UAS-based high resolution mapping of evapotranspiration in a Mediterranean tree-grass ecosystem

Understanding the impact of land use and land cover change on surface energy and water budgets is increasingly important in the context of climate change research. Eddy covariance (EC) methods are the gold standard for high temporal resolution measurements of water and energy fluxes, but cannot resolve spatial heterogeneity and are limited in scope to the tower footprint (few hundred meter range). Satellite remote sensing methods have excellent coverage, but lack spatial and temporal resolution. Long-range unmanned aerial systems (UAS) can complement these other methods with high spatial resolution over larger areas. Here we use UAS thermography and multispectral data as inputs to two variants of the Two Source Energy Balance Model to accurately map surface energy and water fluxes over a nutrient manipulation experiment in a managed semi-natural oak savanna from peak growing season to senescence. We use energy flux measurements from 6 EC stations to evaluate the performance of our method and achieve good accuracy (RMSD ≈ 60 W m−2 for latent heat flux). We use the best performing latent heat estimates to produce very high-resolution evapotranspiration (ET) maps, and investigate the drivers of ET change over the transition to the senescence period. We find that nitrogen and nitrogen plus phosphorus treatments lead to significant increases in ET (P < 0.001) for both trees (4 and 6%, respectively) and grass (12 and 9%, respectively) compared to the control. These results highlight that the high sensitivity and spatial and temporal resolution of a UAS system allows the precise estimation of relative water and energy fluxes over heterogeneous vegetation cover.