Estimating evapotranspiration of arid regions using remotely-sensed data

Classified airborne high resolution multispectral video imagery (pixel size of 0.16 m) and ground reflectance measurements were used to estimate and map the energy balance terms, namely, net radiation (R_n), sensible heat flux (H), and ground heat flux (G) in a desert environment. The data were gathered during summer 1994, at Goshute valley, Nevada, USA.Energy balance fluxes were measured at 10 sites using Bowen ratio and eddycorrelation systems. Ground-based and airborne remotely sensed data were taken at the same time during the experiment. Supervised classification wasconducted on each high resolution image of the sites to estimate the proportions of each surface (i.e. playa, organic soils, Greasewood, Sagebrush, and Shadescale). Surface temperature was mapped for each site using airborne thermal imagery (pixel size of 0.30 m) obtained using an infrared thermal scanner. R_n was estimated using the PIT ratio suggested byJackson (1994) for estimating albedo. H was modelled using vegetation parameters extracted from the multispectral video imagery. The values of G/R_n were exponentially related to the soil adjusted vegetation index (SAVI). Maps of the energy balance fluxes were produced based on the class distribution at each desert site. The good agreement between the observed and estimated surface energy fluxes suggests that maps of surface energy fluxes for sparsely vegetated arid regions could be produced at low cost using airborne sensors and used for input and verification of meso-scale atmospheric and energy balance models.