ms'l. All of these sensors were samples at 10 seconds intervals and the data averaged over 1 
hour periods. 
4- RESULTS AND DISCUSSION 
Values of sensible heat flux were computed using data collected at the 
Railroad valley and Smoke Creek Desert sites, and the model are presented above. The only 
calibration parameter that changes from one the site to the other is the height of the substrate. 
It was set to 0.045 m for Railraod Valley and to 0.03 for Smoke Creek Desert. 
Figure 1 presents a comparison between the model and Bowen ratio 
estimations of sensible heat flux, during 9 days of the experiment in Smoke Creek Desert. In 
spite of some scatters especially during night time, the model seems to reproduce quite 
accurately the observed values of sensible heat flux. The r 2 value is about 0.91 and the Root 
Mean Square Error (RMSE) is about 35.38 Wm'-. Figure 2 exhibits similar comparison for 
Railroad Valley data. At this site, the model estimations of sensible heat flux olso are well 
correlated with the observed data, the r 2 is about 0.93 and the RSME is about 33.33 Wm -2 . 
The model estimates the observed decrease of H, probably due to some scattered showers, 
during the and the 6 ^ day. 
With the exception of some problems during night time, most likely due to the 
limitation of the formulation used to parameterise the aerodynamic resistance, this model is 
sufficiently general to yield satisfactory results for both sites. The average RMSE for both 
sites is about 35 Wm * 2 for a range of sensible heat fluxes values from -50 to 400 Wm* 2 . This 
results confirm that a multi-compartment approach is more suitable for sensible heat flux 
estimation over surfaces with sparse and non-homogeneously distributed vegetation 
(Chehbouni et al., 1993).