123 
rest arrive 
eiver (section 
rpreter to 
étions 4 and 5) 
' pre- 
reas question- 
eight of flight 
ound range dis- 
idth and low 
(DUTSCAT) ma- 
from different 
easured in dif- 
em in every 
ally) repre- 
ENTS USING 
easurements of 
ted with the use 
iment of this 
ite over stands 
R. In 1985 these 
X-band 
e have not been 
ectors with an , 
pace) of 45 dBm 
and 3 of 38 dBm 2 were especially manufactured for this 
purpose. They are dismountable and were assembled in 
situ under the canopy at the forest floor. A 
detailed description of the experiment design, the 
method used to acquire the attenuation factors and 
the detection probabilities of covered corners are 
given by Hoekman (1986). In this section only the 
results of the campaigns and its importance in model 
research will be discussed. 
3.2 Discussion of X-band results 
Table 3 shows the overall results of 2 X-band HH- 
polarized SLAR flights. The other flights have not 
been analysed yet. In each flight the run over the 
6 corners is repeated once. Substantial between 
run variations were noticed and could be explained 
from changing sensor aspect (Hoekman 1986) . For some 
corners no signal could be perceived therefore only 
lower limits could be extracted. The signal from 
corner 3 in the poplar stand probably saturated the 
receiver therefore an upper limit is given. 
Table 3. Equivalent two-way attenuation factors T e q 
(in dB) in X-band, HH-polarization and at 45 degrees 
incidence angle for poplar, oak and pine. 
and crowns are mostly .small and light. Corners 1 till 
4 inclusive and 6 were carefully placed behind groups 
of crowns thereby avoiding illumination of the corners 
through big gaps in the canopy. Corner 5 was placed 
at a large open spot. 
It can be concluded that the attenuation through the 
forest canopy in general fluctuates strong and de 
pends on relative orientations of gaps with respect 
to the sensor. The attenuation of crowns of deciduous 
trees is high. The attenuation of crowns of coniferous 
trees may be lower but still is in the order of 20 dB 
(two-way). Therefore at X -band contribution of scat- 
terers of the understory and forest floor is prima 
rily determined by canopy architecture (i.e. factors 
like crown closure, crown shapes etc.) and angle of 
measurement. 
3.3. Modelling aspects illustrated by the cloud model 
The attenuation properties of vegetation layers is an 
important parameter in many backscatter models. In 
one of these models; the cloud model formulated by 
Attema and Ulaby (1978) and Hoekman et al. (1982), 
gamma of a vegetation layer is modelled as single 
scattering by a low density cloud of small identical 
particles : 
Y veg = °/ 2 q [l-exp(-2NQh/cos0i)] (2) 
Flight over Roggebotzand test area 
where N is the number of particles per unit volume, 
tree species 
run 1 
run 2 
h is the height of the vegetation 
layer, 
corner 
i 
poplar 
>28. 
>28. 
O is the radar cross section of a 
particle ■ 
corner 
2 
poplar 
12.5 
21 .5 
Q is the attenuation cross section of a par- 
corner 
3 
poplar 
< 7. 
10.0 
tide. 
corner 
4 
oak 
22. 
17.7 
corner 
5 
oak 
>22. 
>21 . 
Assuming the soil scattering to add incoherently to 
corner 
6 
oak 
>21 . 
22.6 
the vegetation scattering the total gamma 
value be- 
comes ; 
Flight 
over 
Kootwijk test 
area 
Y ,=Y +exp(-2NQh/cos0 1 -) .Y . 
'total 'veg v v 'soil 
(3) 
tree species 
run 1 
run 2 
corner 
i 
pine 
19.5 
20.4 
With T=exp(-2NQh/cos0^) 
(4) 
corner 
2 
pine 
19.3 
22.2 
corner 
3 
pine 
>24. 
>24. 
corner 
4 
pine 
>17. 
>17. 
corner 
5 
pine 
5.2 
6.2 
this equation is simplified to 
corner 
6 
pine 
11 .0 
18. 1 
Y total" AoP-d^-Vaoil 
(5) 
For an interpretation of these results a description 
of the forest stand architecture is indispensable. 
The stand of poplar (clone 'Robusta') is 25 years 
old, with trees planted in rows every 8 meter. The 
row spacing is also 8 meter, tree height is about 
28 meter. The canopy features no crown closure; 
distances between crowns range from 1 to 2.5 meter. 
The measurement direction was in row direction at 
45 degrees incidence angle. Corner 1 was placed in a 
row. As a result 4 trees (2 crowns and 2 stems) were 
between corner and sensor. This completely blocked 
any perceivable signal from the corner. Corner 3 was 
placed in the middle between 2 rows. Now only a few 
branches were situated between sensor and corner resul 
ting in a low attenuation. Corner 2 was placed be 
tween rows, 2 meters from one row and 6 meters from 
the other. Now the outer parts of 2 crowns were 
between sensor and corner. The big difference in 
measured attenuation factors (12.5 versus 21.5) could 
be explained by between run variation of the relative 
orientation of sensor, crown and corner. 
The stand of oak (Quercus Robur) is 25 years old 
and has a high degree of crown closure; there are 
small gaps (<1 m 2 ) but almost no big gaps. The tree 
height is about 12 meters. The pines (Pinus Sylves- 
tris) at the Kootwijk site are 55 years old and about 
12 meters tall. The site factors do not favour a 
prosperous growth. The degree of crown closure is low 
The parameters T as well as a and Q depend on frequen 
cy, angle of incidence (for non-spherical particles) 
and polarization (for non-spherical particles). 
However the forest canopy is not expected to con 
sist of identical scatterers. By assuming the vege 
tation canopy to be composed of n discrete layers 
with different types of scatterers the cloud model 
equation can be extended. It follows from induction 
that 
Y veg =Yl+TlY2+TlT2Y3+ ‘‘- +TlT2 -• ,T n-l Y n ( 6a ) 
where index 1 stands for the top layer, 
Y i =ai/ 2Q i *t 1_T i3 
(6b) 
aiK * Ti=exp (-2N£Q^d/cos0) 
(6c) 
With _ _ _ _ 
T t =T 1 T 2 T 3 ...T n 
(7) 
it follows Y t *. i“Y +T ^‘Y * i 
'total veg t soil 
(8) 
When the cloud model assumptions are followed, the 
equivalent attenuation factor T eq measured with the 
corner reflector experiment, with corners placed at 
the ground, is the same as the parameter T t of the n- 
layer cloud model. Thus equation (8) can be written 
as