Geosalar: Innovative Remote Sensing Methods for Spatially Continuous Mapping of Fluvial Habitat at Riverscape Scale - Fluvial Remote Sensing for Science and Management - page 189

on a logical statistical design. They therefore stressed the
importance of describing fluvial habitats in a continuous
manner in order to detect unique habitats (e.g. local cool
water input) or disturbance events (e.g. barrier to move-
ments) at specific locations that can affect the distribution
and abundance of fish over the entire riverscape.
However, they acknowledged that implementing their
approach was a challenge due to the lack of appropriate
technology to work at the intermediate scale (10 3

channel is not altered by tributaries and therefore the sus-
pended sediment load can be assumed as constant for the
whole channel. The SMR supports Atlantic salmon and
brook trout ( Salvelinus fontinalis ) populations that have
been the subject of numerous ecological studies by CIRSA
researchers and their students since the creation of the
research group in 1995. Field work for the various com-
ponents of the project was facilitated by CIRSA's research
station located downstream from the confluence between
the main stem and North-East (1100 km 2 ) branches of
the SMR.
In August 2002, during the period of summer low
flow, the XEOS TM imaging system developed by Genivar
Inc. was fitted to a helicopter and used to obtain plan
view digital high resolution optical images covering the
entire 80 kilometres of the main channel of the Principale
branch of the SMR. A first survey conducted at a constant
altitude of 155m above ground resulted in a dataset
comprising of c. 5550 standard colour images with a
spatial resolution of 3 cm. A second survey conducted at
an elevation of 450m above ground generated another
set of 1600 colour images with a spatial resolution of
10 cm. Image format was 3008 pixels × 1960 pixels in the
standard visible bands of red, green and blue. Images were
collected at 60% overlap. An onboard GPS provided the
position of the centre point of each image. These surveys
provided one of the first large scale hyperspatial (defined
in Chapter 8) image datasets reported in the river science
literature. The project also supported a number of field
efforts which provided a range of data concerning the
abundance and spatial distribution of Atlantic salmon at
various life stages.

10 5 m)
encompassing all necessary habitats (Figure 9.1). On the
one hand, traditional field-based methods offer good
ground resolution of fluvial habitat variables at themicro-
habitat scale but they are labour intensive and not well
suited to the continuous characterisation of long river
segments. On the other hand, satellite-based imagery
offer a large-scale synoptic description of entire fluvial
systems but their ground resolution is currently not suf-
ficient for fine-scale habitat modelling purposes. One of
the main focuses of the Geosalar project was therefore to
fill the gap between these approaches by developing a new
set of remote sensing methods allowing the production
of high-resolution spatially continuous maps of fluvial
habitat variables over long river segments. Typically, four
variables are used to describe the physical habitat of
riverine fishes: bed material grain size, water depth, flow
velocity and water temperature. The emphasis of the
Geosalar research effort was put on the quantification of
bed material grain size and water depth. Post-Geosalar
developments later addressed the quantification of flow
velocity and water temperature. This chapter presents
the innovative remote sensing methods that were devel-
oped during the Geosalar project for the quantification
of river habitat variables over long spatially continuous
river segments. The usefulness of these methods is then
exemplified by applying them to the case of the Sainte-
Marguerite River (Quebec, Canada) for the analysis of
Atlantic salmon juvenile and adult habitat.

−

9.3 Grain size mapping

Bed material grain size is one of the most fundamental
descriptors of salmonid habitat. Generally speaking, the
freshwater stage of the salmon's life cycle requires coarse
substrate (Armstrong et al., 2003). The presence of clay,
silt and sand is well established as having a negative
impact on the survival of eggs (Sear, 1993) and juveniles
(a group name for alevins, fry and parr). Consequently
the quality of the substrate is often considered as a pri-
mary indicator of the health of a salmon river. However,
assessing the substrate status for an entire river has always
been problematic. Certain methods, such as the River
Habitat Survey (RHS) protocol developed in the UK
(Raven et al., 1997), employ walking surveys and visual
appraisals in order to get a semi-quantitative sampling

9.2 Study area and data collection

The Geosalar research was conducted on the Principale
branch of the Sainte-Marguerite River (SMR) (48 ◦ 27 N,
69 ◦ 95 W), a gravel-cobble bed river draining a Cana-
dian Shield catchment of approximately 1000 km 2 in the
Saguenay region. Bed material is composed of well mixed
igneous and metamorphic rocks. The lithological com-
position of this mixture is stable along the channel length
and thus no spatially dependent clast colour variations
can be observed. Suspended sediment load along the