500 
Frequency 
Observations 
E(K) 
v 
Mv t 
23.8 GHz 
179 
6.7 
0.60 
0.17 
36.5 GHz 
179 
7.5 
0.81 
0.17 
90 GHz 
364 
4.7 
1.19 
0.17 
Table 1 - PORTOS-derived parameters Mv t and fof the Wang and Schmugge (1980) model. E is the 
rms error on TBs. 
Given the calibrated Wang and Schmugge (1980) model, it is now possible to study the feasability of 
temperature and moisture (of the 5-mm surface layer) retrieval at high frequencies. 
3.2. Surface temperature and surface soil moisture of a sparse canopy from high frequencies 
3.2.1 Models. For rough surfaces, the (specular) Fresnel coefficients-derived reflectivities (Rsp) are not valid. A 
common approach at low frequencies consists in modifying Rsp according to empirical parameters h and Q 
(Choudhury et al. 1979), h depending/or not on the incidence angle. The rough reflectivities (Rp) are expressed 
as 
Rp = (Q Rsq + (1-Q) Rsp) exp(-h) (1) 
For satellite data interpretation, Kerr and Njoku (1990) also use this simple model at high frequencies. In the 
case of sparse canopies, vegetation can be tentatively included into the soil roughness parameters h and Q. One 
may also use a simple emission model for the vegetation layer, like the continuous approach described in 
Wigneron et al. (1993). The problem with the latter approach is the limitation of the validity domain at high 
frequencies. As in Calvet et al. (1994), we suppose that such a model can be employed till 37 GHz. 
3.2.2 Temperature and near-surface soil moisture retrieval. Measurements were performed over a growing 
Sorghum sparse canopy. IR temperature, 5-mm soil moisture, and vegetation biovolume characteristics were 
measured. From the initial bare soil, the plant height and the vegetation coverage reached, respectively, 0.7 m 
and 30%. For 23.8 and 36.5 GHz, h and Q parameters are fitted with the initial bare soil data. The value of h is 
found to be incidence-angle independent. The vegetation continuous-model correlation lengths are obtained at 
the end of the period during moist soil conditions. As far as the 90 GHz frequency is concerned, Eq.(l) is 
employed for both bare soil and developed canopy conditions. The soil-vegetation parameters h and Q are 
determined as linearly dependent on the plant height. 
Once the structure parameters are determined, temperature and near-surface moisture are retrieved from 
the PORTOS TBs and compared with the surface measurements. Figures 4 and 5 show the results for 23.8-36.5 
(simultaneously) and 90 GHz channels, respectively. 
Figure 4 - Temperature and near-surface soil moisture obtained from PORTOS data (fine lines) at 23.8 
and 36.5 GHz over a growing sparse Sorghum using the continuous approach, compared with measurements 
(thick lines).