Mapping the distribution areas of the pale tussock moth (calliteara pudibunda) using multispectral camera-integrated uav (drone): the case of karacabey

In this thesis study, the spatial distribution and mapping of damage caused by the Pale Tussock Moth (Calliteara pudibunda) in Oriental beech (Fagus orientalis) stands located within the boundaries of the Karacabey Forest Enterprise Directorate in the Marmara Region of Turkey were investigated. The timely detection and monitoring of biotic stress induced by forest pests is of critical importance for the sustainable management of forest ecosystems. In this context, high-resolution imagery was acquired using unmanned aerial vehicles (UAVs) equipped with multispectral cameras, and remote sensing-based analyses were conducted to determine the extent of the damage. A total of 17 UAV flights were carried out using a DJI Mavic 3M equipped with a multispectral camera—14 in 2023 and 3 in 2024—particularly over forest compartments within the Güngörmez Forest Sub-district, where the pest damage was observed to be most severe. The flights were conducted at an altitude of 200 meters with 80% front and 70% side overlap. As a result, a total of 18,545 images were collected, generating approximately 170 GB of data. These images were processed using Agisoft Metashape software to generate orthophotos, digital surface models, and vegetation indices such as NDVI, GNDVI, and NDRE. The derived data were then analyzed using object-based classification techniques in the eCognition Developer environment. The classification outputs were spatially assessed in ArcGIS software to produce forest health maps. The analysis revealed that 968,42 hectares of the 1430 hectare surveyed area had experienced significant foliage loss due to pest activity. Among the vegetation indices evaluated, NDVI was found to be the most effective in differentiating between healthy and affected areas and was therefore used as the primary analytical index. Classification accuracy was assessed through comparison with field reference data using confusion matrices, and the resulting kappa coefficients ranged between 0.61 and 0.89, indicating generally strong to near-perfect agreement. Furthermore, the damaged areas were overlaid with forest management plan data to identify the affected stand types and quantify their respective extents. Comparative analyses of RGB imagery from 2024 indicated noticeable vegetation recovery in many of the areas that had been classified as damaged in 2023, suggesting the initiation of a natural regeneration process in the beech forests. The findings of this study demonstrate the efficacy of UAV-based multispectral imaging for monitoring forest pest outbreaks and offer a transferable methodological framework applicable to other pest species and forest ecosystems.