Near-Real-Time Detection of Insect Outbreaks in Urban Trees Using a PlanetScope Time Series

Abstract

A critical challenge for urban forests is the arrival of Toumeyella parvicornis (or pine tortoise scale) in Italy, as this species damages stone pine (Pinus pinea L.), an emblematic Mediterranean species. The aim of this study is to evaluate the effectiveness of remote-sensing data for monitoring pest invasions in the urban area of Rome, using PlanetScope images with a 1-day revisit time and 3 m spatial resolution, making them ideal for detecting outbreaks in complex urban areas. First, we constructed a reference dataset, georeferencing 238 healthy trees in Tenuta San Rossore (Tuscany) and more than 2000 damaged trees in Rome’s green areas. In any case, this dataset of healthy trees—obtained from forest areas—was expected to exhibit higher photosynthetic activity compared to urban green areas. Second, more than 30,000 PlanetScope images were analyzed to test the effectiveness of the Renormalized Difference Vegetation Index in detecting this specific forest disturbance. Finally, different thresholds were examined, allowing for the identification of an optimal threshold to discriminate healthy trees from damaged trees. The index results showed a marked drop during the summer in the infested areas, compared to the healthy areas. The identified threshold provided 99% accuracy in detecting infested trees. The approach applied in this study demonstrated that PlanetScope imagery proved effective in detecting T. parvicornis, leading to promising results.