Geography Reference
In-Depth Information
L1
L2
L3
L4
L5
L6
L7
L8
30
20
10
0
(a)
40
30
20
10
0
80
70
60
50
40
30
Distance upstream (km)
(b)
Figure 9.8
Illustration of the spatial correspondence between the sedimentary link structure of the Sainte-Marguerite River and
a) the distribution of Atlantic salmon parr densities (Data extracted from Bouchard and Boisclair (2008)) b) the centroıd of
spawning sites (blue stars), on each sedimentary link. Adapted from Davey and Lapointe, 2007.
project. The key issue was found to relate to the number
of images in hyperspatial data sets and to the stability of
the lighting conditions. In the case of the hyperspatial
image dataset for the Ste-Marguerite river, the acqui-
sition of 5550 images over 80 kilometres meant that
lighting conditions slightly changed over the duration of
the flight (see the discussionon radiometric normalisation
in Chapter 8). This led to a challenging condition where it
was impossible to calibrate the depthmapping process for
all 5550 images. For example, Figure 9.9 shows calibra-
tion and validation relationships of depth versus image
brightness in the case of raw imagery where radiometric
normalisation is a significant issue. These relationships,
fromCarbonneau et al. (2006), were created fromground
points which span three separate images. An effort was
made to select an area which was thought to be repre-
sentative of the river. This area contained a mid-channel
bar with a typical fining structure going from cobbles
to sparse sandy patches. The bathymetry of the area was
9.4 Bathymetry mapping
Juvenile salmon also express a strong preference for rela-
tively shallow flows and therefore water depth is another
important habitat descriptor (Armstrong et al., 2003).
However, in this case, there is a wealth of literature on the
subject and much detailed literature (see Chapter 3 and
(Legleiter et al., 2009; Legleiter et al., 2011). One classic
approach is to establish an empirical calibration between
geolocated depth measurements and image brightness
values according to the method proposed by Lyzenga
(1978). In this part of the Geosalar project, a Real-Time
Kinematic (RTK) GPS was used to efficiently collect
over 1000 geolocated depth measurements. However,
this empirical approach was initially developed for larger
water bodies sampledwith coarse resolution imagery. As a
result, its application to hyperspatial image data sets posed
certain specific problems whichwere addressed during the
Search WWH ::
Custom Search