Early Identification of Vegetation Pest Diseases Using Sentinel 2 NDVI Time Series 2016–2023: The Case of Toumeyella Parvicorvis at Castel Porziano (Italy)

The aim of this work was to assess the potential of continuous change detection and classification (CCDC) CCDC and trend analysis algorithms on Sentinel 2 NDVI time series (2016–2023) to capture and estimate subtle internal vegetation anomalies, with a focus on disease induced by pests. To explore and characterize long-term vegetation dynamics, Sentinel 2 (S2) time series were analyzed using a processing chain mainly based on three steps: 1) time series segmentation; 2) linear regression (LR) and trending; and 3) classification to extract and map vegetation internal anomalies. The pilot site was selected in a peri-urban area of Rome: Castel Porziano heavily affected by Toumeyella Parvicorvis (TP) in recent years. Results from our investigations highlighted the effectiveness of the S2 time series in sensing subtle but physically significant degradation signals, and the reliability of CCDC and LR to characterize the spatial and temporal evolution of TP even veiled by seasonality and annual cycle behavior, albeit strictly dependent on the period of occurrence of the event.