Geography Reference
In-Depth Information
thus a useful measure of lateral channel activity. Because
elongation rate estimates are less sensitive to photograph
alignment error than are simple lateral shifting measure-
ments, they can be used to estimate reach-average lateral
channel activity even when image rectification quality
is poor. In Figure 11.4b, the reaches with data of low-
est quality are plotted using open diamonds, while the
other reaches used in the regression are plotted using
filled diamonds. The slope of the regression line fit to
the high-quality data only gives E =
the methods were originally developed for orthorectified
images, many of the techniques described here, par-
ticularly those applicable to a single image set, could in
principle be applied to poorly rectified images. Additional
value is of course obtained when image rectification is
sufficiently accurate to allow for temporal change to be
detected.
In this approach, Unitary Geographical Objects (UGO)
are defined as linear and polygonal spatial units that locate
and delineate in space biophysical components of fluvial
systems. Their degree of schematisation depends upon the
spatial resolution of the raw data and of the method used
for extraction. UGO are systematically disaggregated into
elementary spatial units, so-called Disaggregated Geo-
graphical Objects (DGO), to provide linear referencing
systems for attributes along stream networks that allow
for characterisation of UGO continuously from upstream
to downstream (Figure 11.6). DGO usually correspond
to resolution-driven spatial units with a uniform space
step. No clear rules dealing with the disaggregation pro-
cesses have been established. It has been performed by
many authors pragmatically considering segment lengths
varying between 10 m and 1 km long depending on the
channel size and the reach length. In any case, the spa-
tial resolution should be higher than the characteristic
size of the object of interest. Radiometric or planimetric
information can be extracted at the scale of each DGO,
and then statistical tests can be performed on the result-
ing longitudinal series in order to delineate biophysically
meaningful spatial units, the so-called Aggregated Geo-
graphical Object (AGO) (Figure 11.6). AGOs result from
spatial aggregation of DGOs that have been longitudi-
nally differentiated into homogeneous groups. Spatial
aggregation is based on statistical analysis of distributions
generated by one or several attributes from the linear ref-
erenced DGOs. Several statistical techniques can be used
depending on the number of attributes and the character-
istics of their distribution (e.g., Heterogeneity tests such
as Pettitt or Hubert methods, Contrast Enhancing, Spa-
tial constraint clustering, or Hidden Markov Model ... )
(see Leviandier et al., 2011). In this way, identification of
AGO is objective and follows a logic of self-emergence
(Parsons et al., 2004). AGO boundaries thus are likely to
represent process boundaries as long as operators have
selected classification criteria that are morphologically
and/or functionally meaningful.
Such strategies offer new perspectives for characteris-
ing fluvial systems due to the generalisation of methods
to large scale, such as planform measurement. The pro-
cedure for detecting change-points in the longitudinal
0 . 19 at the
95% confidence level. Reaches that plot above the line
are likely either migrating slowly enough that photograph
alignment error causes an upward bias in lateral activity,
or contain an anomalously large number of downstream
translating bends.
In Figure 11.5 we present elongation rates and lat-
eral migration rates computed by simple point-by-point
lateral distance measurements for four separate sets of
photograph pairs along two different river systems. All
of the reaches illustrated in Figure 11.5 show some
combination of elongation and translation, so their reach-
average elongation rates are probably representative for a
broad set of single-thread meandering river floodplains.
For reaches not experiencing channel cutoff, elongation
clearly depends bot h o n the reach average absolute chan-
nel migration rate
5 . 70
±
and on total reach length (a long
channel will elongate more than a short one, everything
else being equal). As a method of characterising mor-
phodynamic activity, they offer an added advantage over
simple lateral shifting measurements in that cutoffs are
plainly visible in the elongation data as abrupt shorten-
ing that occurs at a specific channel coordinate (e.g., at
up channel coordinate 10 km on the Sacramento River
dataset, in Figure 11.5b). Bends that migrate primarily in
the downstream direction without changing form are also
usually visible in the elongation-rate data. These bends
are usually characterised by a rapid shortening followed
by a relatively rapid elongation with little net change
in channel length outside the bend (e.g., at up channel
coordinate 137.7 on the older Sacramento River dataset,
Figure 11.5a).
|
V
|
11.3.4 The aggregation/disaggregation procedure
applied at a regional network scale
The process of disaggregation/aggregation described by
Alber and Piegay (2011) is a way of considering a river
continuum from a multi-scaled perspective, identifying
meaningful geographical objects and characterising them
based on photographs available at a regional scale. While
 
Search WWH ::




Custom Search