Development of micrometeorological models for field-scale determination of evapotranspiration and water status of a discontinuous crop canopy

The feasibility of remote sensing of water status and evapotranspiration in a vineyard canopy was studied by developing mathematical canopy micrometeorology models and testing the models at three commercial vineyard sites subjected to periodic irrigation cutoff. The canopy micrometeorology models served to translate the remotely-sensed signal (infrared exitance) and on-site micrometeorological measurements into an estimate of crop water status (stomatal conductance, gs) and evapotranspiration (latent heat flux density, λE).;Unlike the gs estimates, the quality of the λE estimates were highly dependent on the turbulent transport scheme used in the bottom-up model. The higher order closure scheme provided better estimates of λE than the K-theory scheme, the latter tending to overestimate λE. Unlike the bottom-up model, the top-down model was directly sensitive to errors in canopy H estimates as well as to errors in the measurement of net radiation and soil heat flux. Hence the top-down model tended to have large errors in its ET estimates while the bottom-up estimates (using higher order closure) were more accurate.;The bottom-up model may benefit from additional research concerning the spatial and angular distribution of leaves within typical vineyard canopy hedgerows. Additional knowledge in this area may improve model performance by increasing the accuracy of radiation distribution estimates both within the hedgerow and at the soil surface.;Two classes of canopy models were examined: top-down (big leaf) and bottom-up (multilayer) models. Additionally, two variations of each model class were evaluated against one another: (1) a top-down model with single- and two-layer (soil and foliage) composition and (2) a bottom-up model with K-theory and higher order closure turbulent transport schemes. In general, the differences in the quality of the gs estimates between top-down and bottom-up models was much greater than differences between the two variations of each type of model. The bottom-up model produced more accurate estimates of gs than the top-down model at all three experimental sites. The difference was most pronounced at the 1994 site, which had more compact hedgerows than the other two sites and, hence, more bare soil exposure. At that site, the top-down model greatly underestimated gs while the bottom-up model produced good estimates of g.s.