705 
obtained over homogeneous fields let us 
say, that the backscattering 
coefficient is related to the mean soil 
moisture when fields have a spatially 
distributed water content. 
4.2 Dependence of the physical optics 
model to volumetric water content. 
The physical optics model was evaluated 
by comparing the measured 
backscattering coefficients to each of 
the model’s computed backscatter i ng 
coefficient. The experimental setup 
provided a large sample of different 
surface soil moistures. We used two 
depths of sampling (0-1 cm ; 0-5 cm) to 
measure soil moisture. Surface 
roughness characteristics were limited 
to two apparently isotropic and random 
structures (s=0.011-0.014 m, 1=0.110- 
0.087 m) . 
Figure 8 compares for three angles of 
incidence, the relationship between 
predicted and measured backscattering 
coefficients. Whatever the angle of 
incidence and the soil surface 
moisture, it appears a systematical 
bias between predicted and measured 
backscattering coefficients. These 
discrepancies are related to the 
external calibration needed for 
experimental data. At 50 , the results 
are not satisfactory for the two 
sampling depths. Relationships are in 
fact not linear. These discrepancies 
may be related to the fact that the 
model does not account for multiple 
scatterings and shadowed effect of 
surface roughness which may be not 
negligible for large angles of 
incidence (Vi dal-Madgar, 1988). At 15 
and 23 , the relationships were linear. 
Slopes of the regression line were 
always higher in the case of the 0-1 cm 
depth of sampling. Near surface soil 
moisture ( 6o -1 ) seems to be more 
reliable to the effective surface soil 
moisture that contribute to surface 
scattering. Nevertheless, slopes are 
smaller than one. There is a slight at 
23 and heavy tendency at 15 for 
predicted data to overestimate the 
smallest experimental backscattering 
coefficients. These were always 
obtained in dry soil moisture 
conditions ( 8o - 1 < 0.1 0cm 3 /'cm 3 ) . This 
discrepancies can be related to the 
effect of volume scattering which is 
neglected by the model and becomes more 
important.in dry soil moisture 
conditions. 
In the case of relatively smooth 
surface roughness, the physical optic 
model is able to describe the 
dependence of the backscattering 
coefficient with volumetric water 
content in a limited range of angles of 
incidence. Results show that volumetric 
water contents sampled at 0-1 cm depth 
give more satisfactory results. A 
deeper depth of sampling (0-5 cm) leads 
to overestimate soil moisture that 
contributes to the scattering pattern. 
Figure 8: Dependence of the physical 
optics model to volumetric 
water content. Influence of 
the depth of sampling. 
(5.3 GHz, polarization HH, 
... First Bissector 
0-1 cm Regression line 
0-5 cm Regression line 
X 0-1 cm Sampling depth 
+ 0-5 cm Sampling depth 
Incidence angle 15* 
Incidence angle 23* 
Incidence angle 50* 
Experimental value (dB)