UAS ecophysiology: Imaging spectroscopy can map canopy leaf water potential in a diverse forest ecosystem

Abstract

Forests worldwide are facing increasingly frequent and severe droughts, leading to canopy dieback, tree mortality, and losses in biodiversity and ecosystem function. Understanding spatial patterns of plant water stress in relation to physiological drought responses is essential for identifying species vulnerability and resilience under a changing climate. Here, we used high-resolution short-wave infrared (SWIR) imaging spectroscopy (900–1700 nm) acquired from an Unoccupied Aerial System (UAS) to infer leaf water potential (Ψleaf) across a diverse native Australian forest. Random Forest regression models developed from laboratory-based dehydration experiments against Ψleaf were applied at the species level using a UAS-derived species classification map. Models were evaluated in the field across two sampling periods, one during severe drought and another following rehydrating rainfall. Model performance was strong, with RMSE values below 10% of the full water potential range for three of the five canopy species (Eucalyptus viminalis, Eucalyptus pulchella, and Melaleuca pustulata) and below 18% for the remaining two (Callitris rhomboidea and Allocasuarina verticillata). Combined, the predicted values showed strong agreement with field measurements (R2 = 0.64 across all species, 0.80 excluding A.verticillata). These results demonstrate the potential of UAS imaging spectroscopy to produce fine-scale, spatially explicit maps of plant physiological water stress, enhancing our understanding of forest ecophysiology and improving the monitoring of drought impacts and resilience.