Detecting Agro-Pastoral Abandonment: A Sentinel-2 Multi-Temporal Phenology Approach for Sub-Mediterranean Landscapes

The cessation of agro-pastoral management in sub-Mediterranean mountains drives secondary succession from dry grasslands to closed woodlands, threatening biodiversity and agricultural capacity. This paper presents a Google Earth Engine methodology mapping abandonment status and successional velocity at cadastral resolution using Sentinel-2 (2018–2024). A deterministic cascade exploits the seasonal contrast where managed grasslands are drought-dormant in summer while evergreen maquis remains active, combined with Seasonal Amplitude, Mann–Kendall trend analysis, and the JRC Global Human Settlement Layer, requiring no labelled training data. Applied to Italy's Monti Prenestini district, the method achieves 85.8% overall accuracy (κ = 0.82). Encroached parcels, treated here as a priority restoration class for the study area, occupy moderate, machineryaccessible slopes similar to currently managed land, while most abandoned parcels sit on steeper terrain that constrains recovery. The resulting ranked cadastral database supports parcel-level prioritisation of interventions.