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 PMP estimates. A procedure for estimating
seasonal PMPs for short duration storms on areas up
to 1000 km' in southern Australia is given in Bulletin
53 of the Bureau of Meteorology (1994). Approximate
seasonal estimates are available for longer duration
events in southeast Australia using the GSAM
procedure, but it should be recognised that these
estimates are based on a biased seasonal sample as
the storms were selected on the basis of magnitude
rather than season. Seasonal PMP estimates for
longer duration storms in tropical areas (I.e. GTSM
estimates) are not currently available due to the small
sample used to develop the original method.

. AEP of the Seasonal PMP. At present there is no
accepted procedure for assigning an AEP to a
seasonal PMP. The problem is in fact very difficult,
and is dependent upon design objectives, the
definition of terms, and differences in interpretation of
the generalised PMP procedures. For example there
are at least two possible interpretations of the
seasonal PMP (Weinmann and Nathan, 1997); the
seasonal PMP event is considered to have either (i) a
fixed magnitude equal to the value determined for the
annual PMP event but a different AEP in each
season, or (ii) a fixed exceedance probability equal to
the value determined for the annual PMP event
Weinmann and Nathan reason that the method used
by the Bureau of Meteorology to derive seasonal
PMPs is c10seJ to this second interpretation, as
different PMP depths are being estimated for the
different seasons. Laurenson and Kuczera (1998) also
discuss two alternative interpretations and provide
equations for the calculation of exceedance probability
of seasonal PM P, but they note that their method
produces anomalous resulls when used to coostruct
annual frequency curves.

I
I
1
I
I
I
I

I
I
I
I
I
I
I

(b) Derivation of seasonal design rainfalls

It is seen that the procedure required to derive a
complete seasonal frequency curve of design rainfalls is
nDt straightforward, and is subject to differences in
interpretation, particularly in respect to assigning an AEP to
the seasonal PMP. The basic criterion to be satisfied by
any procedure for estimating seasonal rainfall frequencies
is that, for any given rainfall magnitude, the seasonal
frequencies over all seasons should add up to the AEP of
that rainfall magnitude determined from the analysis of
annual rainfalls.
In the absence of better design information, and noting
the foregoing discussiDn, the following recommendations
should prove adequate for most design problems where
seasonal effects are important (see Section 4.6a).

(i) Large events

Both seasonal and annual frequency analyses should
be undertaken using rainfall data obtained from sites
relevant to the study area. The adopted seasons should
correspond to the same seasons used in derivation of the
seasonal PMP depths, and the seasonal rainfall estimates
should be expressed as fractions of the annual estimates.
The seasonal fractions can then be converted to design
rainfall depths by mumplying by the (annual) design rainfall
values obtained from the standard informatiDn provided in
Book II Section 1 and Volume 2. .

I
I

1

(ii) Rare events

Unless specific regional estimates are avallsble, the
seasonal fractions corresponding to design rainfalls at the
credible limit of extrapolation may be obtained by an

I

~....v" v, - t,.':'llll.<:lIUVII VI L.d'!:f<;: LV L..^U""",,,;:, I ,....VU.;;l

interpolation procedure similar to that used for losses (e.g.
Equation 7), where the lower and upper end points used in
the interpolation are defined by the seasonal fractions
derived for the 1 in 100 AEP and PM P design rainfalls.
Once the seasonal fractions have been obtained by
interpolation, the seasonal design rainfalls are derived by
multiplying by the (annual) design rainfall values at the
credible limit of extrapolation. If the credible limit of
extrapolation is 1 in 100 AEP, then the shape factors (Table
7) may be used directly with the Large and PMP seasonal
design rainfalls.

(/ii) PMP events

Seasonal estimates of the PMP should be obtained
from the Bureau of Meteorology, and they may be plotted at
an AEP equal to that of the annual PMP (as per Figure 4).

3.8 Large to Extreme Rainfalls for Very Long
Durations

(a) Background

For dams with very large storage volumes relative to the
volumes of inflow floods (or dams with liltle or no spillway
provision), or for some very large catchments, it may be
possible that the critical duration of interest may appear
longer than what is available from the generalised design
rainfall information. The longest available storm durations
using procedures in Book II, Section 1 is 72 hours; the
longest available storm durations provided using the GSAM
procedure is 120 hours, and that from the GTSM procedure
is 96 hours. These durations generally relate to the
meteorological limits associated with single storm events,
and thus longer duration design events will involve the
consideration of storm sequences.
The approach to solving design problems involving long
critical durations is in essence a joint probability problem. In
special circumstances the problem may involve the
assessment of joint probabilities of extreme storm
sequences, but when considering issues associated with
reservoir outflow floods, the issue of storm sequences over
extended periods may be implicitly solved by undertaking a
joint probability analysis of inflow floods and reservoir
contents (Section 5.2).

(b) Storm sequences in southeastern
Australia

Analysis of storm data in southeastem Australia
(Bureau of Meteorology, 1998") indicates that about 40% of
large storms are preceded by a rainfall event in the 15 days
prior to the storm (see Section 4.2.1 c). Based on their
magnitude, these antecedent rainfall events appear to
comprise two different populations: most (32% of all large
storms) had accumulated rainfall totals of less than 30% of
the subsequent large storm, but a small proportion (8% of
all large storms) had accumulated rainfall totals of between
30% and 80% of the subsequent large storm. In addition,
Sinclair Knight Merz (1997) undertook an analysis of
antecedent rainfalls in the one to three month period
antecedent to the same large storms and concluded that
the rainfalls in the months immediately preceding a large
storm in this region are not likely to be greater than normal.
While there is too Itttle information with which to derive
quantitative inputs. to a joint probability analysis, it is
reasonably clear that large, long duration events in
southeastern Australia are unlikely to be preceded by
$ignifle$.fil antecedent rarmalls.