Estimation of diurnal hydraulic status of spruce trees using drone-based hyperspectral images and green shoulder indices

Abstract. Tree hydraulic status fluctuates throughout the day as solar radiation and vapor pressure deficit (VPD) regulate transpiration and stomatal conductance. Remote Sensing (RS) can capture these dynamics, particularly through optical wavelengths in the “green shoulder” region (490–550 nm), which are sensitive to physiological changes in vegetation. This study evaluates the performance of the commonly employed Photochemical Reflectance Index (PRI) and four Green Shoulder Curvature Ratio (GSCR) indices derived from hyperspectral imagery as indicators of hydraulic status and photosynthetic efficiency in Norway spruce. Data were collected in the Tönnersjöheden experimental forest in southern Sweden. Four hyperspectral Unmanned Aerial Vehicles (UAV) flights were conducted throughout the same day on two occasions: Week 33 (mid-August) and Week 39 (late September) of 2025. Spectral indices were compared with sap flux density (Js) measurements (Week 33) through logistic regression leaf water potential (Ψleaf) data (Week 39) through linear models. Daily Green Shoulder Indices (GSI) time series were further compared with Photosynthetically Active Radiation (PAR) daily trajectories using the approximate area between curves. GSCR indices consistently outperformed PRI in following diurnal hydraulic status based on Js, while relationships with Ψleaf were mainly insignificant, likely due to the limited number of measurements available and less precise time pairing with UAV imagery. However, linear regression of the Ψleaf-GSI datasets followed similar trajectories as the Js-GSI relationships. GSCR indices generally tracked PAR trajectories more closely than PRI; however, results were mainly inconsistent between the two weeks, suggesting a lack of generalizability.