Geography Reference
In-Depth Information
5
a)
b)
50
NV
MO
VA
NC
WI
CA
IN
WA
4
40
3
30
2
20
1
10
0
0
0.2
0.4
0.6
0.8
1.0
0.2
0.4
0.6
0.8
1.0
Baseflow index
Baseflow index
Figure 7.11. Relationship between the shape parameter
(of the gamma distribution) of the SFDC (FDC for slow flows) and the baseflow index
for 197 catchments across continental USA: (a) variability between catchments using multi-year rainfall
κ
-
runoff data; (b) variability between
years for eight selected catchments. From Cheng et al.(
2012
).
literature has not yet reached consensus on the best
grouping method or on how to select the most suitable
pooling approach, which remains therefore an open prob-
lem in the regionalisation of FDCs.
The process of grouping catchments involves two steps:
(i) the choice of some kind of quantitative similarity index
that will serve as the basis for choosing similar catchments;
and (ii) a grouping method that uses estimates of the
similarity index and organises the catchments into distinct
groups (Blöschl,
2005
a).
Section 7.2.2
presented a survey
of the indices that can be used to group hydrologically
similar catchments. These indices can be based on meas-
ured runoff and reflect the signature that is being predicted.
Alternatively, the grouping can be done using non-
parametric approaches; for example, Ganora et al.(
2009
)
defined a metric to quantify the differences
1997
; Croker et al.,
2003
; Cheng et al.,
2012
), which
although derived from runoff reflects the interaction of
climate and geology.
One way of grouping catchments is by delineating fixed
and contiguous (i.e., geographically identifiable) regions
(e.g., Castellarin et al.,
2004a
; Mohamoud,
2008
; Viola
et al.,
2011
). One can expect that contiguous areas will be
characterised by similar climate, topography and geology
(and all other characteristics that derive from them such as
soils and vegetation) giving rise to similar catchment
hydrological response, and therefore similar FDCs. Of
course, this is not the only possibility. One catchment can
be similar in terms of the processes leading to the FDC to
another catchment that is not necessarily contiguous. Clus-
ter analysis seems to be the prevailing approach concern-
ing the regionalisation of FDCs with which catchments can
be grouped based on the similarity indices discussed above
(see e.g., Castellarin et al.,
2004a
, in central Italy; Sauquet
and Catalogne,
2011
, in France; Tsakiris et al.,
2011
,in
Massachusetts, USA; Ley et al.,
2011
, in Germany).
Several methods have been proposed in the literature for
grouping catchments. There are algorithms that work dir-
ectly on basin characteristics through the definition of a
measure of distance (e.g., Ganora et al.,
2009
; Sauquet and
Catalogne,
2011
). Other clustering algorithms are based on
the utilisation of non-linear approaches such as regression
tree clustering (Sauquet and Catalogne,
2011
) or unsuper-
vised neural networks (self-organising maps; see e.g., Ley
et al.,
2011
). Whether or not the climate or catchment
characteristics or a combination of both should be used in
the cluster analysis depends on the dominant processes.
However, clustering procedures exist
-
measured as a
distance
between the FDCs and then related these dis-
tances to differences among catchments in terms of catch-
ment and climatic characteristics.
If a robust relationship can be found between the slope
of the FDC (or any other FDC similarity index) with
catchment and/or climate characteristics, then the catch-
ment characteristics or climate classes can be used to
delineate the catchments into hydrologically similar
groups. One can also use non-parametric approaches to
delineating catchment groups on the basis of climate, using
several indicators together to group catchments (as was
done for the seasonal flow regime by Coopersmith et al.,
2012
, see
Chapter 6
) or on the basis of surrogates for
climate and catchment characteristics, including combin-
ations of these, as the basis for grouping catchments. An
example is the baseflow index (e.g., Claps and Fiorentino,
-
to assist
in the
Search WWH ::
Custom Search