184 
(4) 
OQ 
For 
“O 
the 
I 
□ 
UJ 
ID 
Z 
< 
CL 
1 
From eq. (2) and (3) the resolution cell follows as: 
Aa = Ax.Ay = cxB a h/sin20 
Which obviously has a minimum at 0 = 45 degre 
example at a flight altitude of 300 m as used 
SIR-B experiment we find that for DUTSCAT: 
T = 100 ns 
0a =0.22 rad 
^^min = ^980 m^ 
In the search for relations between surface proper 
ties and radar backscatter characteristics it must be 
noted that the irregular nature of surfaces in general 
causes the scattered electromagnetic field components 
to be random functions. This means that the scattered 
power as measured by the radar is different for each 
resolution cell as it depends on phase relations with 
in the resolution cell itself. Therefore the estimated 
average of this scattered power, or the radar cross 
section per unit area is the most commonly used para 
meter in radar studies. After an incoherent averaging 
known as "speckle reduction" the standard deviation 
of the average received power P is: 
SCAN 665 0 = 63° 
a = p//n 
(5) 
Figure 2. Averaged received power for scan 665 with 
an incidence angle of 63 degrees. 
3 DATA PROCESSING 
From the NLR we receive preprocessed data of both the 
scatterometer and the airplane's inertial naviagion 
system. These two types of data are time tagged so 
that they can be combined and in the mean time correc 
ted e.g. for the spatial spreading loss and the an 
tenna weighing. This so called radiometric correction 
is expressed by the radar equation: 
with N the number of independent measurements caused 
e.g. by the movement of the airplane. For N we find 
the next expression (Ulaby e.a. 1982): 
,3 4 
P (4tt) R 
r 
(Ay.G./g 2 ) 1 
(8) 
N = Ax-(L a /2) 
(6) 
With L a the antennalength which is approximately: 
L 
a 
A/6 a 
(7) 
with A the free space wavelength which is equal to 
0.25 m for the frequency of 1.2 GHz used by DUTSCAT. 
From (6) and (7) we see that an independent measure 
ment is done after each movement of half the antenna 
length. The received power appears to be exponentially 
distributed and it can be calculated that for N=200 
the averaged power is known with an uncertainty of 1 
decibel within 90% of the time (Smit 1978). The along- 
track distance required in our case to obtain this 
accuracy is 120 m or 1.3 along-track resolution cells 
at an incidence angle of 45 degrees (h=300m). 
2.2 DUTSCAT data 
The DUTSCAT scatterometer has a rather high internal 
pulse repetition frequency of 78.125 kHz so that every 
12.8 ys a pulse is transmitted. The received signal 
is digitized with a sampling rate of 20 MHz which 
means that the sample time is equal to 50 ns. Then a 
coherent averaging is applied to the signals of sub 
sequent pulses to improve the signal to noise ratio. 
After this the previously described speckle reduction 
is performed by averaging incoherently. Additional 
averaging is done to reduce the data rate of the 
digital output. Finally we have a pulse repetition 
frequency of 4.77 Hz, so that effectively the across- 
track direction is scanned once every 210 ms. 
An example of an arbitrary resulting scan is given 
in fig. 2. From the velocity of the airplane, that 
carries the scatterometer, which is known to be about 
50 m/s it directly follows that this particular scan 
(665) is taken at about 7 km after the start of the 
flight track. The information that at the time of 
this scan the incidence angle was 63 degrees is ex 
tracted from flight parameter data that simultaneously 
has been recorded on tape along with the radar data. 
The nominal incidence angle can be selected by the 
operator inside the airplane and varied between 10 
and 80 degrees. During the SIR-B experiment the flight 
track was flown 8 times for one series of measurements 
at intervals of 10 degrees. This was done on two 
days, and in total three times so 24 recordings were 
made. 
P : received power (averaged) 
P : transmitted power 
G ^ : maximum antenna gain 
Jg : Jg 2 (x,y)dx or R.Jg 2 (a,£)da 
g : antenna gain function 
a,£ : azimuth & elevation angle (antenna coordi 
nates) 
Not explicitely written out in this formula is the 
fact that the resulting 0 O or the radar cross section 
per unit area depends on the sample number to which 
it belongs. Variations of the transmitted power P-t 
can be monitored with the aid of the system's inter 
nal calibration. The antennagain G in the direction 
of maximum radiation has been measured. For the 
DUTSCAT antenna (a 0.9 m parabolic dish) a value was 
found of 17 dB. As another part of the external cali 
bration the antenna gain function has been measured 
over 60 degrees. According to eq. (8) this function 
has to be squared and integrated with respect to the 
azimuth angle (along-track or x-direction). The re 
sult (see fig. 3) is a function of the elevation 
angle in antenna coordinates. For the example given 
in fig. 2 the calculated radar cross section 0 Q is 
presented in fig. 4 where of course the most accurate 
values are expected to be around the main direction 
of the antenna. The vertical line in the last figure 
corresponds to an elevation angle in antenna coordi 
nates of -30 degrees below which the gain function 
has not been measured. 
F* 1.2 GHz 
Figure 3. Squared antenna gain function integrated 
with respect to the azimuth angle (-30..+ 30 deg.). 
1«. 
». 
-n. 
-2«. 
-3«. 
5«. 
Figure 4. i 
4 SIR-B ex: 
As already 
SIR-B expe: 
inally, thi 
5 days wit! 
in early Se 
delayed tw: 
and 12 Octi 
ground dati 
field cond: 
the preced: 
failed to ] 
terometer ( 
ameters anc 
facilitate 
The test 
agricultur; 
Figure 5. 1 
The homoger 
soiltype ar 
impression 
literature 
The fligl 
area and he 
measurement 
fields were 
resulted ir 
ness and ve 
The radai 
soils deper 
surface pai 
relative in 
radar paran 
and polari2 
which dat 
soil moistr 
tering mode 
incidence a 
An exampl 
by the cosi 
tering coef