Integrating green-shoulder indices from hyperspectral drone imagery and sap flow monitoring to assess water dynamics in healthy and bark beetle-infested trees

Abstract. Accurate detection of tree physiological stress is essential for understanding health decline and improving disturbance monitoring. Tree stress if often linked to hydraulic dysfunction, yet detecting the onset of this process at canopy scale remains challenging. In this study, we investigated whether spectral indices derived from the green-shoulder region can capture changes in tree hydraulic functioning. We combined hyperspectral drone imagery with continuous sap flow monitoring in a Norway spruce (Picea abies) forest in southern Sweden undergoing bark beetle infestation. Sap flux density (Js) was measured at three depths (1, 2, and 3 cm into the sapwood) to characterize hydraulic responses during infestation progression between weeks 16 and 32 of 2023. During the same period, seven airborne hyperspectral acquisitions were conducted using a VNIR sensor. We calculated indices from the Photochemical Reflectance Index (PRI) and Green Shoulder Curvature Ratio (GSCR) families and examined their relationships with weekly sap flow dynamics. We observed Js declines in one of more sensor depths in infested trees, occasionally leading to complete conductivity collapse and, consequently, tree mortality. Green-shoulder indices strongly responded to Js decline from week 26, especially in the outer sensor depth, indicating that spectral signals tracked changes in tree water transport. Our findings demonstrate that green-shoulder indices can reflect underlying hydraulic processes and provide a fast, scalable approach for monitoring hydraulic stress to support forest health monitoring.