Albedo Accuracy Impact on Evapotranspiration Estimation
Mattar, Cristian¹; Franch, Belen²; Sobrino, J.A.²; Corbari, Chiara³; Jiménez-Muñoz, Juan Carlos²; Olivera, Luis¹; Skokovic, Drazen²; Soria, Guillem²; Oltra-Carrió, Rosa²; Julien, Yves^2
1University of Chile, CHILE; ²University of Valencia, SPAIN; ³Politecnico di Milano, ITALY

Albedo is one of the most essential variables in the surface energy balance. During the last decades several land surface energy models have developed accurate albedo estimations for achieving reliable daily evapotranspiration retrievals. Despite these improvements in the surface energy models, we detected a systematical approach that most studies assume in this framework. It consists of considering that the surface broadband albedo can be estimated directly as a weighted average of surface reflectance. However, this approach does not take into account the surface anisotropy, which is described by the Bi-directional Reflectance Distribution Function (BRDF). The surface albedo is related to land surface reflectance by directional integration and is therefore dependent on the BRDF. In this paper, we analyze the influence of estimating the land surface albedo directly from the surface reflectance or through the BRDF integration in the estimation of energy balance components such as the net radiation, latent and heat flux and consequently in the land surface evapotranspiration. To this end, we processed remote sensing and in-situ meteorological data measured at the agricultural test site of Barrax in the framework of Earth Observation: optical Data calibration and Information eXtraction (EODIX) project. Remote sensing images were acquisitioned for different View Zenith Angles (VZA) by the Airborne Hyperspectral Images (AHS).

Results have shown that albedo estimations derived from BRDF model present stability through every image while albedo estimations using single reflectance presented high variation depending on the VZA. The highest difference was observed in the backward scattering direction along the hot spot region obtaining a RMSE of 0.11 through the AHS image which implied a relative error of 65%. The net radiation presented relative errors from 6 to 17%, obtaining the maximum error in the images that included the hot spot effect. We did not observe significant changes in case of the ground heat flux and the evaporative fraction. However, the sensible heat flux, the latent heat flux and the daily evapotranspiration showed relative errors ranging between 23-39%, 6-18% and 5-15% respectively. However, contrary to albedo and net radiation where the highest error was obtained for the hot spot image, in case of H, LE and daily ET the highest error corresponded to the image with highest VZA in the solar plane. Finally, this work has analyzed the error committed by many evapotranspiration studies that assume the surface as Lambertian and estimate the albedo from a surface reflectance weighted average. The use of BRDF method to estimate surface albedo improves the estimation in other flux variables such as net radiation, sensible and lanten flux, evaporative fraction and even affect the scatter distrbitution in the S-SEBI algorithm