et al. 2005) in westcentral and southwest NB; i.e., Nashwaak Lake (NWL, 46
28
0
20
00 N, 67
06
0 00
00 W) and Charlie Lake sites (CL, 45
53
0 05
00 N, 67
21
0 25
00 W;
Fig. 15.1), (ii) three locations in sparse forests at the Canadian Forces Base
Gagetown, central NB (CFB Gagetown, 46
40
0 N, 66
22
00 W), and (iii) one
location in a bare field at CFB Gagetown. A second evaluation of the method was
conducted by comparing WI with SWC-values generated with a widely used
hydrological model, the Soil Water Assessment Tool, SWAT (Arnold et al. 1998)
for an area in the Black Brook Watershed in northwest NB (47
07
0 N, 67
46
0 W).
Simulated SWC-values were descriptive of averaged soil water conditions within a
1-m soil block, in field conditions typical to the potato-growing area of
northwest NB.
15.3 Methods
Figure 15.2 shows a schematic of workflow associated with deriving the new
WI. The method is divided into three main procedural components, namely (i)
the pre-processing of the RADARSAT-1 data, including extracting radar brightness
values, filtering, co-registration, geo-coding, and creating a multi-temporal image
sequence, (ii) deriving WI-values and associated images, and (iii) conducting
comparisons between radar-based values of WI and estimates of SWC obtained
in the field and with the ArcView™ GIS-version of the SWAT model (i.e.,
AVSWAT_2000; Luzio et al. 2002).
15.3.1 Image Pre-processing
Three calibrated, geo-referenced RADARSAT-1 ScanSAR images provided by the
Canadian Space Agency (CSA) were used to extract radar brightness (β
0 ) in dB
values following steps described in ALTRIX Systems (2000); β
0 is a measure of
radar reflectivity in the slant range and independent of local incident angle. Instead
of using the commonly-used radar backscattering coefficient (i.e., σ
0 ; a measure of
radar reflectivity near ground range and dependent on local incident angle), we
opted to use β
0 in this research. It is well recognized that β
0 is highly influenced by
background features (Bamler 2000), such as surface roughness, local incidence
angle, effective surface cover density and surface-scattering properties (e.g., biomass, leaf density), three-dimensional structure of the scattering surface (canopy
layering, trunk placement, buildings), and dielectric constant of the scattering
material. In general, background conditions vary with variations in SWC and the
physiological status of the surface, particularly with respect to the vegetation cover.
Since we intended to use images obtained during similar flight trajectories and
signal polarization, the use of β
0 for estimating temporal changes in SWC by
tracking changes in β
0 was possible. Note that the resolution of the β
0 images had
15 Development of a New Wetness Index Based on RADARSAT-1 ScanSAR Data
305
28
0
20
00 N, 67
06
0 00
00 W) and Charlie Lake sites (CL, 45
53
0 05
00 N, 67
21
0 25
00 W;
Fig. 15.1), (ii) three locations in sparse forests at the Canadian Forces Base
Gagetown, central NB (CFB Gagetown, 46
40
0 N, 66
22
00 W), and (iii) one
location in a bare field at CFB Gagetown. A second evaluation of the method was
conducted by comparing WI with SWC-values generated with a widely used
hydrological model, the Soil Water Assessment Tool, SWAT (Arnold et al. 1998)
for an area in the Black Brook Watershed in northwest NB (47
07
0 N, 67
46
0 W).
Simulated SWC-values were descriptive of averaged soil water conditions within a
1-m soil block, in field conditions typical to the potato-growing area of
northwest NB.
15.3 Methods
Figure 15.2 shows a schematic of workflow associated with deriving the new
WI. The method is divided into three main procedural components, namely (i)
the pre-processing of the RADARSAT-1 data, including extracting radar brightness
values, filtering, co-registration, geo-coding, and creating a multi-temporal image
sequence, (ii) deriving WI-values and associated images, and (iii) conducting
comparisons between radar-based values of WI and estimates of SWC obtained
in the field and with the ArcView™ GIS-version of the SWAT model (i.e.,
AVSWAT_2000; Luzio et al. 2002).
15.3.1 Image Pre-processing
Three calibrated, geo-referenced RADARSAT-1 ScanSAR images provided by the
Canadian Space Agency (CSA) were used to extract radar brightness (β
0 ) in dB
values following steps described in ALTRIX Systems (2000); β
0 is a measure of
radar reflectivity in the slant range and independent of local incident angle. Instead
of using the commonly-used radar backscattering coefficient (i.e., σ
0 ; a measure of
radar reflectivity near ground range and dependent on local incident angle), we
opted to use β
0 in this research. It is well recognized that β
0 is highly influenced by
background features (Bamler 2000), such as surface roughness, local incidence
angle, effective surface cover density and surface-scattering properties (e.g., biomass, leaf density), three-dimensional structure of the scattering surface (canopy
layering, trunk placement, buildings), and dielectric constant of the scattering
material. In general, background conditions vary with variations in SWC and the
physiological status of the surface, particularly with respect to the vegetation cover.
Since we intended to use images obtained during similar flight trajectories and
signal polarization, the use of β
0 for estimating temporal changes in SWC by
tracking changes in β
0 was possible. Note that the resolution of the β
0 images had
15 Development of a New Wetness Index Based on RADARSAT-1 ScanSAR Data
305