Geography Reference
In-Depth Information
As illustrated by the Chernoff faces example, techniques
to visualise multivariate data may be an easy way to detect
similarities, but are generally qualitative in nature. Because
of subjective interpretation of images, it is also challenging to
rank the importance of different parameters in this way. The
advantage of subjective multivariate data, on the other hand,
lies in the innate tendency of the human brain to identify
patterns and undertake classifications, which is not always
readily reproducible via algorithmic methods. A quantitative
alternative for multidimensional visualisation, which is more
algorithmic, is the Andrews curve (Andrews,
1972
). An
Andrews curve represents the characteristics of one catch-
ment by the sum of a number of harmonics, each of them
weighted with the magnitude of a catchment characteristic.
Similar catchments have similar Andrews curves, allowing
ungauged sites to be directly compared with other catch-
ments that might represent a similar flow regime. An
example of the application of Andrews curves for catchments
in the southern part of Switzerland, in the context of PUB, is
presented in Weingartner (
1999
).
more likely to characterise medium and small catch-
ments. Complex II regimes cannot be directly related
to specific hydrological processes. They result from
the stacking or overlapping of various processes
from the constituent sub-basins.
L
s(
1938
) classification scheme is a development
of Woeikof
'
vovich
'
s work. Whilst elaborated independently, it is
quite similar to the classification of Pardé (
1933
). L
'
'
vo-
vich
s classification is based on the seasonal timing and
intensity of the largest mean monthly runoff (e.g., spring
maximum with more than 50% of total annual runoff).
Furthermore, the main generation processes of runoff are
included (e.g., snow-fed). L
'
s classification scheme
was developed using data from small catchments where
there is a more direct link between dominant processes and
regime type. L
'
vovich
'
s study was the basis for a global
regime classification presented in the Mira-Atlas, pub-
lished in 1964. In the same year, the UNESCO asked the
International Geographical Union (IGU) to found a com-
mission to contribute to the first International Hydrological
Decade (1965
'
vovich
'
74). One of the main tasks of this new IGU
commission was to analyse and map the flow regimes of
the world. There has since been an explosion of studies
focused on regime type and classification. Examples of
studies include Kresser (
1961
), Gottschalk et al.(
1979
),
Aschwanden and Weingartner (
1985
), Haines et al.(
1988
),
Gustard et al.(
1989
) and Krasovskaia and Gottschalk
(
1992
).
-
6.2.3 Catchment grouping
As highlighted in
Chapter 5
, techniques of grouping and
classification underpin most attempts to relate the hydro-
logical behaviour of ungauged catchments to measured
properties of gauged systems. This section focuses particu-
larly on techniques for grouping that are specific to sea-
sonal variations in runoff and the flow regime.
Grouping based on runoff-regime types
A classification of catchments according to a particular
flow regime type offers an immediate approach for
grouping, and has been used for over 100 years to classify
global catchments (Woeikof,
1885
; Arnell et al.,
1993
).
Highly influential regime classification approaches include
those of Pardé (
1933
), and L
Grouping based on runoff: statistical approaches
Using similarity indices to define the flow regime
(
Section 6.2.2
), objective methods to obtain groups with
similar flow regime can be derived. As in the case of
mean annual runoff variability (
Chapter 5
), cluster analy-
sis (CA) is a common tool (e.g., Gottschalk,
1985
; Haines
et al.,
1988
; Guetter and Georgakakos,
1993
; Dettinger
and Diaz,
2000
; Harris et al.,
2000
; Bower et al.,
2004
;
Hannah et al.,
2005
; Monk et al.,
2008
; Laizé and
Hannah,
2010
; Kingston et al.,
2011
). As discussed in
Chapter 5
, CA depends upon the definition of distance
metrics that can be applied between groups: for analysis
of seasonal runoff variability these distances are usually
defined in terms of Pardé
vovich (
1938
). In 1933 Pardé
published his classic regime classification, which inspired
generations of hydrologists. Pardé
'
s classification is quali-
tatively based on the driving processes, timing of the
maxima and minima of runoff and inter-annual variability.
The three basic regime types of Pardé (
1933
) are:
'
Simple regimes with only two hydrological seasons
(high and low water) and one primary driving process.
The respective regime curve has only one peak.
coefficients or Fourier
coefficients.
The application of CA to regional basins rarely results
in spatially coherent groupings of catchments when
applied to either annual runoff or to the seasonal flow
regime. An example of the results of CA (Ward
Complex I regimes with several hydrological phases
induced by the contributions of different processes to
runoff at different points across the hydrological year,
e.g. rainfall, snowmelt. The respective regime curve
exhibits several peaks.
s method,
see
Chapter 5
) applied to long-term river flow regimes
based on their shape (dimensionless form) and magnitude
(size) is shown for 28 Nepalese Himalayan basins in
Figure 6.13
(Hannah et al.,
2005
). The groupings are
'
Complex II regimes can be found in large rivers, which
have tributaries with different types of flow regimes,
whereas simple regimes and Complex I regimes are
Search WWH ::
Custom Search