Geography Reference
In-Depth Information
effective rainfall as a function of rainfall depth, catchment
characteristics and antecedent soil moisture and was
developed from experimental catchment data in the USA
(SCS,
1956
). The degree to which the method can be
generalised beyond its original domain of applicability
has been examined by a number of studies. For example,
Merz and Blöschl (
2009a
) compared the SCS curve
number derived from soil type, land use and antecedent
rainfall according to the SCS curve number method with
the curve number back-calculated from event runoff coef-
ficients for events in numerous Austrian catchments. The
results showed that the back-calculated SCS curve
numbers were not correlated with the predicted SCS curve
numbers. In areas with large rainfall, runoff coefficients
tended to be high, but these were also the forested areas
where the SCS method predicts the smallest runoff coeffi-
cients. A similar result was found by Hoesein et al.(
1989
),
who tested the SCS curve number method with a procedure
analogous to the probabilistic rational method in eastern
Australia. These comparisons illustrate that care must be
taken when extrapolating empirical relationships such as
the SCS curve number method beyond the regime of runoff
processes for which they were developed. More recent
methods estimate event-based model parameters on the
basis of irrigation experiments. Examples include the
(
2003
), who related runoff coefficients and surface rough-
ness to indicators such as vegetation species, land use, soil
texture, drainage density and slope, which can be assessed
during reconnaissance field trips. They then developed a
rule-based method that allowed them to estimate the model
parameters for ungauged basins. The spatial distributions
obtained could then be used to assist in parameterising
distributed models in ungauged catchments (e.g., Rogger
et al.,
2012a b
; see
Chapter 4
).
role of the storm duration and the antecedent wetness
conditions, expressed in the form of the runoff coefficient,
in the mapping of rainfall and flood return periods using
the derived distribution approach.
They found that, unless adjustments are made to the
parameters of the rainfall
runoff model, the return period
of the flood peak could be much higher than the return
period of the design storm (
Figure 9.19a
)
. The ratio of the
return period of rainfall and runoff depends mainly on
the average wetness of the catchment. In arid climates the
return period of the flood peak could be of the order of
hundreds of times that of the rainfall return period, while in
humid climates the maximum flood return period is never
more than a few times that of the corresponding storm.
Figure 9.19b
shows the non-exceedance probability of the
runoff coefficient providing the match of the design storm
and flood return periods obtained in a hypothetical simula-
tion study for varying climate and return periods. There is
no unique non-exceedance probability of the runoff coeffi-
cients that give a 1:1 correspondence of T
P
and T
Q
. For the
driest system, it significantly depends on the return period
(ranging from 0.5 to 0.8), while it is almost constant and
close to 0.8 for the wettest system. In all cases, however, it
is evident that the runoff coefficient that gives the 1:1
matching of design storm and flood return periods is
greater than the median value that has been suggested for
use in a common application of the design storm method
(e.g., Pilgrim and Cordery,
1993
).
For this reason, applications of event-based methods are
usually preceded by analysis of rainfall
-
flood data
from several catchments in the region to enable estimation
of these relationships. Such relationships are either
embodied in regional guidelines (e.g., Australian Rainfall
and Runoff,
1987
) or performed on a case-by-case basis for
the ungauged basin of interest. The model parameters that
transform a T-year storm into a T-year flood peak, i.e. are
return-period neutral, are mapped regionally and are then
used in routine predictions. Although this procedure may
be affected by considerable error due to uncertainties in the
return period (see e.g., Rahman et al.,
2011b
), there is the
potential
for local processes such as floodplain inundations
and obstruction by hydraulic structures to be included. For
whichever model is used, a general recommendation is to
analyse runoff data from similar gauged catchments in the
region whenever possible to estimate the model parameters
in ungauged catchments (IH,
1999
; Blöschl,
2005
).
-
runoff
-
Antecedent soil moisture and mapping of rainfall to
flood return periods
The second major challenge to
the application of event-based methods for flood predic-
tions in ungauged catchments is how the model parameters
change with the event magnitude and how these changes
are related to the return periods of the rainfall and the
floods. In other words, parameter dependence on return
period must be known a priori for its application in
ungauged basins. From a process point of view, the rela-
tionship between the return periods of the rainfall inputs
and flood peak must account for the storm duration, storm
intensity, temporal and spatial storm patterns and the
dynamics of runoff generation, which are controlled by
antecedent wetness conditions, soils and topography, evap-
oration and other processes (Lamb,
2005
). Viglione and
Blöschl (
2009
) and Viglione et al.(
2009a
) analysed the
Estimating the entire population of flood events
One approach that overcomes the problem of assigning
return periods to individual events does so by transforming
the probability distributions of rainfall
into probability
distributions of floods using a rainfall
runoff model (e.g.,
Eagleson,
1972
; Wood,
1976
; Gottschalk and Weingartner,
-
Search WWH ::
Custom Search