The characterization of vegetation cover dynamics plays an important role for both land degradation and climate change studies. In such a context AVHRR/NDVI time series provide a powerful tool for improving knowledge on processes involving vegetation and consequently for supporting restoration policies.

For studying vegetation patterns in the framework of “Methodology Integration of EO Techniques as Operative Tool for Land Degradation Management and Planning in Mediterranean Areas” (Mildmap project – Interreg IIIB Archimed), we analyzed NDVI trends from GIMMS (Global Inventory Monitoring and Modeling System) dataset (1982–2003) available at 8 km resolution over the involved regions (South Italy and Greece).

For such areas, we investigated long-term stationary properties of vegetation dynamics of different covers by estimating the persistence probability according to a mechanical-statistical approach. The followed procedure, considers the times during which vegetation deviates from a given reference status before returning back to it. Our reference map kept the signs of NDVI trends related to the 10-year period (1982–1992); then, the signs of trends concerning longer periods up to 2003 were considered. From the persistence probability, we derived the mean life time of positive and negative trends for the selected covers, by fitting the number of surviving trends per year with the exponential decay law.

Our results show a general healthy level of vegetation and some differences between the two areas mainly for the natural covers (Forests and Maquis). They confirm the valuable use of NDVI time series for capturing recovery and/or endurance signs of vegetation cover against natural or human-induced stresses.