Solar Induced Chlorophyll Fluorescence Downscaling Enables Early Detection of Canopy Drought Stress and Forest Mortality Risk in Deciduous Sal Forests of Bangladesh

Climate change-induced drought is increasing the vulnerability of tropical forests, yet early detection of physiological stress remains challenging because conventional vegetation indices primarily capture structural canopy changes after substantial functional decline has occurred. Solar-induced chlorophyll fluorescence (SIF) provides a direct indicator of photosynthetic activity, but the coarse spatial resolution of current satellite-derived SIF products limits its application in heterogeneous forest ecosystems. This study evaluated the potential of SIF downscaling for detecting canopy drought stress and forest mortality risk in the deciduous Sal (Shorea robusta) forests of the Madhupur landscape, Bangladesh. Satellite-derived SIF observations from the GOSIF v3 dataset were integrated with vegetation indices (NDVI and EVI), ERA5-Land climate variables, and field-based physiological measurements. A machine-learning downscaling framework combining Random Forest, Gradient Boosting Machine, and residual geostatistical interpolation was developed to generate high-resolution (~500 m) SIF products. Model performance was evaluated using cross-validation and ground observations, while drought sensitivity was assessed using the Standardized Precipitation Evapotranspiration Index (SPEI). The Gradient Boosting model achieved the highest predictive performance (validation R² = 0.75). Downscaled SIF showed strong agreement with canopy- and leaf-level fluorescence measurements and was significantly correlated with soil moisture and SPEI. Importantly, SIF declines consistently preceded reductions in NDVI and EVI, indicating earlier detection of drought stress. Areas exhibiting sustained fluorescence reductions were associated with elevated mortality risk, particularly within fragmented forest patches. These findings demonstrate that SIF downscaling can provide an effective early-warning framework linking physiological stress, drought dynamics, and forest mortality risk, supporting conservation planning and climate-resilient management of tropical deciduous forests.