185 
rn 
ise. 
scan 665 with 
ita of both the 
Lai naviagion 
ne tagged so 
ìan time correc- 
; and the an- 
:ric correction 
(8) 
itenna coordi- 
rmula is the 
: cross section 
±>er to which 
;ed power 
^stem's inter- 
:he direction 
, For the 
i) a value was 
external cali- 
>een measured 
this function 
respect to the 
.on) . The re- 
i elevation 
ixample given 
iction 0 o is 
most accurate 
tin direction 
le last figure 
itenna coordi- 
lin function 
J'Jor ::: 
i 
J 
3». 
integrated 
.+ 30 deg.). 
SCAN 665 0 = 63° 
Figure 4. Calculated normalized radar cross section. 
FIELD : 
Figure 6. Along-track field-averaged gamma. 
4 SIR-B EXPERIMENT 
As already told in the introduction unfortunately the 
SIR-B experiment has been partly insuccesful. Orig 
inally, the DUTSCAT measurements should take place at 
5 days within a 10 days flight scheme of the Shuttle 
in early September 1984. However, launching had to be 
delayed twice and the program reduced to 2 days, 11 
and 12 October 1984. On these days the collection of 
ground data of the test area occured under rather wet 
field conditions due to extreme high precipitation in 
the preceding month af September. The Shuttle radar 
failed to produce data for our test area but scat- 
terometer data were gathered along with flight par 
ameters and video recordings. The last are meant to 
facilitate the interpretation of the measurements. 
The test area in the Flevopolder consists of large 
agricultural fields (75 ha). We see it in fig. 5. 
The lengths of the straight lines in this figure are 
proportional to the number of scans where the half 
power azimuthal antenna beamwidth is completely within 
the field. So field-averaging is depending upon 
flightparameters e.g. altitude, heading, pitch, roll, 
as well as incidence angle and of course fieldsize and 
beamwidth. With a videocamera attached to the scat- 
terometer antenna recordings are made containing time 
information from which the position of fieldboundaries 
in the radardata can be extracted independent of the 
measurement geometry. To obtain the earlier discussed 
accuracy of 1 dB in relation with equation (5) to (7) 
more than 10 scans must be available for averaging 
within one field. 
Figure 7 gives an impression of the multi-angular 
gamma values for the 19 different fields. We see that 
the data follow an S-like curve within a range of 
10 dB. 
FIELD 1 ...19 
Figure 7. Gamma versus incidence angle. 
The homogeneity of these fields in terms of flatness, 
soiltype and agricultural practise is unique. For an 
impression of the area the reader is referred to the 
literature (de Loor 1982). 
The flight track is located diagonally over the test 
area and has a length of 10 km. At the time of the 
measurements 19 different fields were covered. These 
fields were surveyed and sampled in detail which 
resulted in data about soil moisture, surface rough 
ness and vegetation (Stroosnijder 1984). 
The radar cross section 0 o of (vegetation covered) 
soils depends on a limited number of predominant 
surface parameters, such as those just mentioned. The 
relative importance of these parameters depends on 
radar parameters such as incidence angle, wavelength 
and polarization. The objective for the experiment for 
which data is available was to produce estimates of 
soil moisture and roughness by using an inverse scat 
tering model and field averaged O values for various 
incidence angles. 
An example of along-track field averaged a divided 
by the cosine of the incidence angle, or the backscat- 
tering coefficient y is given in fig. 6 (<0>=63 ). 
5 SOIL MOISTURE AND ROUGHNESS 
As discussed in 2.1 the across-track ground resolution 
degrades for decreasing incidence angles. Because of 
this resolution problem we will restrict ourselves for 
the moment to large incidence angles. For this case 
the backscattering coefficient y can be written as 
(Attema e.a. 1982). 
y = F.(l-exp(-p 2 )) (g) 
p = 2kacos0, k=2TT/A 
with a the rms value of the height distribution and 
F an unknown intensity factor. Based on experimental 
evidence, especially for bare soil, it can be assumed 
that F is independent of incidence angle and related 
to the soil moisture content (Attema/Krul 1979). From 
eq. (9) we see that many moisture content - roughness 
combinations are possible at one y value and incidence 
angle, so that hardly any discrimination can be at 
tributed to one single y value. Even if a significant 
part of y versus incidence angle is known it might not 
be possible to distinguish roughness from moisture.