Beyond static indicators: a metastable state approach for revealing long-term agroecological trajectories in an arid oasis ecosystem

Arid oasis agroecosystems face increasing climatic and anthropogenic pressures that drive complex ecological transitions, often inadequately captured by conventional remote sensing approaches. This study proposes a metastable-state framework to reconstruct long-term agroecological trajectories using Earth observation data. The Skoura Oasis in southern Morocco was selected as a representative arid ecosystem. Homogeneous landscape units were delineated using object-based analysis of a 0.5 m Pléiades image, while a 40-year satellite archive (1984–2024) was analysed to identify a Metastable Reference State (MRS). Five indicators, MSAVI, FAPAR, NMDI, albedo, and Sandy Feature Index (SFI), were extracted using Google Earth Engine and integrated through Principal Component Analysis. A weighted agroecological distance from the MRS was calculated to quantify ecosystem divergence and temporal dynamics. The framework achieved 88% overall accuracy, a Cohen’s Kappa of 0.84, and a weighted F-score of 88.06%. Results indicate that 54.4% of the oasis is undergoing degradation, 6.8% remains metastable, and 38.7% shows ecological improvement. By incorporating metastability into long-term monitoring, this framework provides a transferable approach for assessing ecosystem resilience, degradation, and agroecological change in arid environments.