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
I
I
I

I
I

I
I
I
I

I

I

I

I

I

I

I

UKAt"1 U

concurrent floods is the same as the correlation between
the areal rainfalls. Alternatively, the correlations between
rainfalls could be used in conjunction with a rainfall-based
method to provide concurrent lIooa hydrographs.

Relevant correlations could be derived by an analysis of
concurrent areal rainfalls over different locations, though
indicative estimates could be obtained (and then
subjectively factored up) from data on the de-correlation
distance of point rainfall maxima. Figure 10 illustrates such
relationships for Tasmania, Queensland and Victoria. As it
is expected that the correlation between areal rainfalls is
greater than that between point values, the relationships
presented in Figure 10 could be assumed to represent a
lower limit of correlations for the derivation of concurrent
floods.

There is some evidence to suggest (e.g. Nandakumar
et aI., 1997) that the log-Normal distribution for the marginal
distributions of rainfalls provides a satisfactory
approximation over the range of AEPs of interest; given the
uncertainty of the correlation structure and the practicai
benefits of such an assumption, it is thus considered that
adoption of a bivariate log-Normal distribution is
acceptable. If required, rainfalls could be input into
hydrograph models to yield information on hydrograph
shape as well as peak, but in most situations it is likely that
the inherent uncertainties and relative importance of the
inputs will undermine the value of such effort.

5.4 Consideration of Snowmelt

5.4.1 Overview

Snowmelt can have an appreciable impact on the timing
and magnitude of floods, though there are only a small
number of areas in Australia where it needs to be
considered. A large number of different methods are
available for estimating snowmelt. The variety of available
methods reflects the different purposes for which they have
been developed, and the different data resources available
for their use. While there is a considerable body of
literature concerned with the simulation and quantification
of snowmelt processes, there is unfortunately little
guidance on estimating the snowmelt component of design
floods.
The snowmelt algorithms used in the established flood
event models can be broadly divided into two groups. One
group of models is based on a femperature index approach
in which temperature alone is used as a surrogate for the
energy available for snowmelt. Another group of snowmelt
algorithms is based on an energy balance approach in
which energy fluxes are calculated explicitly using

1.0
0.8
c
.2
.. 0.6
~
0 0.4
0
0.2
0.0
0

Queensland

1.0
08
0.6
0.4
0.2
0.0
2000 0

t500K VI - estimation 01 Large to t:.xtreme ....looOS

physically-based process equations. The results of a recent
international comparison of snowmelt runoff models (World
Meteorological Organisation, 1986') indicate that the
temperature index approach has an accuracy comparable
to more complex energy budget formulations.
Unfortunately, however, the method does not lend itself to
hourly computations (which are required for flood event
estimation purposes) because it is the radiation component
which is mainly responsible for the hour-to-hour variations
(Rango and Martinec, 1995).

5.4.2 Selection of Snowmelt Model

The selection of an appropriate method for snowmell
estimation is subject to the following two conflicting
requirements: (I) the need to model as accurately as
possible the snowmelt process; and, (ii) the need to adopt a
parsimonious model for use in design. The resolution of
these two conflicting requirements is a common problem in
engineering hydrology, and the accepted philosophy of
approach is to match model complexity with the nature of
the available data. While the adoption of a complex,
physically-based model may appear theoretically
appropriate, in practice without the data to confirm
component processes such models may perform no better
than over-parameterised conceptual models. Parsimony in
design snowmeit estimation is particularly important
because, compared to rain-only flood event models, there
is a considerable increase in the number of factors that
influence the transfer from rainfall to runoff. The salient
factors depend on the nature of the transfer function used,
but in general it is necessary to consider carefully the
inputs related to initial depth and density of the snowpack,
the nature and duration of antecedent conditions prior to
the rainfall event, windspeed, and the temperature
sequence.
The most appropriate method to use for the derivation
of snowmelt design floods will depend largely on the nature
of the available data. Practitioners are encouraged to
review carefully the type of data that can be obtained for
the site of interest, and to select a model that is
commensurate w~h the complexity of the available data. A
number of suitable models are commercially available (e.g.
U.S. Army Corps of Engineers, 1990; Quick, 1993), though
there is little documented experience with their application
to Australian conditions.

5.4.3 Application to Extreme Events

It is general international practice to maximise all salient
factors contributing to rain-an-snow runoff (e.g. US Army
Corps of Engineers, 1960; NERC, 1975; Bergstrom et al.,
1996). Typically, the antecedent snowpack is set equal to

Tasmania

1.0
0.8
0.6
0.4
0.2
0.0
SOO 0

600

Victoria

1500

1 00 200 300 400
Distance (km)

Distance (km)

5110

1000

Oisftlnce (km)

- Approximate median

200

400

- Approximate 90% prediction limits

Figure 10 Variation of correlation between point rainfall maxima and distance for Queensland, Tasmania and
Victoria (after McConachy, 1997; N, Nandakumar, pers. comm.).