FLOOD01798

• County: Statewide
• Title: Australian Rainfall and Runoff 1998, Revision of Book VI - Estimation of Large to Extreme Floods
• Date: 11/28/1998
• Prepared By: Rory Nathan, Sinclair Knight Merz
• Floodplain - Doc Type: Educational/Technical/Reference Information

I
I
I
'J
f

I
I

I
I
I
I
t
I
I

I
I

I

I
I

L-'1'V"Il I.....

routing characteristics of larger events may be considerably
different. Thus while calibration of a model provides
valuable information on the flood response of a catchment,
caution is needed when using the model to estimate design
floods of much larger magnitude. Extrapolation beyond the
range of observed events will introduce considerable
uncertainty into the estimated parameter values.

The following sections provide specific guidelines on
calibration methods and are aimed at minimising the
uncertainties in the selection of design values,

(b) Calibration to observed flood events

The model should be calibrated to events over a range
of flood magnitudes up to the largest observed event, and
the results be analysed for the presence of non-linearities.
The representation of the calibration results on a log-log
plot of storage versus discharge may be helpful in this
respect, and the variation of parameter values with flood
magnitude should be assessed. If appropriate data is not
available at the site of interest, consideration should be
given to undertaking such analyses on a similar catchment
with an observed very large event.

(c) Reconciliation with flood frequency
estimates

The reconciliation of rainfall-based estimates with flood
frequency estimates can provide important information on
flood response characteristics for Rare to Extreme design
events. With this approach, design rainfall information is
prepared for a specified exceedance probability, and then
used with a given set of parameter values and input
assumptions to derive a design hydrograph. The peak (or
volume) of the design hydrograph can then be compared to
the corresponding design flood obtained from at-
site/regional flood frequency analyses (Book IV, Section 2),
or paleohydrological procedures (Section 4.9). The model
inputs associated with the greatest uncertainty can be
varied within appropriate limits to ensure agreement
between the selected design flood estimate. The nature of
the model inputs to be varied depends on the model
selected, but as discussed in Section 4.2.2 (c) it is
preferable that routing parameter values be obtained from
calibration to major historic events, and that loss parameter
values be obtained from achieving reasonable agreement
with flood frequency based design flood estimates. It is
recommended that this comparison be undertaken for a
range of exceedance probabiiities to ensure a consistent
variation of parameter values with flood magnitude (Boyd
and Cordery, 1989).

The approach is suited to ungauged catchments using
regional-only flood frequency methods, as well as sites with
limited information. It is considered to be particularly useful
when combined with flood frequency information based on
paleoflood estimates. The degree to which the rainfall-
based and flood frequency estimates should agree is of
course dependent on the reliabiiity of the flood frequency
estimate, and consequently it is recommended that
confidence limits be derived for the flood frequency
estimates in order to assess the significance of any
differences. Guidance on the relative weight to be given to
results obtained from flood frequency and rainfall-based
methods is provided in Book III Section 2.

The attraction of this approach is that it provides a
mechanism for reviewing the efficacy of the AEP-neutral
design inputs. In effect, the approach treats the flood model
as a transfer fuoction, where the parameter values are
selacted to ~ns<>re thal tlre a4<lpled c!eslgn inputs yield the
desired design outputs. Any errors in the design inputs thus
have the potential to be offset by the selection of parameter
values, though hydrological judgement should be used to

.............." VI - .....:>.",.""'....."...' ........~......~... ...^,,""....., I ,.............

ensure that the parameter values of the transfer function
are still consistent with the conceptual understanding of
catchment processes. It is desirable that the rainfall-based
design floods be matched to the largest possible values
that can be estimated credibly by flood frequency methods.
This will help ensure that design floods of lower
exceedance probabilities incorporate the appropriate
degree of non-linearily associated with the transformation
of design rainfalls to design floods.

(d) Selection of model parameter values for
design

The parameter values for design flood estimates in the
Large to Extreme event range should be selected on the
basis of the most relevant calibration results from (b) or (c)
above, supplemented by information from other
considerations. Il is evident that the appropriate degree of
model non-linearity for Rare to Extreme events depends on
topographic and hydraulic factors; a hydraulic analysis may
thus provide valuable information on the general form of the
storage-discharge reiationships characterising a routing
element. As an example, the analysis of hydraulic channel
and flood plain flow characteristics may shed some light on
the nature of non-linearity in the streamflow routing
elements in the extreme flood range.
It is recognised that in many cases the constraints on
the study budget will limit the extent to which the above
measures can be considered. It will thus be necessary to
place a greater reliance on experience gained from earlier
studies and insight obtained from an analysis of relevant
regional information. The following two sections give
recommendations based on the results of previous studies
with runoff routing models and unit hydrograph models.

4,3.5 Specific Recommendations for Runoff
Routing Models

(a) Gauged catchments

As discussed by Pilgrim (1986), a value of the exponent
m in the power law storage-discharge relation (Equation
3.2, Book V) of less than 0.8 is generally conservative, in
that extreme floods are overestimated. The recommended
procedure described below for parameters associated with
the PMP design flood applies directly to the RORB model
(Laurenson and Mein, 1997), as most published information
relates to this model.

(i) Where most of the valleys in the catchment are V-
shaped with only small flood plains:

. if the peak of the calibration flood is greater than
40% of the flood resulting from the PMP, and the
calibrated m is less than or greater than 0.8, then
adopt the calibrated value; and,

. otherwise, use m = 0.8.

(ii) Where many of the valleys in the catchment have
appreciable flood plains, the value of m in the above
recommendations should be changed to 0.85 rather
than 0.8. A value greater than 0.85 could be adopted,
but this value has been recommended as providing a
reasonable degree of conservatism.

Il should be noted that the above recommendations for
m relate to all floods beyond the' credible limit of
extrapolation. If a different value of m is selected for floods
of a lesser magnitude, then the dimensionless coefficient k
in the power law storage-ctischarge relation (Equation 3.2,
Book V) should be adjusted to ensure that the magnitude of
flow at the credible ~mit of extrapolation is <lIlehanged when
used with the new value of m. An initial estimate of the
required value of k can be obtained by means of Equation