PUB in practice: case studies - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 309

Table 11.9. Cross-validation performance of estimating the 100-year flood in ungauged basins in Austria
as a function of mean annual precipitation (MAP) (only gauging stations with more than 40 years of flood
records were used)

MAP (mm) Number of catchments Rbias (%) RRMSE (%)

< 800 mm 32 11.7 64.5

800 - 1500 mm 204 8.37 33.5

> 1500 mm 62 6.40 34.9

All catchments 298 8.32 38.4

floods in wet catchments are estimated more accurately than
in drier catchments). At the local scale, more detailed ana-
lyses involving rainfall

damage to public infrastructure estimated to exceed Aus$5
billion (PWC, 2011 ). While the majority of this flood
damage occurred in urban areas such as Brisbane, which
are located in well-gauged river basins, a significant portion
of it was experienced in ungauged basins distributed
throughout the state. Limited design flood information for
these areas may not only contribute to a larger flood damage
bill, but also to greater trauma and personal loss suffered by
people in these areas when a major flood event occurs.
Flood management involves various activities, e.g., flood
risk assessment, flood forecasting, flood emergency manage-
ment, flood plain development control, flood protection and
insurance. For all of these studies, past information/data on
flooding is a vital input. Due to the sheer size of Australia,
recorded flood data have limited spatial coverage and hence
' Prediction in Ungauged Basins ' is an important issue.
Australian Rainfall and Runoff (ARR), the national
guide for flood estimation, has a dedicated chapter on
'

runoff modelling may be needed
for specific projects to account for additional local effects.
An important step in estimating the flood discharges was
the involvement of the Hydrographic Services. The inter-
play of formal regionalisation methods based on hard data
with the understanding of local hydrologists based on
proxy data turned out to be extremely helpful in estimating
the flood discharges. The flood discharges estimated in this
study were transformed to flood hazard zones by other
project partners using hydraulic modelling, and were visu-
alised on the Web ( www.hochwasserrisiko.at ) with public
access. These zones are the main information for comply-
ing with the preliminary assessment requirements of the
Flood Directive in Austria. The flood discharges from the
HORA project are also used for producing the detailed
flood hazard maps required by the Flood Directive. The
results from the flood regionalisation are much appreciated
by end-users of the water authorities and consultants in
Austria, both in the context of implementing the Flood
Directive and for other purposes of water resources man-
agement. In the meantime, the estimates of the HORA
project have become a standard of flood estimates in Aus-
tria for a range of purposes.

-

. However, these methods were
published in 1987. Recognising that the national flood
database has been augmented with an additional 25 years
of data and that there have been significant new develop-
ments in regional frequency analysis, the National Com-
mittee on Water Engineering of Engineers Australia
initiated a review of the 1987 procedures.
Although regional flood frequency analysis (RFFA)
methods are often intended for

regional flood methods '

'
purposes, their application and importance is much broader
in that they can serve as a benchmark for more advanced
methods based on design storms and rainfall

' quick flood estimation

11.11 REVISION OF AUSTRALIAN
RAINFALL AND RUNOFF FOR
IMPROVED FLOOD
PREDICTIONS
a. rahman, k. haddad, e. weinmann and
g. kuczera

runoff model-
ling. The current upgrade of regional flood estimation
methods in Australia is intended to be released as an

-

' easy

to use web-based software to make design flood frequency

information more accessible to the Australian community.

'

The issue from societal and hydrological perspectives
Australia is the driest inhabited continent, with highly vari-
able year to year water availability, leading to cycles of
significant droughts and flooding. Episodes of serious
flooding can affect large areas, such as experienced during
the 2010

Description of the study area
Australia has an area of 7.7 million km 2 , with a diverse
landscape and climate. Up to about 75% of the country is
classified as

. More than 80% of the
continent receives annual rainfall less than 600 mm; only
Antarctica

' semi-arid ' and ' arid '

11 season in Queensland, when flooding
impacted around 70% of the state and resulted in total

-

receives less rainfall (on average) than