Mapping Boreal Forest Vitality Using Drone-Based Multispectral Time-Series and Object-Based Classification

Abstract. Monitoring tree vitality in boreal forests is increasingly critical under escalating climate stress and pest disturbances. This study presents a novel UAV-based framework integrating multispectral imaging, object-based classification, and in-situ physiological sensing for individual tree health assessment. Implemented at the Svartberget research park in northern Sweden, the approach leverages 15 UAV-acquired time-series image stacks (2024–2025) combined with sap flow, dendrometer, and stem water content and potential measurements across spruce, pine, and birch. Statistical analysis of reflectance profiles revealed strong spectral separation, particularly in the red edge and near-infrared bands, enabling early discrimination of vitality classes. Preliminary findings demonstrate the framework’s capacity to detect physiological stress signals and inter-species differences, offering a pathway toward operational forest health monitoring. The method is designed for upscaling to the 2,300 ha Attsjö super test site, where multi-sensor data fusion will support landscape-level implementation. This work highlights the value of UAV–sensor integration for precision forestry and climate-resilient ecosystem management.