UAV-based imagery for monitoring and predicting vegetation water stress across landscape-scale gradients

The extreme summer droughts across Central Europe (i.e., 2003, 2015, and 2018), driven by anthropogenic climate change, emphasized the urgency of understanding and predicting ecosystem responses to extreme droughts. Water limitation during severe drought limits photosynthesis and respiration by closing stomata, induces xylem cavitation, and reduces plant carbon balance, which leads to seasonal decreases in productivity and long-term increases in tree vulnerability to major disturbances and mortality. Drought-induced stress can be measured by remote sensing as it influences physical leaf properties and alters leaf spectral responses in both the visible and thermal part of the spectrum. The physiological responses to drought events not only depend on their timing, i.e. recurrence and duration, but also on their geography (landscape-scale heterogeneity). For example, large gradients in soil depth, slope, and exposition in mountainous landscapes can therefore cause differential vegetation responses to drought across scales of 101-104 m. Combining both vegetation (spectral) indices, and a highly variable geography, offers a non-destructive and rapid method for investigating plant physiological processes under a wide range of drought stress. Our research maps temporal variations and landscape-scale heterogeneity in vegetation water stress using UAV-based multispectral remote sensing. To investigate landscape-scale heterogeneity of drought impacts, the study is carried out at various sites in Valais, Switzerland, with different elevations, soil and plant rooting depths, slopes, and various species exhibiting varying responses to drought stress. All sites are part of a larger tree monitoring network and provide co-located plant and soil-point measurements of water stress. Here we outline the heterogeneity in the study locations, the methodological approach, as well as tree responses during recent summers at these sites. The collected data will be used to develop predictive models for water stress using UAV imagery, aiming to upscale the effects of water stress on vegetation functioning across a heterogeneous landscape.