Integrating thermal stress indexes within Shuttleworth–Wallace model for evapotranspiration mapping over a complex surface

Abstract

The main goal of this work was to evaluate the potential of the Shuttleworth–Wallace (SW) model for mapping actual crop evapotranspiration (ET) over complex surface located in the foothill of the Atlas Mountain (Morocco). This model needs many input variables to compute soil (rss) and vegetation (rsv) resistances, which are often difficult to estimate at large scale particularly soil moisture. In this study, a new approach to spatialize rss and rsv based on two thermal-based proxy variables was proposed. Land Surface Temperature (LST) and Normalized Difference Vegetation Index (NDVI) derived from Landsat data were combined with the endmember temperatures for soil (Ts min and Ts max) and vegetation (Tv min and Tv max), which were simulated by a surface energy balance model, to compute the soil (Ts) and the vegetation (Tv) temperatures. Based on these temperatures, two thermal proxies (SI ss for soil and SI sv for vegetation) were calculated and related to rss and rsv, with an empirical exponential relationship [with a correlation coefficient (R) of about 0.6 and 0.5 for soil and vegetation, respectively]. The proposed approach was initially evaluated at a local scale, by comparing the results to observations by an eddy covariance system installed over an area planted with olive trees intercropped with wheat. In a second step, the new approach was applied over a large area which contains a mixed vegetation (tall and short) crossed by a river to derive rss and rsv, and thereafter to estimate ET. A Large aperture scintillometer (LAS) installed over a line transect of 1.4 km and spanning the total area was used to validate the obtained ET. The comparison confirmed the ability of the proposed approach to provide satisfactory ET maps with an RMSE, bias and R2 equal to 0.08 mm/h, 0.06 mm/h and 0.80, respectively.