentinel Hub proved to be very high because both NDVI and NDSI products showed mismatch values under 15% which validated the reliability of the applied workflows. The NDVI data indicates that vegetation growth began earlier in years with short snow cover duration which supports the documented pattern of alpine vegetation moving to higher elevations (Lamprecht et al. 2018; Choler et al.,2021). The complete analysis of climate-driven landscape transformation emerges from studying vegetation and snow dynamics because it demonstrates how snowmelt impacts plant ecosystems. This work presents a methodological approach which uses adaptive thresholding and multi-source validation and cloud-based processing to create a transferable and reproducible system for extended alpine monitoring. The research indicates GEE with Otsu adaptive thresholds provides the optimal combination of analytical consistency and scalability but Sentinel Hub delivers the most effective results for operational visualization and public dissemination of EO-based indicators. The dual-platform system provides support for upcoming climate monitoring and risk assessment and conservation planning activities in mountain areas with limited data availability.