Geography Reference
In-Depth Information
1.0
0.5
0.0
0
5
10
0
20
Fraction of catchment area covered by lakes (%)
40
60
Mean basin slope (%)
1.0
0.5
0.0
0
1000
1500
2000
0
1
2
3
Annual mean degree days (K
.
d
.
10
3
)
Mean annual precipitation (mm)
1.0
0.5
0.0
0
5
10
Area (10
4
km
2
)
Figure 9.12. Specific 100-year floods plotted against catchment characteristics for 151 catchments in Quebec, Canada. From Shu and
Ouarda (
2007
).
particularly when there is considerable scatter in the rela-
tionship. It is therefore essential to interpret the relation-
ships found in regressions from a hydrological perspective.
This will assist in understanding the limits of predictability
of this relationship and the associated uncertainties. The
interpretation can take advantage of the similarity meas-
ures discussed in
Section 9.2.2
. As an example,
Figure
9.13
shows a power law relationship between the param-
eters of the generalised extreme value (GEV) distribution
and annual maximum peak, indicating a strong log-log
linear relationship between the location and scale param-
eters of the GEV distribution with respect to catchment
area. The interesting thing is now to interpret, say, the
slope of the relationship of the location parameter (top
panel) with respect to the slopes found by other authors
around the world in a comparative analysis. For example,
Merz and Blöschl (
2008a
,
b
) found for a region in Austria
that the slope strongly depended on the regional rainfall
regime. In a region such as Buwe (
Figure 9.8
), where
convective precipitation is the main flood generation
mechanism, the relationship between specific floods and
area was steep, while in a region with synoptic rainfall,
such as Gurk (
Figure 9.8
), the relationship was much
flatter. The different scaling of the floods is a reflection
of the scaling of the driving processes and space-time
interconnections of the rainfall
runoff process (Skøien
and Blöschl,
2006
). Alternatively, the characteristics of
such relationships can be interpreted by process reasoning,
making use of the derived flood frequency approach as
dicussed above (e.g., Robinson and Sivapalan,
1997a
).
These two methods, comparative and process-based, are
complementary. Both shed light on the driving processes,
but from different angles.
-
9.3.2 Index flood methods
Growth curves
In the previous sections we have seen how the selection of
stations to be used in a regression analysis is a key issue.
The idea of pooling together data from similar catchments
Search WWH ::
Custom Search