second one is the Tortuosity index (T) which is defined 
by the ratio : 
L -Lo 
T = 
Lo 
where L is the actual length of the profile and Lq is the 
projected horizontal length of the profile. One major 
interest of the tortuosity index is related to its smaller 
sensitivity to field variability compared to other 
roughness indices (Bertuzzi et al., 1990b). 
Soil surface changes resulting from the sealing processes 
was evaluated on three localized 1 m 2 surfaces of 
reference using the method proposed by Boiffin (1984, 
1986). According to visual cntera, the evolution of the 
soil surface structure during slaking can be classified in 
five typical stages. (1) So, initial soii surface resulting 
from tillage. (2) Si strutural crust. Soil aggregates and 
clods disintegrate due to raindrop impacts.Continuous 
patches appear and expand due to interstitial infilling. 
(3) S\ + local appearance of depositional crusts. (4) S2 
depositional crust; the fragmented layer becomes 
continuous; depositional areas are formed in small 
surface depressions where puddles appear during 
rainfall (5) SU the soil surface appears to be completely 
close and continuous. 
3. RESULTS AND DISCUSSIONS 
3.1 Ground measurements 
Mean values of soil volumetric water content were 
estimated from the 10 samples. They were plotted 
against the degree of slaking in Figure 3. The vertical 
bar corresponds to + /- standard deviation of the 10 
samples. In wet condition, soil draining contributes to 
decrease soil moisture very quickly during the diurnal 
phase of the day. Dry condition corresponded to mean 
water contents less than 0.19 cm +3 .cm' 3 . This limited 
value may appear high. They must analysed in the 
context of the chosen sampling procedure. Whatever 
the spectral bands, it is not possible to attribute a real 
physical sense to calculated soil moisture. If fact, soil 
moisture gradients are generally very high in the top 
soil layers particularly in dry conditions. The sampling 
depth averaged the effect of water gradients. For optical 
measurements, dry conditions always corresponded to 
uniform and light soil surface. 
Table 2 shows the decrease of the two roughness indices 
as functions of the cumulative rainfall and the stage of 
slaking. These are mean values computed from the 
results of the twelve sampled roughness profiles. 
3 2 Microwave measurements 
In Figure 4, the backscattering coefficient is plotted 
against the Tortuosity index. The vertical bar displays 
the magnitude of the three backscatter measurements 
obtained during the diurnal phase of the day for each 
soil moisture condition and the corresponded stage of 
slaking. Whatever the angle of incidence, a change in 
soil moisture conditions increases the curve level and 
Figure 3 : Mean volumetric water content as 
a function of the stage of slaking. 
Table 2. Mean roughness indices as functions 
of cumulative rainfall and the stage 
of slaking. 
Stage 
Rain 
(mm) 
s 
(m) 
T 
So 
0 
0.013 
1.79 
Si 
30 
0.012 
1.60 
su 
64 
0.010 
1.38 
S2 
100 
0.009 
1.20 
S 2 + 
124 
0.009 
1.18 
influences 
the variability 
of the 
backscatter 
measurements. It is always smaller in wet condition 
although the volumetric water content decreases rapidly 
( Figure 3) 
At 50° of angle of incidence, the backscattering 
coefficient, exhibits a small increase with the Tortuosity 
index. But, the sensitivity of the backscattering 
coefficient is the order of magnitude of the observed 
variability. At nadir (0°), the backscattering coefficient 
is more sensitive to roughness changes due to slaking 
whatever the soil moisture conditions. 
In term of the standard deviation of height (s), the 
decrease of surface roughness (about 4 mm) due to 
slaking may appear very small. Consequently, only 
backscattering measurements of the coherent 
component obtained in the specular direction are 
sufficiently sensitive to detect roughness changes due to 
slaking. This will be probably verified whatever the 
initial roughness conditions. 
117