703 
Physical sense or non-sense of such 
measurements have to be investigated to 
analyse the relationships between soil 
surface spatially distributed water 
contents and backscattering coefficient 
measurements. 
(3) Practical applications of 
microwave remote sensing in monitoring 
soil moisture are generally not limited 
to the optimal configuration as 
described above. In these conditions, 
microwave measurements are also 
affected by soil surface roughness. 
The use of general theoretical models, 
that predict the backscatteri ng 
coefficient according to the radar 
configuration (frequency, polarization, 
angle of incidence, soil surface 
moisture and roughness) will be useful 
to evaluate the effects of surface 
roughness and to isolate the effects of 
soil moisture. Such procedure will 
require accurate measurements of soil 
roughness parameters needed by the 
model. 
We present in this paper empirical and 
theoretical results collected from 1986 
to 1989 to evaluate the ability of 
microwave measurements to provide 
accurate soil moisture estimates at a 
field scale on bare soil under natural 
conditions. 
2. MATERIALS AND METHODS 
2.1 Test site 
Experiments were carried out on a bare 
field located at Montfavet (France). 
The soil was a clay loam (27% clay, 
61.7% fine and coarse loam, 11.1% 
sand). Several fields (0.1 ha) were 
tilled with a rotary digging machine, 
and the cloddiness of the soil surface 
was controlled to obtain various 
surface roughness conditions. Soil 
clods were arranged in an apparently 
random way and resulted from soil 
break-up tillage implements (no tillage 
directions). Rainfall was applied on 
fields when necessary to obtain various 
soil moisture contents. Rainfall was 
simulated using a sprinkler irrigation 
equipment composed of aluminium frame 
(20 m wide) with regularly distributed 
sprinklers and supported by two 
pneumatic tyred wheels on each side. 
The irrigation equipment was moved 
accross the experimental fields on two 
tramlines located on each side of the 
field. After wetting, intermediate and 
dry surface soil moisture conditions 
were obtained by natural evaporation. 
2.2 Ground Observations 
Ground observations of soil moisture, 
soil surface roughness and dry bulk 
density were performed on the 
experimental fields concurrently with 
scatterometer measurements. 
Once a day, 9 to 18 precise 
gravimetric water content profiles were 
sampled from 0 to 10 cm (0-1; 1-2; 2-3; 
3-4; 4-5; 5-7; 7-10 cm) at random 
locations to estimate the mean and the 
standard deviation of the soil water 
content at a field scale. 
Dry bulk density was measured using a 
field gamma-ray transmission apparatus 
having a precision of 20-30 kg*rrr 3 
(Bertuzzi et al., 1987). Volumetric 
water content profiles were easily 
estimated by combining gravimetric and 
dry bulk density measurements. 
An automated non-contact laser profile 
meter was used to sample the roughness 
profiles (Bertuzzi et al., 1990). The 
laser detector was supported by an 
aluminium carriage that automatically 
moved along an aluminium frame. This 
equipment was linked to a drive and 
data-logging unit in conjonction with a 
portable computer. Elevation data were 
recorded with a 2 mm sampling interval 
and an accuracy better than 0.25 mm. 
2.3 Scatterometer measurements 
Microwave measurements were made using 
the scatterometer RAMSES designed by 
the Centre National d’Etudes Spatiales 
(C.N.E.S). It was a frequency-modulated 
continuous wave, mu 11ifrequency and 
multipolarization system. The 
scatterometer was mounted on a mobile 
platform of a crane-boom. The antenna 
was at a 19.70 m height above the soil 
surface. It had a 3 dB beam width of 
2.65 at 5.3 GHz. The radial soil 
surface resolution varied from 2.0 m at 
nadir to 4.3 m at 50 of incidence. 
Independent measurements of 
backscattering coefficients were 
sampled along tracks ranging from 4.5 
to 12 m, and the average value was 
calculated. 
3. THEORY 
3.1 Microwave penetration model 
Several attempts have been made to 
provide more or less precise microwave 
penetration models, including effects 
due to simple soil surface reflection 
or mu 11i-ref 1ection within the soil 
layers (Pausader, 1982). To analyze the 
relationships between soil water 
content profile and microwave 
penetration depth, let us summarize the 
main steps of the calculations 
(Bruckler et al., 1988). 
The air was regarded as a real medium 
and the soil a complex one. We assumed 
that the soil depth was divided into 
’n’ thin soil layers. At each soil 
layer boundary, the incident wave 
amplitude was partly transmitted, and 
partly attenuated within each layer. 
Only one reflection was taken into 
account at each boundary between two 
different layers. 
Starting from the basic Snell- 
Descartes law, it was possible to 
calculate the transmitted amplitude 
from the soil surface to the soil 
layers and the attenuated amplitude 
within each soil layer. Using an 
iterative numerical procedure for 
calculating the transmission and 
attenuation coefficients for each soil