Detecting vegetation anomalies in mediterranean ecosystems in southern Italy using sentinel-2 NDVI time series and principal component analysis

Vegetation is central to the functioning of ecosystems, supporting biodiversity and human livelihoods. However, it is highly sensitive to climate variability, land-use changes, and natural disturbances. Vegetation anomalies, or areas where plant growth significantly deviates from expected patterns, can indicate ecological stress, altered productivity and potential land degradation. Therefore, monitoring these anomalies is critical for understanding ecosystem resilience and informing sustainable management. This study examines vegetation anomalies in two adjacent hydrographic basins in southern Italy (Basilicata region) over a seven-year period (2017-2023), using Sentinel-2 imagery and the Normalised Difference Vegetation Index (NDVI). Selective Principal Component Analysis (PCA) was applied to enhance the detection of localised temporal changes, reduce data dimensionality and extract key ecological signals, including seasonal dynamics, interannual variability and abrupt disturbances. The results reveal the spatial and temporal patterns of vegetation anomalies across different ecosystems, emphasising the impacts of climate variability and land-use practices, such as land abandonment. This study demonstrates the effectiveness of multitemporal PCA in identifying significant deviations in vegetation dynamics. It offers a valuable framework for assessing the functionality, resilience, and management of ecosystems under changing environmental conditions.