1038 
in -i 
TH 
$¡3 T 1 1 i ' i 1 
'0 20 <0 60 80 
GRAZING ANGLE 
a) 12% of weight 
«H—■—i—*—i—■—;—•—i 
1 0 20 40 60 80 
GRAZING ANGLE 
b) 7% of weight 
8 I 1 ' 1 ' 1 1 1 
1 0 20 40 60 80 
GRAZING ANGLE 
c) 2% of weight 
Figure 11. Backscattering coefficient of bare soil versus grazing angle (90°-9) 
for different moisture contents. The Netherlands ROVE program 1977. 
Figure 12. Backscattering coefficient y versus inci 
dence angle for different soil roughness values. The 
points represent actual measurements at 3 cm wave 
length. 
One of the more striking things is that, for all 
fields, the curves for the HH and W polarization 
combinations are nearly the same. The cross-polar 
measurements show the same type of behaviour but at 
a 6-10 dB lower level. From the wave interaction point 
of view this means that the objects do not show a 
dominant orientation, at least not at 3 cm wavelength. 
The shape of the y-curves is to a large extent in 
fluenced by the roughness, the most characteristic 
profile corresponding with the smoothest surface. Al 
though there seems to be a lot of information in the 
way radar scattering depends on the incidence angle 
it is not realistic to assume that in the near future 
this incidence angle can be scanned over an appreci 
able interval. 
Not only the shape but also the level of the curves 
is changing with roughness. It should be noticed in 
this respect that, although fields 1 and 2 have 
approximately the same roughness, there is a differ 
ence in the look direction of the scatterometer. For 
field 1 the plough-furrows are perpendicular to this 
direction whereas they are parallel to it for field 
2. Obviously the latter situation makes the surface 
roughness less effective. 
The amount of scattered power however is not only 
dependent on the roughness of the soil but also on 
its moisture content. Soil moisture affects the radar 
response in such a way that y changes for all inci-» 
dence angles by about the same amount. This soil 
moisture influence is clearly illustrated by the ex 
amples in fig. 11. The minor deviations are caused by 
roughness changes due to rainfall. 
Supposing that eq.(4) were a good description of 
0° for bare soil then the moisture influence could be 
represented by a common factor m: 
0° = m[ F(0)exp(-p 2 )+1/2 {1-exp(-p 2 )}] (7) 
It is interesting to note that, under this assumption, 
an incidence angle 0^ satisfying the equation 
F(0 )= 1/2 
c 
will make 0° independent of roughness. The model des 
cribed by eq.(7) was used successfully (Attema e.a. 
1982) to fit bare soil measurements in case of small 
surface roughness (fig. 12). 
A second type of data collected by scatterometers 
is presented in fig. 13. The curves show the develop 
ment of a number of crop types during the growing sea 
son. Since the shape of the curves, as we will see, 
cannot be explained by vegetation effects alone also 
a bare soil field was measured for reference purposes. 
Each marked point in the figure indicates a measure 
ment. 
Up until June 2 the coverage of the soil by vege 
tation is still small and therefore the underlying 
soil plays a dominant role in y. The peak on May 7 
is due to a variation in soil moisture caused by a 
10 mm rainfall shortly before the measurement. 
Further peaks in the bare soil response later in time 
are indications for sudden variations in soil moisture 
as well. The gradual decrease in the bare soil scat 
tering is caused by the effect that, due to rainfall 
and slaking, the soil roughness decreases through 
time. 
After June 2 differences begin to occur due to the 
increasing contribution of the vegetation. In this 
period the soil moisture peaks are smoothed by the 
attenuation in the canopy. As can be seen the total 
range in y is in the order of 20 dB (mid-season val 
ue) . It is within this range that discrimination for 
classification and monitoring of crops has to take 
place. This requires adequate resolution and stabil 
ity of operational radar systems. 
Data like the one shown in fig. 13 was used to 
simulate crop classifications (Smit 1979). From this 
study it was learned that, although the differences 
in scattered power are small, classification of crop 
WR HAY 
1980 
LEGEND ■ O SUGfiRBEET 
Figure 13. Back 
plots. Incidenci 
0 o 
Figure 14. Geom< 
types should be 
results being tl 
greatly improve; 
analysis. Althoi 
add some inform; 
the classificat: 
vations can be < 
growing season c 
imitated to ver: 
study (Hoogeboor 
Multitemporal 
their full extei 
with sufficient 
ask for a raise 
offer at the san 
memory function 
ory function wi] 
important elemei 
sensing systems, 
respect to monii 
for meaningful c 
in different yec 
pean or even on 
An important £ 
the relative imj 
contributions ir 
of the, so-calle 
The underlying i 
trie constant oi 
er than the diel 
canopy is usuall 
volume. Therefoi 
water cloud, the 
by the vegetatiLi 
assumed that thi 
droplets with th 
random distribut