previous 
| satellite 
sponding 
vest time 
m’s Barn 
caf Area 
e in good 
standard 
rved and 
10delling, 
‘RS SAR 
ved using 
and ERS- 
ugar beet 
This was 
1 into the 
Barn in 
collected 
n ERS-1 
d weekly 
Systems, 
and Red 
e for 30° 
urements 
ce to the 
um 12 m 
ophysical 
ger set of 
of image 
The SAR precision images are multi-look, ground range, digital 
images generated from raw SAR image mode data using up-to- 
date auxiliary parameters and corrected for antenna elevation 
gain and range spreading loss. A full scene of ERS-1 and ERS- 
2 SAR normally has a coverage of 100 km in ground range and 
at least 102.5 km in azimuth (along track). The Ascending and 
Descending ERS-1/2 SAR overpasses for East Anglia are at 
approximately 2200 GMT and 1053 GMT, respectively. Six 
ERS-1 images for 1995 and seven ERS-1 and three ERS-2 SAR 
scenes for 1996 were used for this study. 
4. ERS-1 AND ERS-2 SAR CALIBRATION 
The calibration constant was given as 59.96 dB for ERS-1 and 
55.61 dB for ERS-2 the mid-swath position (Bird, 1997). The 
methodology defined by Laur et al. (1997) to derive 
backscatter coefficients in ERS-1 and ERS-2 SAR products 
was followed as the initial calibration precision procedure. 
Therefore, to calculate the radar backscatter coefficient o? (dB) 
for ERS-2, the following equation was used : 
o? (dB)-10log;o(«I»)»-10log;o(sin 0/sin23?)-55.61 dB (1) 
where <I> is the mean pixel intensity and « is the local 
incidence angle. 
The two sugar beet fields where the ground data collection. 
campaign took place were identified on the ERS-1 and ERS-2 
SAR images. An area for each field was defined on the 
ERDAS IMAGINE image processing system together with a 
report on the number of pixels in the selected area, the mean 
value of pixel amplitude and the standard deviation. The 
average radar intensity for each field was calculated as the sum 
of the square of mean and the square of standard deviation. 
The local incidence angle for each field was computed for the 
centre pixel column of the field. These values were substituted 
into equation (1) to calculate the backscatter coefficient. 
5. WATER CLOUD RADAR MODEL 
In the original form of the Water cloud radar model, the radar 
cross section per unit area, was expressed in power units 
(m?/m?) to reflect the physical relationships between the radar 
backscatter coefficients and the target parameters (Attema and 
Ülaby, 1978). 
Xu et al. (1996) modified a version of the cloud model which 
has been derived Leeuwen and Clevers’ (1994). This was used 
for estimation of LAI using measured values of o? and values 
of soil moisture content: 
  
_cos0 | o ( mi / m? )- A. cos0 
In 1 
2B C'(1-- D'. m;)- A. cos0 
(2) 
L= 
Where, A=0.3259, B=0.167, C'20.0452 and D” = 0.0603 
A, B and C' were coefficients fitted in a previous experiment 
on nine fields in 1994 (Xu et al, 1996) and D' was 
predetermined from published values on soil backscatter 
(Wooding et al., 1993). 
IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India,2002 
   
6. CANOPY COVER ESTIMATION 
The canopy cover fraction or the fraction of intercepted solar 
radiation was estimated based on Beer’s law (Monsi and Saeki, 
1953) using following formula (Steven et al., 1986): 
f=1-exp (-K.L) (3) 
Where f is the fraction of incident solar irradiance intercepted 
or absorbed by leaves and L is the leaf area index. K is 
normally a constant for a given crop type, (K=0.7 for sugar 
beet, Scott and Jaggard (1992)). 
7. YIELD MODEL 
The yield model, which provides a forecast of harvestable 
sugar per hectare, is represented by the equation (Werker and 
Jaggard, 1997): 
Y = k * [s (S. f. e. p)dt (4) 
Í sowing 
Where: 
k = a correction factor to account for losses 
S = solar radiation 
f = fraction of radiation intercepted by crop 
e = conversion coefficient (to dry matter) 
p = partitioning coefficient 
The way in which crop cover (f) varies with time for sugar beet 
is modelled by Werker and Jaggard (1997) using a 
parameterised profile of crop cover which is initially defined in 
terms of default parameters which are driven by thermal time. 
The crop cover values obtained from a sequence of satellite 
observations taken through the growing season are utilised by 
adjusting the f-profile to fit these values. The adjusted profile is 
then used as input to the sugar beet yield model (equation 4), 
along with weekly meteorological data (used up to the current 
date), climate data (used to estimate future weather up until 
harvest) and sowing dates. 
8. RESULTS 
8.1 1995 experiment 
8.1.1 Crop cover estimation by LAI using ERS-1 SAR for 
two beet fields 
LAI was calculated for each field on the 6 dates by equation (2) 
using soil moisture. Correlation coefficients of 0.69 and 0.63 
were found between the measured and predicted LAI for field 
A and B, respectively. Brown et al. (1992) and Bouman and 
Hoekman (1993) observed that radar backscatter of sugar beet 
was often attributed to leaf area and water content. Estimates of 
f from the radar data with the conversion from LAI as in 
equation 3 for the Xu et al.(1996) model. Correlation 
coefficients of 0.65 and 0.64 were found for field A and B, 
respectively between predicted values with crop cover 
determined from the leaf area measurements. 
Meadows and Wright (1994) observed that for a 5 ha field, 
speckle can affect radar backscatter up to 0.27 dB (6% error). A 
Lee-sigma filter improved the appearance of the image and 
made field identification easy in this study, but had a non- 
significant effect on crop cover estimation.