1173 
Fig. 5: Adjustment of (Ts-Ta)/(Tsmax-Ta) to the 
volumetric soil moisture. Soil temperatures 
were measured with a hand-held thermo- 
radiometer on Kendall site (MF 5). 
Fig. 6: comparison between remote estimate of 
sensible heat flux (ERS-l/TM) and eddy 
correlation measurements on Lucky Hills site 
(MF 1). 
5. CONCLUSION 
This study has presented a multi-sensor scheme, combining radar and thermal data, to retrieve 
the main driving variables of two-layer model suited to sparse vegetation of semi-arid areas. The use of ERS-1 
SAR images was a priori a powerful means to monitor soil moisture on these areas considering the generally 
low development of vegetation biomass and the resulting low vegetation effects on a°. The results has shown 
in fact that the particular structure of this vegetation, particularly dead material buildup characteristic of semi- 
arid rangeland, plays a major role on backscattering behavior and thus that the amount of vegetation biomass 
was not the main driving variable. Moreover the evolution in time of this structure makes difficult the multidate 
use of these images without a modelling approach. At last soil roughness spatial variability has appeared as a 
dispersion factor if not accounted for. 
However a statistical model has been designed with simple assumptions about vegetation biomass 
and provided good estimate of soil moisture that can be used in conjunction with thermal data for sensible heat 
flux modeling. The proposed approach needs yet further validation with various vegetation and moisture 
conditions to be fully operational. 
6. ACKNOWLEDGEMENTS 
This research was made possible by the cooperative spirit of Steve Land of EOSAT Corp. and 
Guy Duchossois of the European Space Agency who provided Landsat TM and ERS-1 images at no cost. 
Support was also provided by the NASA Interdisciplinary Research Program in Earth Sciences (NASA Ref. 
Num. IDP-88-086), the NASA Eos Program (NASA Ref. Num. NAG-W2425) and NSF (BSC-8920851).