Multispectral drone images for the early detection of bark beetle infestations: assessment over large forest areas in the South-Eastern Alps

Aurora Bozzini ^1* Langning Huo ² Stefano Brugnaro ³ Giuseppe Morgante ¹ Henrik Jan Persson ² Valerio Finozzi ⁴ Andrea Battisti ¹ Massimo Faccoli ¹

• ¹ Department of Agronomy, Food, Natural Resources, Animals and the Environment, School of Agricultural Sciences and Veterinary Medicine, University of Padua, Legnaro, Veneto, Italy
• ² Swedish University of Agricultural Sciences, Uppsala, Uppsala, Sweden
• ³ Geologist Flight Instructor & Remote Sensing, San Giorgio delle Pertiche - Padova, Italy
• ⁴ Regione del Veneto - U.O. Fitosanitario, Treviso, Italy

Introduction: European forests face increasing threats from climate change-induced stressors, which create favorable conditions for bark beetle outbreaks. The most critical spruce forest pest in Europe is the European Spruce Bark Beetle (Ips typographus L.). Effective forest management of this beetles’ outbreaks necessitates timely detection of recently attacked spruce trees, which is challenging given the difficulty in identifying symptoms on infested tree crowns, especially over large areas. This study assessed the detectability of infested trees over large spruce dominated areas (20-60 ha) using high-resolution drone multispectral imagery. Methods: A multispectral sensor mounted on an Unmanned Aerial Vehicle (UAV) was used to capture images of the investigated spruce stands weekly during June 2023. These were used to compute the reflectance of all single trees, derive vegetation indices, and then compare these between bark beetle infested trees and healthy ones. Results: The results showed that it was possible to separate the spectral features of recently infested trees from the healthy trees during the final developmental stage of the first beetles’ generation, despite the limitations due to difficulties in image processing over large areas. The best performing vegetation indices included NDRE (Normalized Difference Red Edge index) and GNDVI (Green Normalized Difference Vegetation Index), which allowed the earlier separation between infested and healthy trees. Discussion: The study shows that the use of UAV high-resolution imagery can present some limitations when performing early detection over larger areas. The integration of sensors focused on narrower spectral windows around the Red-Edge and Green bands and other remote sensing methods (e.g., satellite imagery) could help overcome these limitations and improve early-detection over large forest areas. The proposed early-detection approach will increase the understanding of which factors to consider when performing early detection with remote sensing techniques. In particular, it will add insights when upscaling to larger spatial scales, providing useful guidance for the management of areas suffering pest outbreaks.