This paper reports the application to vegetation canopies of a coherent model for the propagation of electromagnetic radiation through a stratified medium. The resulting multi-layer vegetation model is plausibly realistic in that it recognises the dielectric permittivity of the vegetation matter, the mixing of the dielectric permittivities for vegetation and air within the canopy and, in simplified terms, the overall vertical distribution of dielectric permittivity and temperature through the canopy. Any sharp changes in the dielectric profile of the canopy resulted in interference effects manifested as oscillations in the microwave brightness temperature as a function of canopy height or look angle. However, when Gaussian broadening of the top and bottom of the canopy (reflecting the natural variability between plants) was included within the model, these oscillations were eliminated. The model parameters required to specify the dielectric profile within the canopy, particularly the parameters that quantify the dielectric mixing between vegetation and air in the canopy, are not usually available in typical field experiments. Thus, the feasibility of specifying these parameters using an advanced single-criterion, multiple-parameter optimisation technique was investigated by automatically minimizing the difference between the modelled and measured brightness temperatures. The results imply that the mixing parameters can be so determined but only if other parameters that specify vegetation dry matter and water content are measured independently. The new model was then applied to investigate the sensitivity of microwave emission to specific vegetation parameters.

Abstract. The use of a large-aperture scintillometer to estimate sensible heat flux has been successfully tested by several investigators. Most of these investigations, however, have been confined to homogeneous or to sparse with single vegetation-type surfaces. The use of the scintillometer over surfaces made up of contrasting vegetation types is problematic because it requires estimates of effective roughness length and effective displacement height in order to derive area-average sensible heat from measurements of the refractive index. In this study an approach based on a combination of scintillometer measurements and an aggregation scheme has been used to derive area-average sensible heat flux over a transect spanning two adjacent and contrasting vegetation patches: grass and mesquite. The performance of this approach has been assessed using data collected during the 1997 Semi-Arid Land-Surface-Atmosphere field campaign. The results show that the combined approach performed remarkably well, and the correlation coefficient between measured and simulated area-average sensible heat flux was -0.95. This is of interest because this approach offers a reliable means for validating remotely sensed estimates of surface fluxes at comparable spatial scales. IntroductionThe turbulent heat fluxes near the ground surface are strongly affected by the ability of the surface to redistribute the radiative energy absorbed from the Sun and the atmosphere into sensible and latent heat. These fluxes play a key role in regulating the energy balance of the atmosphere, which in turn drives atmospheric circulation. For this reason, recent efforts have concentrated on improving the parameterization of landsurface processes in atmospheric models by taking into account the effect of surface heterogeneities on the exchanges processes [Avissat, 1995]. The problem, however, is the difficulty in validating model simulations at regional and certainly at the global circulation model (GCM) scale. On the other hand, it is necessary to validate GCM output because unless these models can reliably simulate the observed water and energy cycles in the present climate, future predictions of climate change are The objective of this study is to use a large-aperture scintillometer (LAS) to estimate areally averaged sensible heat flux over a transect made up of two adjacent patches (a grasscovered patch and a primarily mesquite-covered patch) with contrasting water status and roughness length. Scintillometerbased sensible heat flux is compared to a weighted average of those measured over each individual patch using two independent three-dimensional eddy correlation systems. The experiment took place in the San Pedro Basin within the context of 2505

Using the proposed Soil Moisture and Ocean Salinity (SMOS) mission as a case study, this paper investigates how the presence and nature of vegetation influence the values of geophysical variables retrieved from multi-angle microwave radiometer observations. Synthetic microwave brightness temperatures were generated using a model for the coherent propagation of electromagnetic radiation through a stratified medium applied to account simultaneously for the emission from both the soil and any vegetation canopy present. The synthetic data were calculated at the look-angles proposed for the SMOS mission for three different soil-moisture states (wet, medium wet and dry) and four different vegetation covers (nominally grass, crop, shrub and forest). A retrieval mimicking that proposed for SMOS was then used to retrieve soil moisture, vegetation water content and effective temperature for each set of synthetic observations. For the case of a bare soil with a uniform profile, the simpler Fresnel model proposed for use with SMOS gave identical estimates of brightness temperatures to the coherent model. However, to retrieve accurate geophysical parameters in the presence of vegetation, the opacity coefficient (one of two parameters used to describe the effect of vegetation on emission from the soil surface) used within the SMOS retrieval algorithm needed to be a function of lookangle, soil-moisture status, and vegetation cover. The effect of errors in the initial specification of the vegetation parameters within the coherent model was explored by imposing random errors in the values of these parameters before generating synthetic data and evaluating the errors in the geophysical parameters retrieved. Random errors of 10% result in systematic errors (up to 0.5°K, 3%, and ~0.2 kg m -2 for temperature, soil moisture, and vegetation content, respectively) and random errors (up to ~2°K, ~8%, and ~2 kg m -2 for temperature, soil moisture and vegetation content, respectively) that depend on vegetation cover and soil-moisture status.