Geography Reference
In-Depth Information
characteristics of the studied object. The analysis of
backscattering coefficient permits the identification and
characterisation of different types of land cover, such as
wetlands or agricultural practices (Touzi et al., 2007).
Almost no specific studies have been realised on
narrow strips of riparian vegetation because of the
complexity of this geographical object and the available
SAR data. Some studies of floodplain environments
show that inundation and vegetation structure can be
accurately characterised with radar data (Townsend,
2001; Townsend, 2002). Nevertheless, the latest radar
satellite, which acquires polarimetric data in a high
or very high spatial resolution, appears promising for
small riparian corridor characterisation as it brings
complementary information to classical data used for
riparian studies (optical data, LiDAR, ground control
campaign, soil maps etc). For example, with Radarsat-2
data, different types of scattering of the riparian
vegetation can be extracted, bio physical parameters such
as moisture and biomass can be retrieved and combined
with other information sources, permitting a finer study
and monitoring of riparian vegetation. Figure 10.10
presents two composite colours of a Radarsat-2 image
(mode fine Quad-Pol; 10/02/2011) on the Yar watershed
in Brittany (France). On the first colour composite
(Figure 10.10a) which represents the entire scene, the
colour red is associated with HH polarisation, the
colour green with the HV polarisation and the colour
blue with the VV polarisation. At this scale, we can
distinguish quite precisely the vegetation (green) and
bare soil (magenta) surfaces. In order to identify riparian
vegetation we zoomed on the north of the Yar watershed
(Figure 10.10b). We applied the Freeman decomposition
(Freeman et al., 1998), which allows the extraction of
the power contributions due to rough surface (Red),
volume scattering (Green) and double-bounce (Blue).
Bare soils are represented in red and magenta because
of the high power contribution of the rough surface.
Riparian vegetation, characterised by volume scattering,
appears here in green and is clearly identified. Finally,
urban surface appears on blue because of a main
contribution of the double-bounce scattering. With
this type of data compounded with four polarisations,
different polarimetric decomposition can be applied
(Cloude and Pottier, 1997; Touzi et al., 2007, Freeman
et al., 2007
Inverse model can also be applied to retrieve soil or
vegetation moisture, forest biomass etc (Beaudoin, 1994).
10.6 Perspectives: from images to
indicators, automatised and
standardised processes
The efficiency of image use is linked to our capacity
to provide a standard that can be specific to a given
question, but also simple and relevant. There is a large
panel of applied uses ranging from simple and general
indicators (e.g. corridor width or vegetation continuity)
to very complex or specific needs such as senescence
characterisation (Lonard et al., 2000) or identification of
potential restoration reaches (Mollot and Bilby, 2008).
The efficiency is also linked to our ability to merge sensor
data produced with field and ancillary data (Munne et al.,
2003; Gonz alez del T anago and Garcıa de Jal on, 2006;
Debruxelles et al., 2009). For river and forest managers
the questions we need to address are : i) 'which indica-
tors should we choose or define?' and, ii) 'how do we
extract them?'. The first question needs an active dia-
logue between scientists (from several disciplines) and
managers. To answer the second, we can use GIS and
remote sensing tools that have been recently developed
and can provide regularly spaced information for the
assessment of riparian conditions (Tiner, 2004; Claggett
et al., 2010; Wiederkehr et al., 2010; Tormos et al., 2010;
Alber and Piegay, 2011). Indeed, besides all that the tech-
nical possibilities offer by remote sensing, image uses
for riparian area understanding and management require
some progress, notably with respect to automation and
standardisation of processes and indicator calibration.
They are very important both for river network managers
in charge of national and regional planning (upscaling
analysis from reach to network scale) and for technical
transfer to reach-scale managers. For example, the com-
bination of LiDAR data and satellite images has been
shown as a very interesting approach at reach scale, but to
be able to pass from one image to a set of images and then
to the overall network, it needs specific tools in terms
of data fusion, calibration and management (Hall et al.,
2009; Arroyo et al., 2010). At the reach scale, there is a
large gap between tools used or under development by the
scientific community and standard approaches accessible
for managers (e.g. historical analysis of aerial photos).
Another question is thus how can managers have access
) and permit a good discrimination of
geometrical parameters of the studied object such as the
type of scattering (surface, volume or double bounce).
...
Search WWH ::
Custom Search