Using spectral and thermal UAS data to infer the influence of shaded and unshaded urban vegetation on evapotranspiration and land surface temperature

As the urban population has become predominant globally, heat stress and its negative consequences on human health have grown due to increasingly dense and artificial environments. Urban green infrastructures (UGI) mitigate heat stress by providing cooling services through evapotranspiration (ET) and by blocking solar radiation through shading. Even though ET is a crucial component of urban water and energy regimes, our understanding of the role of vegetation on urban water cycling is still poor when observed through remote sensing. To better understand the seasonal and diurnal variability of ET from urban vegetation, a comprehensive sampling campaign combining an unmanned aerial aircraft system (UAS) and field-based measurements in an urban ecohydrological research observatory in Berlin (Germany) was conducted. The sampling was undertaken throughout an entire growing period (from April to October 2019) to characterize the seasonality of both climatic drivers and phenological effects on ET. Three vegetation types were sampled in the study site  (grassland, forest, and shrubs). 

Field-based measurements included sap flow and stomatal conductance (LI-6800 gas exchange), to capture monthly and diurnal dynamics of transpiration, leaf area index (LAI), grassland vegetation height as well as soil moisture. Soil moisture and sap flow were available at hourly resolution while LAI, stomatal conductance and vegetation height were measured at monthly intervals. The images were captured by UAS flights with multi-spectral (Tetracam MCA) and thermal (Flir Tau 2) cameras on a monthly basis and, on some dates, at multiple times during the day to capture diel variability. UAS data were divided into shaded and unshaded areas within the three vegetation classes. ET estimates from UAS observations were derived using the inference method based on vegetation indices (VI) as described in (Nouri et al., 2013), Eddy flux data was used to validate modeled ET and also provided hydroclimatic data . 

Results showed clear differences for ET and land surface temperature (LST) between vegetation classes throughout the year, with trees and shrubs showing lower overall temperatures and higher ET estimations than grassland during the observation period. The influence of shadow on modeled ET and observed LST also became apparent for all classes, especially when multiple UAS observations were taken during a single day. Shaded areas exhibited lower overall LST and ET than non-shaded areas, with the starkest contrast exhibited for grassland where shaded areas showed up to 50% lower LST and estimated ET was reduced by up to 25%. Both ET and LST showed correlation to the measured sap flow and stomata conductance at both diurnal and seasonal temporal scale.

Our findings provide important insights into the influence of  different urban vegetation types in both ET and LST with respect to shaded and unshaded surfaces. Our study also highlights the importance of a detailed understanding of UGI characteristics and its cooling potential for further improvements in urban green management. Moreover, it could improve models of the urban water cycle and is important for upscaling ET to a broader city scale.