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
1

nonnal range of observations must be extrapolated to the
range of interest in this Book.

(c) Design flood characteristics

In many cases, the flood hydrograph is required as well
as the peak discharge and in some cases may be more
important For the design of a dam spillway or a detention
basin, floods calculated from a range of design rainfall
durations should be routed through the storage for a variety
of combinations of spillway and gate configurations,
operating procedures and dam crest heights to determine
the optimum design. Different durations of design rainfalls
may be critical for different configurations and combinations
of conditions, which should all confonn with the
recommendations of ANCOLD (1998). The complete
hydrograph of the design flood is also required for flood
studies where flow profiles in natural or constructed
channels are to be calculated by unsteady flow prDcedures.

I
I

1

(eI) Special circumstances

The recommendatiDns in this section apply only to the
direct estimation of floods from design rainfalls, without the
consideration of other factors. However, there may be
some cases where some other set of circumstances may
be critical for design. One example is dams with very large
storages where it is necessary to take explicit account of
initial storage level or rainfall sequences over very long
durations. The assessment of consequences on
communities downstream of a storage may require
consideration of concurrent floods in adjacent catchments.
Also, a series of dams along a given stream requires
special consideration, as failure of an upstream dam could
impose more severe conditions on a downstream dam than
its nonnal design flood. It is the responsibility of the
designer to consider all circumstances that are critical for
design. A number of issues related to these and other
design considerations are discussed in Section 5.

I
I
I
1
1

4.2 Losses and Rainfall Excess

I

4.2.1 General Considerations

A loss model is needed to partition the design rainfall
input into rainfall excess (runoff) and loss. General
guidance on loss mDdelling is provided in Book II Section 3,
and the following discussion focuses mainly on those
aspects that are related to the estimation of Large to
Extreme floods.

I
I

(a) Importance of design losses - Large to
Extreme events

Like temporal patterns of storm rainfan, design losses
are highly variable and can have an appreciable impact on
both the peak and volume of the resulting flood. A given
rainfall occurring on a dry watershed produces a
significantly smaller flood than the same rainfall occurring
on a wet watershed. For more frequent events, loss may be
the most important factor. Joint probability approaches
(Weinmann et aI., 1998) are better able to deal with the
high variability of design losses as, in contrast to the design
event approach, they IIse a probability diStribution of loss
values, rather than a single representative value. However,
the importance of losses diminishes with decreasing AEP,
and for Extreme events it is Iikety that losses are of lesser
importance than temporal patterns. For the estimation of
Large to Extreme floods, the use of single-valued
fejlr<i5enlatille *siglllosses is lhos generaVy ade~ta.

For extreme rains and floods, a much greater proportion
of a catchment may become saturated during the event

I

I

I
I

I

---.,.. ...-...,......-.. -. --';3-'- -....-...-. .----

than is the case for most floods in the observed range.
Also, during extreme rainfalls, vegetation may be stripped
from the catchment, thus resulting in an increase in the
speed and volume of the overland flow component of runoff
(Kemp and Daniell. 1997). Any evidence considered
relevant to the changed behaviour of the catchment under
extreme rainfall conditions should be considered when
estimating the resulting design flODds.

(b) Losses associated with design storms and
design bursts

(i) Design storms versus design bursts

When considering the adoption of design losses it is
necessary to understand the distinction between design
bursts of rainfall, and design stonns. A schematic diagram
illustrating the difference between the two concepts is
shown in Figure 7. Design bursts are generally preceded by
SDme lower intensity rainfall (pre-burst rainfall), they do not
represent complete storms. The selection of design loss
values must take into consideration the manner in which
the design information was derived, and whether the losses
are to be applied to design storms or design bursts.

(Ii) Initial losses for design storms and design
bursts

In the context of the above distinction between design
storms and design bursts, it is clear that generally the burst
initial loss is less than the storm initial loss, as the pre-burst
rainfall within a complete storm tends to satisfy part of the
initial loss requirement The difference between IL. and IL.
increases with decreasing burst duration.

Intrinsically the initial loss concept relates to complete
storms, as the catchment saturation process depends on
the total rainfall from the beginning of the stonn event (and
pre-storm rainfall). Initial loss values from calibration to
historic events are also based on the analysis of complete
storms. However, as the design rainfall and temporal
pattern information presented in Section 3 generally relates
to design bursts, derived loss values are not directly
compatible with the design rainfall data. There are two
basic options' for overcoming this incompatibility:

. transform the design rainfall burst into complete design
storms, then apply storm initial losses (in this
transfonnation a constant ratio between burst rainfall
and storm rainfall is assumed, regardless of event
frequency).

. use design rainfall bursts with burst initial losses, I.e.
use initial loss values that have been adjusted for the
occurrence of pre-burst rainfall.

This distinction between the storm and burst initial loss
concepts has important implications for the manner In
which design inputs are combined (e.g. see Hill et aI.,
1998) and specific recommendations for both concepts are
provided below.

(ill) Continuing tosses for design storms and
design bursts

For the continuing loss rate or the proportional loss
fraction, there is no systematic difference between the loss
values associated with design storms or design bursts, aOO
the two are treated as equivalent

(c) Variation of loss values with event
magnitude

The discosslon in (b) above makes it clear that stofm
initial loss (IL.) and burst initial loss (14) are expected to
show a variation with event magnitude. The two types of