In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
2. STUDY SITE AND DATA COLLECTION 
Two sites were selected near Ottawa, Ontario, Canada for field 
and satellite data collection, the Canadian Food Inspection 
Agency (CFIA) research farm and a region of private producers 
east of Casselman, Ontario. The terrain across these two study 
sites can be considered level to very gently sloping (<2%) with 
an average field size of 23 hectares. This region of eastern 
Canada consists largely of com and soybean annual crop 
production. 
Ground truth measurements were performed on several selected 
sites. Within the CFIA site, 13 fields including 16 com and 21 
soybean sample sites were selected. At the Casselman site, 20 
fields including 10 com and 10 soybean sample sites were 
visited. Total LAI was measured at each sample site using an 
LAI-2000 (Li-Cor, Inc., Lincoln, NE) plant canopy analyser 
throughout the growing season. Volumetric surface soil 
moisture was measured coincident with each SAR acquisition, 
using Delta-T Theta probes with 6-cm waveguides. At each 
site, mean soil moisture was calculated from ten replicate 
moisture measurements. 
SAR images were acquired by TerraSAR-X, PALSAR/ALOS, 
and RADARSAT-2 satellites. During the 2008 growing season, 
two RADARS AT-2 Fine beam mode quad-pol images (July 6 
and 9) and one PALSAR/ALOS (July 2) were acquired over the 
CFIA site; two RADARS AT-2 Fine beam mode quad-pol 
images (August 23 and August 26) and one TerraS AR-X 
stripmap image (August 21) were acquired over the Casselman 
site. The pixel spacing of TerraSAR-X, PALSAR/ALOS, and 
RADARSAT-2 is 3 m, 12.5 m and 8 m, respectively. 
Characteristics of the SAR images used in this study are 
summarized in table 1. 
Date 
SAR sensor 
mode 
poi. 
incident 
angle 
07-02- 
2008 
PALSAR 
LI.5 
HH,HV 
34° 
05-23- 
2007 
PALSAR 
LI.5 
HH,HV 
21° 
07-09- 
2008 
Radarsat-2 
FQ20 
Quad- 
pol 
o 
O 
07-06- 
2008 
Radarsat-2 
FQ6 
Quad- 
pol 
26° 
08-21- 
2008 
TerraSAR-X 
stripmap 
HV/VV 
44° 
Table l.Main characteristics of SAR images used in this study. 
Optical images were acquired by the Compact Airborne 
Spectragrahic Imager (CASI) and SPOT-4 multi-spectral 
satellite. CASI hyperspectral data were acquired on August 21, 
2008 over the Casselman site. SPOT-4 multi-spectral data were 
acquired on July 6, 2008 over the CFIA site. From the CASI 
and SPOT-4 data, the Modified Triangular Vegetation Index 
(MTVI2) (Haboudane et al., 2004) was calculated. A non-linear 
curve fitting procedure was used to establish an empirical 
equation for LAI estimation from MTVI2 (Liu et al., 2009): 
LAI = -6.247 x ln(0.946 - 0.643 x MTVI2) 
Using this formula, LAI maps for the entire study site were 
generated from the optical data, with an RMSE of 0.76 and an 
R 2 of 0.85. 
3. METHODOLOGY 
3.1 Image processing 
Radiometric calibration of TerraSAR-X and PALSAR/ALOS 
images was carried out using the follow equations. These 
equations were used to convert the digital number of each pixel 
DNi into a backscatter coefficient ( a ). 
For TerraSAR-X, 
cr° (dB) = 201og 10 DN. + 101og l0 (CalFact) + lOlog lo (sin(0 ( .)) 
CalFactor is given in thè TerraSAR-X data delivery package 
annotation file. It is processor and product type dependent. 
For PALSAR/ALOS, 
of (dB) = 101o glo (DN? ) 4- CF 
The calibration factor (CF) for PALSAR L1.5 products is -83 
dB. The ALOS and TerraSAR data products were delivered in 
ground range. A 3 X 3 Enhance Lee filter was applied to both 
the ALOS and TerraSAR data to reduce speckle noise. 
RADARSAT-2 fine quad-pol SLC data were provided as 
compressed stokes matrix values for each slant range pixel. 
Prior to extracting the backscatter information, a boxcar filter 
with a 5 by 5 kernel size was applied to the polarimetric SAR 
scattering matrix data to suppress SAR speckle. After filtering 
the covariance matrix was converted to a symmetrized 
covariance matrix. From the symmetrized 3 by 3 covariance 
matrix, intensity backscatter (HH/HV/W) was extracted. All 
the data were then geometrically corrected and geo-referenced 
using national road network vector data. 
3.2 Water cloud model 
The water cloud model was introduced first by Attema and 
Ulaby (1978). In the general version of the water cloud model, 
the power backscattered by the whole canopy (cr°) can be 
represented as the incoherent sum of contributions of the 
vegetation, ( a o ) ; and the underlying soil, ( a 0 ). This study 
veg soi 
selected the model modified by (Prévôt et al., 1993) as it 
incorporates LAI as a descriptor of vegetation development. In 
this model, SAR backscatter from a canopy at a given incidence 
angle can be written as: 
cr° = AL e cos é?(l - exp(-2BL t cos 9)) + o° oi¡ exp(-2BL / cos(0) 
where x is the two-way attenuation through the canopy layer, L 
is the LAI, expressed in (m 2 m' 2 ) , the backscatter coefficients 
a 0 , and are expressed in power units. A,B, C,D and E are 
model coefficients to be defined by experimental data. A, B and