Detection of green attack and bark beetle susceptibility in Norway Spruce: Utilizing PlanetScope Multispectral Imagery for Tri-Stage spectral separability analysis

The detection of susceptible and attacked trees is a key factor in the management of bark beetle infestations. The challenge of early detection of infestations, due to invisible changes in the canopy color, consequently hinders the control of outbreaks in a timely manner. While many studies have examined the spectral characteristics during the green-attack stage, with no discernible needle discoloration, few have focused on the temporal-scale spectral characteristics from pre-infestation to the infestation stages, accurately detected in the field. We investigated the spectral differences among three classes of forest areas: healthy, susceptible to bark beetle attacks, and green-attacked trees using individual wavebands and spectral vegetation indices (SVIs). We assumed that the spectral characteristics of these forests vary significantly due to different temporal-scale susceptibility to the bark beetle attacks. We used 16 imageries, between 2 April and 5 September 2020, acquired from PlanetScope satellite. The spatial position of attacked trees during the growing season was recorded in the field using the ArcGIS Collector App. Ancillary datasets of forest management units were used to randomly select non-attacked trees from areas with properties similar to the attacked areas. The statistical analyses were performed on the final cleaned datasets to assess the separability among the tree classes. Our findings indicate that the Green and Red wavelengths are promising bands to distinguish healthy, susceptible, and green-attacked trees. We also found the existing spectral differences between healthy and susceptible trees before bark beetle attacks, suggesting the presence of abiotic stress to facilitate the infestation. Among different SVIs, the Enhanced Vegetation Index (EVI) and Visible Atmospherically Resistant Index (VARI) were found to be capable of specifically detecting susceptible trees to the infestation. However, it is important to note that the scope of our study is limited to a one-year period and does not focus on individual trees but rather on groups of trees, indicating that future research could benefit from a multi-year analysis at the tree level to enhance the understanding and management of bark beetle dynamics.