Estimates of Evapotranspiration with a One- and Two-Layer Model of Heat Transfer over Partial Canopy Cover

One of the applications of remotely sensed surface temperature is to determine the latent heat flux (LE) or evapotranspiration (ET) from held to regional scales. A common approach has been to use surface-air temperature differences in a bulk resistance equation for estimating sensible beat flux, H, and to subsequently solve for LE as a residual in the one-dimensional energy balance equation. This approach has been successfully applied over uniform terrain with nearly full, actively transpiring vegetative cover; however, serious discrepancies between estimated and measured ET have been observed when there is partial canopy cover. In an attempt to improve the estimates of H and as a result compute more accurate values of ET over partial canopy cover, one- and two-layer resistance models are developed to account for some of the factors causing the poor agreement between computed and measured ET. The utility of these two approaches for estimating ET at the field scale is tested with remotely sensed and micrometeorological data collected in an and environment from a furrowed cotton field with 20 percent cover and a dry soil surface. The estimates of LE are compared with values measured using eddy correlation and energy balance methods. It is found that the one-layer model generally performed better than the two-layer model under thew conditions; but only when using a bluff-body correction to the resistance based on a conceptual model of beat and water vapor transfer at the surface taking place by molecular diffusion into Kolmogorov-scale eddies. The empirical adjustment to the surface resistance with the one-layer approach assumed to be applicable for a fairly wide range of conditions was found to be inappropriate. This result is attributed to the significant size of the furrows relative to the height of the vegetation. Furthermore, a sensitivity analysis showed that the one-layer model with the empirical adjustment for the resistance was significantly affected by the changes in the surface roughness, whereas the physically based bluff- body correction was relatively insensitive to thew variations. For the two-layer model, a large change in the input variable for computing soil evaporation had a relatively small impact on the computed fuxes while a significant change in the leaf area index appeared to amplify the deviations between measured and modeled LE-values.