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

are able to reproduce calibration events equally well,
adoption of the same loss parameters for derivation of
Extreme design floods may produce significantly different
design floods. For example, a specific set of loss
parameters for the IL-CL and IL-PL models may yield
similar flood peaks for the 1 in 100 AEP design event, but if
the same parameters were retained to derive the 1 in 10'
AEP flood, the different loss models would produce
markedly different floods. The impact of model structure on
the estimated rainfall excess must be carefully understood
and the results over a range of conditions should be
checked against the results obtained using other models.

I
II

!

I

I

(ii) Range of application of derived loss values

The considerations in Section 4.2.1 indicate that the
appropriate loss values for a catchment depend on the
following: (i) the type of design rainfall data used (design
storm or design burst), (ii) the event magnitude and
duration (particularly for burst initial loss), (Hi) the season in
which the event falls, and (iv) the characteristics of the
catchment The designer should therefore make sure that
the loss values derived using the procedures in Section
4.2.2 are applicable to the specific design situation of
interest and, if necessary, adjust the derived losses before
applying them in design flood estimation. Section 4.2.1
gives guidance on the factors to be considered. At this
stage there is no evidence to suggest that different design
loss values are applicable for catchments of different size.

I
I

I

I
I

(Iii) Checking of design loss values using
independent flood frequency estimates

Where credible design flood estimates are available
from another, independent method, the procedure outlined
in Section 4.2.2 (c) should also be applied for the
independent testing of design loss values before they are
adopted. Beyond the credible limit of flood extrapolation, it
is not possible to test the AEP-neutralily of the adopted loss
values, and thus it is necessary to adopt a more
prescriptive, conservative approach, based on the
recommendations in sub-sections (b, iv) and (d, iii).

I
I

(iv) Changes to catchment conditions

Changes may occur in the catchment during the design
life of the structure for which the design event is required.
These changes might include urbanisation, destruction of
vegetation cover by clearing or fire, soli conservation
measures, and construction of reservoirs. While their
possible effects on design losses should be considered, the
major effects of such changes are on the response times of
runoff. These are considered in Section 4.3.3(b,iv). Losses
during relatively minor floods may be affected appreciably
by these catchment changes, but in general there will be
very little if any effect on the low design burst (IL,) losses
assumed in Extreme floods.

I

I
I

I

(b) Recommended initial loss values for use
with design bursts (IL,,)

For regions without information on pre-burst temporal
patterns it is not possible to apply directly the initial loss
values derived from an analysis of complete storms.
The selection of inilialloss for use with design bursb of
rainfall is problematic as the depth of rainfall antecedent to
the burst varies with both storm duration and event
magnitude. Traditionally, it has been assumed that the net
bias resulting from the application of storm losses obtained
from calibration -with design bursts is negligible (see
Section 4.2.2 a). However, the available evidence for flood
events more frequent than the 1 in 100 AEP event
suggests that the losses obtained from calibration to large

I

I

I

............"...-...."""....."..."...........\:1..."......"""",.."',,"""....,,.

historic floods are too low (e.g. Walsh et aI., 1991; Hill et
aI., 1996').

The expected reduction of IL, with reducing burst
duration and increasing event magnitude (see Section
4.2.1c,ii) means that the following recommendations have
to differentiate between event magnitudes.

(i) Large to Rare events

IL. values suitable for derivation of floods more frequent
than 1 in 100 AEP should be based on recommendations
contained in Book II, Section 3, or other relevant design
data for the region, as deemed appropriate.

Where possible, reconciliation with independently
derived design flood estimates should also be attempted,
as described above in 4.2.2(c). After adoption of a suitable
value of continuing loss, IL, values should be selected by
matching rainfall-based estimates with design floods
derived by independent procedures. In many cases the
credible limit of flood extrapolation is restricted to events
with AEPs equal to or more frequent than 1 in 100. It is
possible however that regional frequency or
paleohydrological procedures could be used to estimate
floods rarer than the 1 in 100 AEP event, and thus
reconciled loss values could be obtained for lower AEP
events.

(ii) Extreme events

IL, values should be varied gradually between the
values adopted for Rare and PMP events. In the absence
of any scientific justification, it is suggested that losses
between the two limits are determined from a simple
interpolation procedure. For example, if the initial loss value
for the 1 in 100 AEP event is 10 mm and that for the most
Extreme design event (with an AEP of 1 in 10') is 0 mm,
then Extreme loss values can be interpolated from a line
drawn on log-Normal probabilily paper between 10 mm at 1
in 100 AEP and, say, 0.1 mm at 1 in 10' AEP (the initial
loss of 0.1 mm is an approximation of 0 mm in the
logarithmic domain).

Alternatively, it may be assumed that the losses vary
linearly on a log-log plot of losses versus AEP; this
assumption is more consistent with the interpolation
procedure used for design rainfalls (Section 3.6), and is
also more amenable to calculation. For example, if initial
loss values L, and L, were assigned, respectively, to events
with AEPs of 1 in Y 1 and 1 in y" then the loss value to be
used in conjunction with a design burst of intermediate AEP
1 in Y could be interpolated using the following equation:

log(l ) = lag(l ) + (lag(Y) -lag(Y )) log(l2) -Iagel,)
Y' 1 log(Y2)-log(Y,)
(7)

A zero loss value is again to be approximated by a
small value, say 0.1 mm. The practical difference between
the use of Equation 7 and the assumption of log-Normal
variation is negligible given the uncertainty of loss rate
variation. A worked example describing the derivation of
intermediate losses is provided in Section 6.3.2.

(iii) PMP Design Flood

Very low values of IL, are recommended as it is
assumed that the pre-burst rainfalls associated with the
PMP design burst will either partly (longer duration bursts)
or fully (short duration bursts) satisfy soil moisture deficits.
In conformity with the adopted policy of aiming for
reasonable conservatism in the absence of more relevant
information, conservatively low estimates are generally
recommended. For PMP design burst durations
approaching the duration of the observed storms, the IL"