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

seasonal loss values suitable for design, and efforts should
be made to seek out relevant regional information where
available. Where there is clear evidence of seasonal
differences in losses, and where the seasDnal variation of
other design factors is being allowed for (see Sections 3.7
and 4.6), the loss values from the appropriate season
should be applied.

4.2.2 Methods for Derivation of Design Loss
Values

The methods for deriving suitable loss rates for use with
design rainfall bursts or storms can be divided into three
groups, as described below.

(a) Analysis of event data at site of interest.

The most common method for estimating event losses
is to derive the loss values as part of the flood-event
modelling process. This could be on the basis of events
selected on flood or rainfall magnitude, where generally
attention is focused on a small number of large events.
Discussion on the issues associated with the derivation of
losses obtained from the analysis of events selected on the
basis of flood magnitude is provided in BODk II, Section 3.
When applied to design storms, it is likely that initial losses
derived in this manner will be underestimated due to the
bias towards wet antecedent conditions.
An aiternative approach is to select events on the basis
of rainfall magnitude rather than flood size, to ensure that
the at-site analysis of concurrent storm and streamflow
data provides estimates of loss values that are not biased
towards wet conditions.
Instead of determining losses from the analysis of a
small number of selected events, it is possible to derive
losses from the direct examination of a large number of
concurrent rainfall hyetographs and streamflow
hydrographs. An example of this type of analysis is
provided in Hill et aL (1 996'). Considerable effort is
required to derive at-site loss values for an individual
gauged catchment by this method, and it is likely that such
an investigation could only be justified in a minority of
cases.

I
1
I

1

I
I

I

I
I

I

I

(b) Derivation of regional prediction equations

Repetition of the at-site loss estimation procedure for
many sites in a homogeneous region allows the estimation
of loss values applicable to catchments of similar
characteristics (see Book II Section 3, and for example
Walsh et aI., 1991). In some cases it has been possible to
derive prediction equations that incorporate directly
differences in catchment characteristics (e.g. Hill et aI.,
1996'), but in general the relationship between catchment
characteristics and loss values is limited (see Book II
Section 3, and for example Cordery and Pilgrim, 1983).

The direct estimation of loss values from regionai
prediction equations is limited to regions with similar
characteristics, and to the range of event magnitudes
represented in the data set analysed. However, with
appropriate interpretation, such estimates equations can
provide guidance for the selection of loss values in other
situations.

I

I

I

I

I
I

(c) Reconciliation with independent flood
frequency estimates

(i) Basis of procedure

Wilhthis approliCh, Illss parametefS are ~acted to
ensure reasonable agreement between flood estimates
obtained using rainfall-based procedures and those
obtained using reliable independent methods. For events

I

__......" >I, - ~"'~""CI~'''''''''' ~~.::II~.... ~~.............. ..............

up to the credible limit of extrapolation of floods, the
independent estimates can be obtained using al-
site/regional flood frequency, Dr paleohydrol09ical,
procedures. It should be recognised that there may be
considerable uncertainty in the flood frequency estimates
and thus it is only necessary to obtain reasonable
agreement within the sampling bounds of each procedure
(see point ii, below).
It is preferable that Independent estimates of design
floods be used to validate, rather than calibrate, the loss
parameters derived from at-site or at-site/regional analyses;
however, as lon9 as the routing parameters of the flood
event model have been found from calibration to major
historic events, Ihen matching independent design flood
estimates within sampling bounds is an acceptable
approach for the selection of loss values.

Loss parameters are nDt the only input that may need to
be aitered to achieve a match between the two estimates,
particularly if the AEPs of the floods used to calibrate the
routing parameters are much smaller than the probability at
which the two estimates from the two approaches are of a
similar accuracy. Further discussion on this issue is
provided in Section 4.3A

(n) RangeofappHcaUon

The information provided in Book III Section 2 should be
used for guidance on the "probability of indifference", the
AEP at which the rainfall-based estimates are considered
as accurate as flODd frequency estimates. It should be
noted, however, that the recommendations provided in
Book III Section 2 only apply to flood frequency estimates
derived using data recorded at the site of interest Use of
at-site/regional flood frequency or paleohydrologicai
procedures will appreciably reduce the probability of
indifference, and judgement will be required to determine
the sampling bounds within which agreement between the
two estimates is desirable.

4.2.3 Guidelines for Selection of Design Loss
Values

Specific guidance on the use of loss values is divided
into four sections. Firstly, in (a), general guidelines are
provided on the selection of design loss values for Large to
Extreme events, Specific recommendations are then made
in (b) and (c) for the selection of design initial losses for
design storms and design bursts, respectively. Finally, the
recommendations in (d) relate to the selection of continuing
loss design values. In (c) and (d), different design situations
for the selection of design loss values can be distinguished
depending Dn the event magnitude class (Figure 1). Where
appropriate, recommendations are given for specific
geographic regions, consistent with lhe availability of
design information for different parts of Australia.

(a) General Guidelines

(I) Selection of loss model

Guidance on the common types of loss models used in
practice may be found in Book II Section 3. The specific
recommendations in (b), (c) and (d) relate to loss values for
the IL-CL model, as a large body of relevant experience
has been accumulated over many years. However, other
loss models may be used if they can be shown to be more
appropriate in the specific situation. The following aspects
should be considered when selecting a loss model and
specific loss values for the derivation of design floods.
OiIferent loss models wilt intrQ(\wqe differing degrees of
non-linearity into the transfer between design rainfalls and
the resulting hydrograph. Thus even if different loss models