Content uploaded by James Henry Ricketts
Author content
All content in this area was uploaded by James Henry Ricketts on Jan 03, 2015
Content may be subject to copyright.
Climate change impacts on
fi re-weather in south-east
Australia
K. Hennessy, C. Lucas* N. Nicholls* J. Bathols, R. Suppiah
and J. Ricketts
CSIRO Marine and Atmospheric Research
* Bushfi re CRC and Australian Bureau of Meteorology
www.csiro.au
ACT Government
Climate Change Cover.indd 1Climate Change Cover.indd 1 25/1/06 9:09:48 AM25/1/06 9:09:48 AM
2
Enquiries should be addressed to:
Kevin Hennessy
CSIRO Marine and Atmospheric Research
PMB No 1, Aspendale, Victoria, 3195
Telephone (03) 9239 4536
Fax (03) 9239 4444
E-mail Kevin.Hennessy@csiro.au
Important Notice
© Copyright Commonwealth Scientific and Industrial Research Organisation
(‘CSIRO’) Australia December 2005
All rights are reserved and no part of this publication covered by copyright may be reproduced or copied in
any form or by any means except with the written permission of CSIRO.
The results and analyses contained in this Report are based on a number of technical, circumstantial or
otherwise specified assumptions and parameters. The user must make its own assessment of the suitability
for its use of the information or material contained in or generated from the Report. To the extent permitted by
law, CSIRO excludes all liability to any party for expenses, losses, damages and costs arising directly or
indirectly from using this Report.
Use of this Report
The use of this Report is subject to the terms on which it was prepared by CSIRO. In particular, the Report may only be
used for the following purposes.
x this Report may be copied for distribution within the Client’s organisation;
x the information in this Report may be used by the entity for which it was prepared (“the Client”), or by
the Client’s contractors and agents, for the Client’s internal business operations (but not licensing to
third parties);
x extracts of the Report distributed for these purposes must clearly note that the extract is part of a larger
Report prepared by CSIRO for the Client.
The Report must not be used as a means of endorsement without the prior written consent of CSIRO.
The name, trade mark or logo of CSIRO must not be used without the prior written consent of CSIRO.
ISBN 1 921061 10 3
For more information about climate change, see http://www.cmar.csiro.au/impacts/index.html and
http://www.bom.gov.au/climate/change
Climate Change Cover.indd 2Climate Change Cover.indd 2 25/1/06 9:09:52 AM25/1/06 9:09:52 AM
CONTENTS
Executive summary 5
Technical summary 6
1 Introduction 11
2 Methodology for an updated fire risk assessment in south-east Australia 13
3 Results 21
4 Discussion 30
5. Gaps in knowledge and research priorities 31
6 References 32
Appendix 1 IPCC scenarios of global warming 35
Appendix 2: Soil Dryness Index 37
Appendix 3: Keetch Byram Drought Index 46
Appendix 4: Forest Fire Danger Index 55
Appendix 5: Grassland Fire Danger Index 72
3
4
ACKNOWLEDGMENTS
This report is a co-operative, inter-jurisdictional initiative which has been jointly funded by the
Commonwealth, NSW, Victorian, Tasmanian and ACT Governments.
Comments on this report were gratefully received from Laura Holbeck (Australian Greenhouse Office in the
Department of the Environment and Heritage), Alasdair Wells and David Palmer (Tasmanian Department of
Primary Industries, Water and Environment), Oliver Woldring (NSW Greenhouse Office), Rod Anderson,
Duncan Pendrigh and Jack Holden (Victorian Department of Sustainability), Ross Bradstock (NSW
Department of Environment and Conservation), Malcolm Gill (CSIRO Plant Industry) and John Raison (Ensis
Environment).
The CSIRO climate models used in this report were developed with assistance from the Australian
Greenhouse Office.
EXECUTIVE SUMMARY
Fire risk is influenced by a number of factors – including fuels, terrain, land management,
suppression and weather. This study assesses potential changes to one of these factors, fire-weather
risk, associated with climate change. Fire-weather risk relates to how a combination of weather
variables influences the risk of a fire starting or its rate of spread, intensity or difficulty of
suppression. The study is based in south-east Australia, an area projected to become hotter and drier
under climate change.
The study uses fire danger indices, such as the Forest Fire Danger Index (FFDI) and Grassland Fire
Danger Index (GFDI), to provide an indication of fire risk based on various combinations of
weather variables. These variables include daily temperature, precipitation, relative humidity and
wind-speed.
Fire danger indices are calculated for historical weather records from 1974-2003 for sites in New
South Wales, the Australian Capital Territory, Victoria and Tasmania. Two climate models are then
used to generate climate change scenarios for 2020 and 2050, including changes in average climate
and daily weather variability. Fire danger indices are then calculated for 2020 and 2050.
This study is a significant methodological improvement on earlier fire risk assessments in Australia.
It avoids biases from:
• using raw daily climate model data that may not be representative of observed climate, or
• inadequate assessment of changes in extreme weather events through failure to take sufficient
account of likely changes in daily weather variability.
A key finding of this study is that an increase in fire-weather risk is likely at most sites in 2020 and
2050, including the average number of days when the FFDI rating is very high or extreme. The
combined frequencies of days with very high and extreme FFDI ratings are likely to increase 4-25%
by 2020 and 15-70% by 2050. For example, the FFDI results indicate that Canberra is likely to have
an annual average of 25.6-28.6 very high or extreme fire danger days by 2020 and 27.9-38.3 days
by 2050, compared to a present average of 23.1 days. The increase in fire-weather risk is generally
largest inland. Tasmania is likely to be relatively unaffected.
The study also indicates that the window available for prescribed burning may shift and narrow. It
is likely that higher fire-weather risk in spring, summer and autumn will increasingly shift periods
suitable for prescribed burning toward winter.
A number of uncertainties remain when assessing potential changes to fire-weather risk associated
with climate change. These uncertainties relate to:
• the quality of data for some weather variables
• the possibility of different results arising from the use of other climate models
• changes in seasonal indicators used for fire preparedness planning
• changes in rainfall thresholds required to control fires
• changes in ignition and fuel load
• changes in El Niño-Southern Oscillation events under climate change.
5
TECHNICAL SUMMARY
Since 1950, rainfall has decreased in south-east Australia, droughts have become more severe and
the number of extremely hot days has risen. The effect of these changes on fire frequency and
intensity is not evident, although it is clear that hotter and drier years have greater fire risk. Climate
change projections indicate that the south-east is likely to become hotter and drier in future. The
aim of this study is to assess potential changes in fire-weather risk associated with future climate
change, due to the enhanced greenhouse effect. Fire weather is only one of the important factors
determining fire risk and fire behaviour – fuels, terrain and suppression are also critical, but these
have not been assessed in this report. This is just a first step toward better informing fire
management agencies and researchers about climate change risks. Ongoing engagement between
scientists and fire management agencies is needed to maximise the value of this assessment.
The weather variables required for this analysis were daily maximum temperature, precipitation,
3 pm relative humidity and wind-speed. For the 30-year period 1974-2003, data for all four weather
variables were available at 17 sites in New South Wales (NSW), the Australian Capital Territory
(ACT), Victoria and Tasmania, namely:
• NSW: Coffs Harbour, Cobar, Williamtown, Richmond, Sydney, Nowra, Wagga, Bourke,
Cabramurra
• Victoria: Mildura, Melbourne, Laverton, Sale, Bendigo
• ACT: Canberra
• Tasmania: Hobart, Launceston
The maximum daily Forest Fire Danger Index (FFDI) and Grassland Fire Danger Index (GFDI)
were calculated at each site for “present” conditions (1974-2003). The FFDI and GFDI are used
operationally to monitor fire risk, schedule prescribed burning and declare Total Fire Ban days.
Climate change scenarios for 2020 and 2050 were generated from two CSIRO climate models.
These scenarios included changes in average climate and daily weather variability, and were applied
to observed daily weather data. This method is unique in Australian fire risk assessments, and
perhaps internationally. It avoids the limitations of two other methods commonly used: (1) biases
often found when using raw climate model data that include changes in daily variability, and (2)
inadequate assessment of changes in extreme weather events when applying changes in monthly-
average climate to observed daily data. Our method includes changes in daily variability without the
biases from raw climate model data, giving more reliable fire risk projections.
6
The choice of climate simulations for this study was constrained by a number of factors; (1) models
that perform well over south-eastern Australia, (2) availability of simulated data with fine resolution
(grid-spacing of 50 km or less), and (3) availability of simulated daily weather data from which to
compute changes in daily variability. An assessment of the performance of 20 models over south-
eastern Australia showed that 13 adequately reproduced observed average patterns of temperature,
rainfall and pressure. Ten of these were global climate models with a grid-spacing of 200-400 km
and monthly data, but only three had a grid-spacing of about 50 km and daily data. One of the 50
km simulations was based on a CSIRO model (DARLAM) that has been superseded, so the other
two 50 km simulations (CCAM) were used. CCAM is a global atmosphere-only model with fine
resolution over Australia that can be driven by boundary conditions from a global climate model
(including ocean, atmosphere, ice and land). One CCAM simulation was driven by CSIRO’s Mark2
global climate model and the other was driven by CSIRO’s Mark 3 global climate model,
henceforth called CCAM (Mark2) and CCAM (Mark3). Both perform well over south-east
Australia, although CCAM (Mark 2) has a better simulation of average temperature. On this basis,
slightly more confidence could be placed in results from CCAM (Mark2). Their climate projections
are considered independent. Regional climate change patterns from each model were scaled to
include the full range of IPCC SRES scenarios of greenhouse gas and aerosol emissions, and the
full range of IPCC uncertainty in climate sensitivity to these emissions (Appendix 1).
The climate change scenarios were applied to observed daily weather data at 17 sites. The FFDI and
GFDI results were calculated in three ways.
• Annual-average cumulative FFDI and GFDI, denoted ΣFFDI and ΣGFDI
• Monthly-average FFDI and GFDI
• Daily-average FFDI and GFDI
The “present” average ΣFFDI in inland areas is around 3000-5000, while southern and coastal areas
have values around 1700-2600. For CCAM (Mark2), the values rise by around 2-10% by 2020 and
5-25% by 2050. For CCAM (Mark3), the values rise by around 3-10% by 2020 and 8-30% by
2050.
Annual-average ∑FFDI at 17 sites for present (1974-2003) conditions, and percentage changes for
2020 and 2050, for low and high rates of global warming.
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low %
2020
high %
2050
low %
2050
high %
2020
low %
2020
high %
2050
low %
2050
high %
Canberra
2913 4 8 10 26 4 10 11 29
Bourke
5869 4 9 9 25 3 7 7 19
Cabramurra
501 5 10 10 26 7 14 15 40
Cobar
5818 4 10 10 26 3 8 8 22
Coffs Harbour
2002 2 5 5 12 3 6 6 15
Nowra
2507 1 4 4 13 2 6 6 18
Richmond
3049 4 8 8 20 4 8 8 21
Sydney
2158 2 4 5 12 3 7 7 19
Wagga
4047 4 8 9 23 4 9 9 25
Williamtown
2641 2 5 5 13 3 7 7 18
Bendigo
2854 3 8 8 22 3 8 8 23
Laverton
2456 3 8 8 21 4 9 9 24
Melbourne
2121 3 8 8 21 3 8 8 22
Mildura
5898 3 7 7 17 3 8 8 21
Sale
2207 3 8 8 21 4 8 8 23
Hobart
1723 0 0 0
-
1 0 1 1 2
Launceston
1677 1 3 3 8 3 6 6 17
7
The monthly-average FFDI results show that most sites currently have the highest fire danger in
spring and summer (blue curves in plot below). A spring peak is distinctive for coastal NSW sites,
whereas the summer peak is typical of southern and inland sites. In 2020 and 2050, the curves move
upward, indicating higher fire danger, particularly in spring, summer and autumn. Periods suitable
for prescribed (control) burning are likely to move toward winter.
Monthly-average FFDI at Melbourne for “now” (1974-2003), 2020 and 2050, based on the CCAM
(Mark3) climate change scenarios.
The daily-average frequency distributions of FFDI have five intensity categories: low (less than 5),
moderate (5-12), high (13-25), very high (25-49) and extreme (at least 50). At all sites, except
Hobart, Launceston and Cabramurra, there is an increase in the frequency of very high and extreme
days by 2020 and 2050. These are the categories of most interest to fire management agencies. By
2020, the combined frequencies of very high and extreme FFDI generally increase 4-20% for
CCAM (Mark2) and 6-25% for CCAM (Mark3). By 2050, the increases are generally 15-55% for
CCAM (Mark2) and 20-70% for CCAM (Mark3).
Average number of days when the FFDI rating is “very high” or “extreme” under present conditions
(1974-2003) for the years 2020 and 2050.
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low
2020
high
2050
low
2050
high
2020
low
2020
high
2050
low
2050
high
Canberra
23.1 25.6 27.5 27.9 36.0 26.0 28.6 28.9 38.3
Bourke
69.5 75.2 83.3 84.0 106.5 73.9 80.3 80.6 96.2
Cabramurra
0.3 0.3 0.4 0.4 0.7 0.4 0.4 0.5 1.0
Cobar
81.8 87.9 96.2 96.6 118.3 86.6 92.8 93.0 108.6
Coffs Harbour
4.4 4.7 5.1 5.1 6.3 4.7 5.6 5.6 7.6
Nowra
13.4 13.9 14.7 14.8 17.5 14.2 15.6 15.6 19.9
Richmond
11.5 12.9 14.0 14.1 17.5 13.1 14.3 14.4 19.1
Sydney
8.7 9.2 9.8 9.8 11.8 9.5 11.1 11.3 15.2
Wagga
49.6 52.7 57.3 57.6 71.5 52.8 57.4 57.7 71.9
Williamtown
16.4 17.2 18.2 18.4 20.9 17.3 19.4 19.4 23.6
Bendigo
17.8 19.5 21.3 21.4 27.3 19.7 21.9 22.0 29.8
Laverton
15.5 16.4 17.3 17.3 21.2 16.6 17.8 17.8 22.3
Melbourne
9.0 9.8 10.7 10.8 13.9 9.8 11.1 11.2 14.7
Mildura
79.5 83.9 89.5 89.9 104.8 84.6 90.7 90.9 107.3
Sale
8.7 9.3 10.0 10.1 12.1 9.6 10.7 10.8 14.0
Hobart
3.4 3.4 3.4 3.4 3.4 3.4 3.5 3.5 3.5
Launceston
1.5 1.5 1.5 1.6 2.0 1.6 1.9 1.9 3.1
8
Changes in the frequencies of extreme FFDI days are generally largest inland, e.g. at Bourke,
Cobar, Mildura and Wagga. By 2020, the increases are generally 10-30% for CCAM (Mark2) and
15-40% for CCAM (Mark3). By 2050, the increases are generally 20-80% for CCAM (Mark2) and
40-120% for CCAM (Mark3). At many sites, there is a doubling (or greater) of the number of
extreme days by 2050 for the high scenario. Tasmania is relatively unaffected. In Hobart, the rise in
temperature is offset by a rise in humidity.
The magnitude of the grassland fire danger index is always higher than the FFDI since the GFDI is
more strongly influenced by wind-speed and we have assumed a worst-case scenario of 100%
curing. By 2020, the number of very high or extreme GFDI days increases by around 0-15% for
CCAM (Mark2) and 5-20% for CCAM (Mark3). By 2050, the increases are generally 5-30% for
CCAM (Mark2) and 15-40% for CCAM (Mark3).
Average number of days when the GFDI rating is “very high” or “extreme” under present conditions
(1974-2003) and for the years 2020 and 2050.
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low
2020
high
2050
low
2050
high
2020
low
2020
high
2050
low
2050
high
Canberra
96.8 100.3 103.7 104.0 113.1 103.5 110.3 110.6 129.0
Bourke
90.6 97.5 102.9 103.3 117.9 97.7 102.7 103.0 117.0
Cabramurra
11.6 11.6 11.8 11.8 12.6 12.5 13.8 13.9 18.6
Cobar
112.8 124.1 129.0 129.4 146.6 124.0 129.5 130.1 148.1
Coffs Harbour
86.4 99.9 101.8 101.8 109.1 101.5 105.2 105.6 117.7
Nowra
71.7 80.3 81.7 81.8 86.3 83.5 88.5 88.9 104.0
Richmond
40.4 44.1 44.8 44.8 47.1 45.3 47.4 47.5 55.1
Sydney
116.2 117.6 120.0 120.1 126.8 122.1 129.3 129.7 153.5
Wagga
104.6 110.7 114.4 114.4 123.5 112.5 118.7 119.0 134.2
Williamtown
123.1 132.2 134.9 135.1 144.1 135.0 141.8 142.5 162.9
Bendigo
61.1 63.6 65.8 65.9 72.4 65.0 69.5 69.7 81.7
Laverton
110.1 109.4 111.7 111.9 118.6 111.8 117.4 117.9 131.7
Melbourne
38.7 41.2 41.2 42.2 45.7 42.3 45.0 45.2 54.5
Mildura
146.7 149.1 153.6 153.9 165.6 150.6 157.6 157.0 174.6
Sale
95.4 102.5 104.0 104.1 109.3 104.9 110.2 110.3 124.2
Hobart
67.5 67.5 67.2 67.2 66.1 68.1 68.8 69.0 71.5
Launceston
73.3 73.4 72.3 72.3 69.4 78.5 85.0 85.5 102.8
By 2020, the number of extreme GFDI days increases by around 5-20% for CCAM (Mark2) and
10-30% for CCAM (Mark3). By 2050, the increases are generally 10-30% for CCAM (Mark2) and
30-80% for CCAM (Mark3).
A number of uncertainties remain:
• Quality of observed daily wind and humidity data at most sites in Australia
• The effect of scenarios based on other climate models
• Assessment of changes in the range (and sensitivity) of seasonal indicators used by fire
management agencies for fire preparedness planning.
• Changes in rainfall thresholds required to control fires
• Changes in ignition (natural and anthropogenic)
• Changes in fuel load, allowing for carbon dioxide fertilization on vegetation
• Potential impacts on biodiversity, water yield and quality from fire affected catchments,
forestry, greenhouse gas emissions, emergency management and insurance.
9
Priorities for further research include:
• Testing and rehabilitation of observed humidity and wind data
• Creation of climate change scenarios from other models
• Fine scale fire modelling that captures vegetation and terrain features and fire management
• Hydrological and ecological modelling to assess impacts on water and biodiversity
• Using satellite remote sensing to monitor the extent and nature of fire, recovery of
vegetation after fire, and greenhouse gas emissions from fire.
10
1 Introduction
Since 1950, Australia has warmed by 0.85
o
C, rainfall has decreased in the south-east, droughts have
become hotter (Nicholls, 2003) and the number of extremely hot days has risen (Nicholls and
Collins, in press). The effect of these changes on fire frequency and intensity in the south-east is not
clearly evident, partly due to confounding factors such as fire management and arsonists. However,
it is clear that hotter and drier years have greater fire risk (BTE, 2001). Climate change projections
indicate that Victoria and NSW are likely to become hotter and drier in future (CSIRO, 2001;
Hennessy et al,, 2004; Suppiah et al, 2004), while Tasmania is likely to become warmer and wetter
(McInnes et al., 2004). The aim of this study is to assess potential changes in fire-weather risk
associated with future climate change, due to the enhanced greenhouse effect. It represents a
resource for ongoing engagement with fire management agencies to plan for the impacts of climate
change. However, the report is not intended to provide management recommendations to agencies.
Bushfires have been part of Australia’s environment for millions of years. Our natural ecosystems
have evolved with fire, and our landscapes and their biological diversity have been shaped by both
historical and recent patterns of fire (Cary, 2002). South-eastern Australia has highest risk in spring,
summer and autumn (Figure 1). This region has the reputation of being one of the three most fire-
prone areas in the world, along with southern California and southern France.
Figure 1: Seasonal pattern of fire danger. http://www.bom.gov.au/climate/c20thc/fire.shtml
Very little of the Australian continent is free from fires - scrub-fires may sweep even the arid
regions in years when good wet season rains are followed by a long dry spell. In the spring of 1974,
15 percent of the land area of Australia burned after prolific growth during the preceding wet
summer dried off and ignited. The Black Friday fires in Victoria (1939), the 1967 fires in Tasmania,
and the Ash Wednesday fires in Victoria and South Australia (1983) have each killed more than 60
people. Throughout the 20th Century, many other fires have claimed lives, destroyed people’s
homes and livelihoods, and reduced thousands of hectares of forest to charcoal and ash. From 1960-
2001, there were 224 fire-related deaths, 4505 injuries and $2475 million in damages (McMichael
et al., 2003). More than half the fire-related deaths, injuries and costs were in Victoria. The insured
costs of fire damage from 1967-2005 are shown in Table 1. Many of these fires occurred during
droughts associated with El Niño events. The most costly fires occurred in 1983 ($138 million) and
2003 ($342 million). Damage to plantation timber was a significant component of the costs in 1983.
There are also periods where Australia has not been affected by any large bushfires (e.g. 1970-76,
1998-2000), mainly due to the wetter La Niña conditions.
11
Table 1: Insured costs of fire damage from 1967-2005 (http://www.idro.com.au/disaster_list)
Date Location Original cost* $m
Feb 1967 Hobart TAS 14
Feb 1977 Western VIC 9
Feb 1980 Adelaide Hills SA 13
Feb 1983 VIC 138
Feb 1983 SA 38
Sep 1984 NSW 25
Feb 1987 Southern TAS 7
Jan 1990 VIC 10
Oct 1991 Central coast NSW 12
Jan 1994 Sydney NSW 59
Jan 1997 Ferny Creek VIC 10
Dec 1997 Sydney NSW 3
Dec 2001 Sydney NSW 69
Oct 2002 Sydney NSW 19
Jan 2003 Northeast VIC
Southeast NSW
12
Jan 2003 Canberra ACT 342
Jan 2005 Eyre Peninsula SA 27
* cost at time of event, not adjusted for inflation.
1.1 Previous assessments of climate change impacts on fire risk
Global warming is likely to increase fire frequency and severity. Various overseas studies have
shown this, e.g. Stocks et al (1998), Goldammer and Price (1998), Wotton et al (2003), Brown et al
(2004), Pearce et al (2005). In Australia, the McArthur Mark 5 Forest Fire Danger Index (FFDI) is
used operationally by weather forecasters and fire services throughout eastern Australia to
determine fire hazard and declare Total Fire Ban days. The FFDI has been closely related to the
probability of asset destruction in the Sydney region (Bradstock and Gill, 2001) (Figure 2). Beer et
al (1988) and Beer and Williams (1995) assessed the potential change in fire danger using the FFDI
and various climate change scenarios. Beer and Williams (1995) found that the annual accumulated
FFDI increased by at least 10% for a doubling of carbon dioxide concentration over most major
forest fire zones in southern and eastern Australia.
0
0.2
0.4
0.6
0.8
1
0 102030405060
Forest Fire Danger Index
Propotion of fires that are destructive
Figure 2: Trend in the proportion of unplanned fires in Sydney which resulted in house destruction
between 1954 and 1995, in relation to FFDI (Source: Bradstock and Gill, 2001).
12
Williams et al (2001) used simulated weather data from the CSIRO9 climate model for present and
enhanced greenhouse conditions (circa 2050) to assess changes in the daily FFDI at eight Australian
sites: Katanning (WA), Normanton (northern Qld), Miles (southeast Qld), Alice Springs (NT),
Hobart (Tas), Mildura and Sale (Vic). An increase in fire danger was simulated at all sites.
Cary (2002) used simulated weather data from the CSIRO regional climate model (DARLAM) for
present and enhanced greenhouse conditions over the ACT. Changes by the year 2070 were:
• 0.6 to 3.4
o
C warmer
• 1-2% lower relative humidity
• 0-25% less rain in Jan-Oct, 8-10% more rain in Nov-Dec
• No change in wind-speed.
These changes were applied to daily weather data and fed into the ANU FIRESCAPE model to
assess changes in the FFDI. The results were
• 5-20% increase in annual accumulated FFDI
• 12-70% decrease in years between fires at same location
• 7-25% increase in fire-line intensity.
FIRESCAPE is also being tested over southwest Tasmania, central Australia and the greater Sydney
basin. Since it runs at 1 km resolution, it requires detailed topographic and vegetation information,
and significant computer resources.
2 Methodology for an updated fire risk assessment in south-east Australia
There are various ways of using climate model information in fire risk models. These include:
1. Generate daily weather data using a climate change model for present and enhanced greenhouse
conditions, then use these data as input to a fire risk model (e.g. Beer and Williams, 1995;
Williams et al,, 2001). While this has the advantage of capturing changes in relationships
between weather variables, and changes in daily weather variability, it has the disadvantage of
being biased by errors in the simulated baseline climate, i.e. the simulation may be too
warm/cold or wet/dry on average. For example, Cary (2002) found that some errors in the
simulated baseline climate were as large as the changes in climate projected due to a doubling of
carbon dioxide. While corrections can be applied, residual errors remain (especially for extreme
fire danger events), and the spatial resolution is very coarse (about 300 km between data points).
2. Compute the changes in monthly average weather variables from a climate model, then apply
these changes to observed daily weather data, which are then input to a fire risk model (e.g.
Cary, 2002). This has the advantage of avoiding biases that exist in the simulation of baseline
conditions, while having the disadvantages of assuming (i) that existing relationships between
weather variables will be maintained in future, and (ii) there will be no change in daily weather
variability. These disadvantages are generally not considered serious. Another disadvantage is
the limited availability of sites with daily records of the four key weather variables, i.e.
temperature, rainfall, relative humidity and wind-speed.
3. Compute changes in daily weather variability from a climate model, then apply these changes to
observed daily weather data. This avoids biases in the simulated baseline climate and avoids
disadvantages (i) and (ii). Including changes in daily weather variability and the behaviour of
extreme events are obviously important for fire-weather risk.
Method 3 was used in the present study. It is unique in Australian fire risk assessments. There were
four main steps, as described below.
13
2.1 Select sites with high quality observed daily weather data
The weather variables required for this analysis were daily maximum temperature, precipitation,
minimum (3 pm) relative humidity and maximum wind-speed. A 30-year period centred on 1990
was needed since climate change projections are relative to 1990. The Bureau of Meteorology has
developed high quality data sets for daily temperature and rainfall (Haylock and Nicholls, 2000;
Collins et al., 2000). Creation of a high quality dataset for humidity is being undertaken by the
Bureau of Meteorology, sponsored by the Bushfire CRC. Preliminary humidity data for selected
sites were made available for the present study. The quality of the humidity data is acceptable at
most sites. However, at Cabramurra, there is only one humidity observation per day (at 9 am) in
much of the record, and since humidity is quite high early in the morning, this may underestimate
fire danger results at that site. Due to resource limitations, wind-speed data have not been
homogenized, so there are some data problems due to changes in instrumentation and observer
practices. While the wind data are usable, they may introduce errors within the analysis. For
example, at Richmond, some wind data are missing in the 1970s. Also, at Melbourne, the wind-
speed dropped after 2000 due to a shift in instrumentation. However, temperature and humidity are
the most important drivers of fire-weather (Beer and Williams, 1995; Williams et al., 2001).
For the period 1974-2003, data for all four weather variables were available at 17 sites (Figure 3):
• NSW: Coffs Harbour, Cobar, Williamtown, Richmond, Sydney, Nowra, Wagga, Bourke,
Cabramurra
• Victoria: Mildura, Melbourne, Laverton, Sale, Bendigo
• ACT: Canberra
• Tasmania: Hobart, Launceston
Figure 3: Locations of the 17 sites used in this study.
2.2 Derive daily soil moisture, drought and fire danger indicators for present conditions
At each site, observed weather data were used to compute daily values of the Mount Soil Dryness
Index (SDI), Keetch Byram Drought Index (KBDI), FFDI and Grassland Fire Danger Index
(GFDI). The average number of days exceeding low, medium, high, very high and extreme levels
was calculated.
In preparation for the climate change scenarios, observed frequency distributions of the four
weather variables were computed for each of the 12 calendar months. For each day in the 30 years,
the weather variables were assigned to one of ten deciles by comparing the value for that day with
the frequency distribution for that month.
14
Deciles are defined as
• Decile 1: the lowest 10% of values
• Decile 2: values in the lowest 10-20%
• Decile 3: values in the lowest 20-30%
• Decile 4: values in the lowest 30-40%
• Decile 5: values in the lowest 40-50%
• Decile 6: values in the highest 40-50%
• Decile 7: values in the highest 30-40%
• Decile 8: values in the highest 20-30%
• Decile 9: values in the highest 10-20%
• Decile 10: the highest 10% of values.
In previous fire risk assessments, where changes in mean monthly climate have been applied to
observed daily weather data (e.g. Cary, 2002), each decile has been changed by the same amount.
This assumes no change in future weather variability, so the shape of each monthly frequency
distribution remains unchanged while the mean increases or decreases (Figure 4a).
In this study, each decile has been changed by different amounts, according to changes in the mean
and variability simulated by two climate models, so the shape of each monthly frequency
distribution changes (Figure 4c). This improves the reliability of fire-weather projections since
changes in extremely high temperature and wind-speed, and extremely low humidity, are critical for
fire risk.
Figure 4: The effects on extreme temperatures when (a) the mean increases with no change in
variance, (b) the variance increases with no change in the mean, and (c) when both the mean and
variance increase, leading to more record hot weather. Source IPCC (2001).
15
2.3 Create climate change scenarios for each decile
The choice of climate simulations for this study was constrained by a number of factors: (1) models
that perform well over south-eastern Australia, (2) availability of simulated data with grid-spacing
of 50 km or less, and (3) availability of simulated daily weather data from which to compute
changes in daily variability. An assessment of the performance of 20 climate models over south-
eastern Australia showed that 13 adequately reproduced observed average patterns of temperature,
rainfall and pressure (McInnes et al., 2005). Ten of these were global climate models with a grid-
spacing of 200-400 km and monthly data, but only three had a grid-spacing of about 50 km and
daily data. One of the 50 km simulations was based on a CSIRO model (DARLAM) that has been
superseded, so the other two 50 km simulations (CCAM) were used.
CCAM is a global atmosphere-only model, developed by CSIRO, that can be driven by boundary
conditions from a global climate model (including ocean, atmosphere, ice and land) (McGregor and
Dix, 2001). At 50 km resolution, CCAM has a better representation of climate and topographic
processes than most global climate models. One CCAM simulation was driven by CSIRO’s Mark2
climate model and the other was driven by CSIRO’s Mark 3 climate model, henceforth called
CCAM (Mark2) and CCAM (Mark3). Both perform well over south-east Australia, although
CCAM (Mark 2) has a better simulation of average temperature. Hence, slightly more confidence
can be placed in results from CCAM (Mark2). The ability of the models to reproduce observed
wind and humidity has not been tested, although validation of the pressure patterns implies that
wind patterns are well simulated. Their climate projections are considered independent since the
Mark 2 and Mark 3 models have different parameterisations of physical processes. Regional climate
change patterns from each model were expressed as a change per degree of global warming. This
allows the results to be linearly scaled for any future year using the IPCC (2001) global warming
estimates (Mitchell, 2003), which include the full range of IPCC SRES (2000) scenarios of
greenhouse gas and aerosol emissions, and the full range of IPCC (2001) uncertainty in climate
sensitivity to these emissions (Whetton, 2001; Appendix 1).
At each of the 17 sites in both simulations,
• “Present” deciles were calculated for daily maximum temperature, rainfall, wind-speed and
relative humidity, for each calendar month in a 30-year period centred on 1990.
• For each year in the period 1962-2100, deciles were calculated for each climate variable for
each month, using a 3-year window centred on the year of interest
• For each year in the period 1962-2100, the change in each decile relative to the “present”
values was calculated for each month.
• For each year in the period 1962-2100, the annual global mean warming was calculated.
• We assumed that there is a linear relationship between annual global mean warming and
regional climate change (Whetton, 2001; Mitchell, 2003; Whetton et al, in prep.). For each
year in the period 1962-2100, the regional decile changes were regressed against the global
warming values. This gave a decile change per degree of global warming for each variable.
• Regional projections are presented as low-high ranges (probabilities are not available). The
low regional projection is based on a low global warming projection (low emission scenario
with low climate sensitivity), while the high regional projection is based on a high global
warming projection (high emission scenario with high climate sensitivity) – see Appendix 1.
Regional projections for 2020 and 2050 were computed by multiplying the regional decile
changes per degree of global warming by the IPCC low-high global warming values for the
years 2020 and 2050, namely 0.37-0.85
o
C by 2020 and 0.88-2.24
o
C by 2050.
Rather than showing climate change scenarios for each site, Figure 5 shows examples of monthly
climate change scenarios for the four capital cities (Canberra, Sydney, Melbourne and Hobart).
16
JFMAMJJASOND
Max. temp. change (
o
C)
0
1
2
3
4
5
6
Canberra
JFMAMJJASOND
Precipitation change (%)
-40
-20
0
20
40
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-40
-35
-30
-25
-20
-15
-10
-5
0
5
10
Mean 2020 low
Mean 2020 high
Mean 2050 low
Mean 2050 high
Extreme 2020 low
Extreme 2020 high
Extreme 2050 low
Extreme 2050 high
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
JFMAMJJASOND
Max. temp. change (
o
C)
0
1
2
3
4
5
6
JFMAMJJASOND
Precipitation change (%)
-40
-20
0
20
40
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-40
-35
-30
-25
-20
-15
-10
-5
0
5
10
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
Figure 5a: Climate change scenarios for Canberra in 2020 and 2050, relative to 1990. Mean changes
in maximum temperature, rainfall, relative humidity and wind-speed summarise the changes across
deciles 1 to 10. Extreme changes represent decile 10 for maximum temperature and wind-speed, and
decile 1 for relative humidity.
17
JFMAMJJASOND
Max. temp. change (
o
C)
0
1
2
3
4
5
Melbourne
JFMAMJJASOND
Precipitation change (%)
-30
-20
-10
0
10
20
30
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-30
-25
-20
-15
-10
-5
0
5
10
15
Mean 2020 low
Mean 2020 high
Mean 2050 low
Mean 2050 high
Extreme 2020 low
Extreme 2020 high
Extreme 2050 low
Extreme 2050 high
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
JFMAMJJASOND
Max. temp. change (
o
C)
0
1
2
3
4
5
JFMAMJJASOND
Precipitation change (%)
-30
-20
-10
0
10
20
30
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-30
-25
-20
-15
-10
-5
0
5
10
15
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
Figure 5b: Climate change scenarios for Melbourne in 2020 and 2050, relative to 1990. Mean changes
in maximum temperature, rainfall, relative humidity and wind-speed summarise the changes across
deciles 1 to 10. Extreme changes represent decile 10 for maximum temperature and wind-speed, and
decile 1 for relative humidity.
18
JFMAMJJASOND
Max. temp. change (
o
C)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Sydney
JFMAMJJASOND
Precipitation change (%)
-20
0
20
40
60
80
100
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-30
-25
-20
-15
-10
-5
0
5
10
Mean 2020 low
Mean 2020 high
Mean 2050 low
Mean 2050 high
Extreme 2020 low
Extreme 2020 high
Extreme 2050 low
Extreme 2050 high
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
JFMAMJJASOND
Max. temp. change (
o
C)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
JFMAMJJASOND
Precipitation change (%)
-20
0
20
40
60
80
100
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-30
-25
-20
-15
-10
-5
0
5
10
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
Figure 5c: Climate change scenarios for Sydney in 2020 and 2050, relative to 1990. Mean changes in
maximum temperature, rainfall, relative humidity and wind-speed summarise the changes across
deciles 1 to 10. Extreme changes represent decile 10 for maximum temperature and wind-speed, and
decile 1 for relative humidity.
19
JFMAMJJASOND
Max. temp. change (
o
C)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Hobart
JFMAMJJASOND
Precipitation change (%)
-20
0
20
40
60
80
100
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-30
-25
-20
-15
-10
-5
0
5
10
Mean 2020 low
Mean 2020 high
Mean 2050 low
Mean 2050 high
Extreme 2020 low
Extreme 2020 high
Extreme 2050 low
Extreme 2050 high
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
CCAM (Mark2)
JFMAMJJASOND
Max. temp. change (
o
C)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
JFMAMJJASOND
Precipitation change (%)
-20
0
20
40
60
80
100
JFMAMJJASOND
Windspeed change (%)
-8
-4
0
4
8
12
Month
JFMAMJJASOND
Relative humidity change (%)
-30
-25
-20
-15
-10
-5
0
5
10
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
CCAM (Mark3)
Figure 5d: Climate change scenarios for Hobart in 2020 and 2050, relative to 1990. Mean changes in
maximum temperature, rainfall, relative humidity and wind-speed summarise the changes across
deciles 1 to 10. Extreme changes represent decile 10 for maximum temperature and wind-speed, and
decile 1 for relative humidity.
20
Some brief observations can be made from the figures above:
• The mean warming is around 0.5-1.5
o
C by 2020 and around 1.5-3.0
o
C by 2050
• Mean rainfall tends to increase in autumn and decrease in spring-summer in Canberra and
Melbourne
• Mean rainfall tends to increase in autumn-winter in Sydney and Hobart in CCAM (Mark3)
but decrease in CCAM (Mark2)
• Mean wind-speed tends to decrease throughout the year at each site in CCAM (Mark2) but
increase in CCAM (Mark3)
• Mean humidity tends to decrease all year except around March at each site in CCAM
(Mark2), while increases are more prevalent in CCAM (Mark 3) in Hobart, Melbourne and
Sydney
The mean results represent changes spread across all ten deciles, while the extreme changes
represent changes in decile ten (very high) for temperature and wind-speed, and decile 1 (very low)
for relative humidity. The mean and extreme changes can be significantly different for some cases.
For example, the spring-summer increases in extremely high temperature are larger than the
increases in mean temperature. Similarly, the decreases in extremely low humidity are larger than
the decreases in mean humidity.
2.4 Apply the climate change scenarios to observed weather data
Changes in each monthly decile were applied to the observed daily weather data, thus creating
“new” 30-year weather data centred on 2020 (low and high global warming) and 2050 (low and
high global warming). For example, if the maximum temperature on 19 January 2003 was 27.5
o
C,
and this was in decile 7 of January observed maximum temperatures, and the CCAM (Mark2)
decile 7 high scenario for January in 2050 was a warming of 2.3
o
C, then the "new" temperature for
a simulated 19 January would be 29.8
o
C. The “new” weather data were then used in calculations of
SDI, KBDI, FFDI and GFDI for 2020 and 2050.
3. Results
McArthur Mark 5 Forest Fire Danger Index (FFDI; Noble et al,, 1980) is defined as:
FFDI = 2exp(0.987logD – 0.45 + 0.0338T + 0.0234V – 0.0345H)
where: H = relative humidity from 0-100%
T = air temperature in degrees Celsius
V = average wind-speed 10 metres above the ground, in metres per second
D = drought factor in the range 0-10
The drought factor, D, can be defined in different ways, e.g. the Keetch Byram Drought Index
(KBDI; Keetch and Byram, 1968) and the Mount Soil Dryness Index (SDI; Mount, 1972), The
KBDI is a function of days since last rain, assuming a 200 mm soil moisture capacity. Mount
(1972) modified the KBDI on the basis of Tasmanian experience and developed the SDI (Beer et
al,, 1988). In the present study, we have used the Griffiths (1999) drought factor because this is
standard practice in the Bureau of Meteorology. It uses the KBDI and includes the effect of
evapotranspiration. The difference between SDI and KBDI lies in the way evapotranspiration is
handled. KBDI uses an exponential relationship, while SDI uses a linear regression. Both have their
shortcomings, but SDI is probably most applicable to Tasmania, where it was developed, and least
applicable in inland NSW. SDI is almost always higher than KBDI, and results in slightly higher
FFDI values. Although we use Griffiths drought factor, results are presented in the appendices for
SDI and KBDI since these were required by the sponsors of this study.
21
The McArthur Mark 4 Grassland Fire Danger Index (GFDI; Purton, 1982) is defined as:
GFDI=10
x
where x = (-0.6615+1.027log10(Q)-0.004096(100-C)
1.536
+0.01201T+0.2789√V-0.9577√RH)
and Q is fuel quantity (t/ha)...[we assume a standard 4.5 t/ha]
C is curing factor (0-100%) [we assume 100% fully cured]
T is temperature (Celsius)
V is wind speed (km/hr)
RH is relative humidity (%)
The degree of grassland curing refers to the proportion of cured and/or dead material in a grassland
fuel complex, and has a significant effect on fire behaviour, in particular potential fire spread
(Anderson and Pearce, 2003). The GFDI results are sensitive to the curing factor, as shown in
Figure 6. Our assumption of 100% curing represents a worst-case scenario. An assumption of 80%
curing could reduce the GFDI results by about 60%. In the absence of a pasture growth model, the
choice of curing factor is arbitrary. It is likely that the 100% assumption overestimates GFDI in
winter/spring and has greater accuracy in summer/autumn. A change in climate could lead to earlier
and /or more efficient curing, or to a change in vegetation type which has different curing
properties. Hence, the GFDI results should be treated with caution.
0 20 40 60 80
100
Curing factor
0
0.2
Fraction of GFDI for 100% curing
0.4
0.6
0.8
1.0
Figure 6: Relationship between GFDI and curing factor.
3.1 Drought factors: SDI and KBDI
Appendices 1 and 2 show frequency distributions of SDI and KBDI at each of the 17 sites, for
present, 2020 and 2050 conditions. In future, there is a strong tendency toward higher values. In
some cases, the frequency of reaching 150-200 (very dry) doubles by 2050 in the high scenarios.
3.2 Forest fire danger index
The FFDI results were calculated in three ways.
• Annual-average cumulative FFDI, denoted ΣFFDI
• Monthly-average FFDI
• Daily-average FFDI.
Table 1 shows the “present” average ΣFFDI at each site. Inland areas have values around 3000-
5000, while southern and coastal areas have values around 1700-2600. Site selection was
22
constrained by availability of high quality data, so the sites used are not always representative of
large areas. For example, Melbourne ΣFFDI values are less than those at nearby Laverton because
the former are based on data from a highly urbanised area with lower wind-speeds and higher
rainfall. Hence Melbourne values are more representative of inner suburbs while Laverton values
are more representative of outer Melbourne suburbs. Sydney values are less than those at nearby
Williamtown because the former are based on data measured near Botany Bay with higher
humidity. Richmond values are representative of outer Sydney suburbs and the Blue Mountains.
All sites, except Hobart, show an increase in annual ΣFFDI in both 2020 and 2050. For CCAM
(Mark2), the increases are generally 2-10% by 2020 and 5-25% by 2050. For CCAM (Mark3), the
increases are slightly greater: generally 3-10% by 2020 and 8-30% by 2050. Hobart shows
negligible change since small increases in temperature are offset by increases in humidity.
However, ΣFFDI is a fairly conservative measure of fire risk because it hides information about
monthly and daily extremes. Therefore, monthly and daily FFDI values were examined.
Table 1: Annual-average ∑FFDI at 17 sites for present (1974-2003) conditions, and percentage
changes for 2020 and 2050, for low and high rates of global warming, in CCAM (Mark2) and CCAM
(Mark3).
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low %
2020
high %
2050
low %
2050
high %
2020
low %
2020
high %
2050
low %
2050
high %
Canberra
2913 4 8 10 26 4 10 11 29
Bourke
5869 4 9 9 25 3 7 7 19
Cabramurra
501 5 10 10 26 7 14 15 40
Cobar
5818 4 10 10 26 3 8 8 22
Coffs Harbour
2002 2 5 5 12 3 6 6 15
Nowra
2507 1 4 4 13 2 6 6 18
Richmond
3049 4 8 8 20 4 8 8 21
Sydney
2158 2 4 5 12 3 7 7 19
Wagga
4047 4 8 9 23 4 9 9 25
Williamtown
2641 2 5 5 13 3 7 7 18
Bendigo
2854 3 8 8 22 3 8 8 23
Laverton
2456 3 8 8 21 4 9 9 24
Melbourne
2121 3 8 8 21 3 8 8 22
Mildura
5898 3 7 7 17 3 8 8 21
Sale
2207 3 8 8 21 4 8 8 23
Hobart
1723 0 0 0
-
1 0 1 1 2
Launceston
1677 1 3 3 8 3 6 6 17
Appendix 3 shows monthly-average FFDI and daily-average frequency distributions of FFDI at
each site. Most sites currently have highest fire danger in spring and summer (blue curves). The
spring peak is distinctive for coastal NSW sites, whereas the summer peak is typical of southern and
inland sites (c.f. Figure 1). In 2020 and 2050, the curves move upward, indicating higher fire
danger, particularly in spring, summer and autumn. At NSW coastal stations, the largest increases
occur in spring. Periods suitable for prescribed (control) burning are likely to move toward winter.
The daily-average frequency distributions of FFDI have five intensity categories: low (less than 5),
moderate (5-12), high (13-25), very high (25-49) and extreme (at least 50). At all sites, except
Hobart, Launceston and Cabramurra, there is an increase in the frequency of very high and extreme
days by 2020 and 2050. These are the two categories of most interest to fire management agencies.
By 2020, the combined frequencies of very high and extreme FFDI (Table 2) generally rise 4-20%
for CCAM (Mark2) and 6-25% for CCAM (Mark3). By 2050, the increases are generally 15-55%
for CCAM (Mark2) and 20-70% for CCAM (Mark3).
23
Changes in the frequencies of extreme FFDI days (Table 3) are largest inland, e.g. at Bourke,
Cobar, Mildura and Wagga. By 2020, the increases are generally 10-30% for CCAM (Mark2) and
15-40% for CCAM (Mark3). By 2050, the increases are generally 20-80% for CCAM (Mark2) and
40-120% for CCAM (Mark3). At many sites, there is a doubling (or greater) of the number of
extreme days by 2050 for the high scenario. Tasmania is relatively unaffected. In Hobart, the rise in
temperature is offset by a rise in humidity.
Table 2: Average number of days when the FFDI rating is “very high” or “extreme” under present
conditions (1974-2003). Results are also shown for the years 2020 and 2050, for two climate models
(CCAM Mark2 and CCAM Mark3), and two rates of global warming (low and high).
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low
2020
high
2050
low
2050
high
2020
low
2020
high
2050
low
2050
high
Canberra
23.1 25.6 27.5 27.9 36.0 26.0 28.6 28.9 38.3
Bourke
69.5 75.2 83.3 84.0 106.5 73.9 80.3 80.6 96.2
Cabramurra
0.3 0.3 0.4 0.4 0.7 0.4 0.4 0.5 1.0
Cobar
81.8 87.9 96.2 96.6 118.3 86.6 92.8 93.0 108.6
Coffs Harbour
4.4 4.7 5.1 5.1 6.3 4.7 5.6 5.6 7.6
Nowra
13.4 13.9 14.7 14.8 17.5 14.2 15.6 15.6 19.9
Richmond
11.5 12.9 14.0 14.1 17.5 13.1 14.3 14.4 19.1
Sydney
8.7 9.2 9.8 9.8 11.8 9.5 11.1 11.3 15.2
Wagga
49.6 52.7 57.3 57.6 71.5 52.8 57.4 57.7 71.9
Williamtown
16.4 17.2 18.2 18.4 20.9 17.3 19.4 19.4 23.6
Bendigo
17.8 19.5 21.3 21.4 27.3 19.7 21.9 22.0 29.8
Laverton
15.5 16.4 17.3 17.3 21.2 16.6 17.8 17.8 22.3
Melbourne
9.0 9.8 10.7 10.8 13.9 9.8 11.1 11.2 14.7
Mildura
79.5 83.9 89.5 89.9 104.8 84.6 90.7 90.9 107.3
Sale
8.7 9.3 10.0 10.1 12.1 9.6 10.7 10.8 14.0
Hobart
3.4 3.4 3.4 3.4 3.4 3.4 3.5 3.5 3.5
Launceston
1.5 1.5 1.5 1.6 2.0 1.6 1.9 1.9 3.1
Table 3: Average number of days when the FFDI rating is “extreme” under present conditions (1974-
2003). Results are also shown for the years 2020 and 2050, for two climate models (CCAM Mark2 and
CCAM Mark3), and two rates of global warming (low and high).
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low
2020
high
2050
low
2050
high
2020
low
2020
high
2050
low
2050
high
Canberra
2.2 2.6 3.0 3.0 4.7 2.5 3.5 3.5 5.7
Bourke
6.4 7.6 8.8 8.9 14.4 7.5 8.8 8.9 14.2
Cabramurra
0 0.0 0.1 0.1 0.1 0.0 0.0 0.0 0.1
Cobar
8.5 9.8 12.2 12.4 20.5 9.7 12.2 12.3 19.9
Coffs Harbour
0.5 0.6 0.6 0.6 0.7 0.6 0.7 0.7 0.9
Nowra
2.5 2.7 3.1 3.1 3.5 3.1 3.4 3.4 5.3
Richmond
1.5 1.7 1.9 1.9 2.4 1.8 2.0 2.0 3.2
Sydney
1 1.0 1.1 1.1 1.4 1.2 1.4 1.4 2.5
Wagga
6.3 7.0 8.0 8.2 13.2 6.9 8.3 8.4 14.5
Williamtown
2.8 3.1 3.3 3.3 4.2 3.3 3.7 3.7 5.5
Bendigo
1.6 1.9 2.2 2.2 3.2 2.0 2.3 2.3 3.9
Laverton
3.4 3.6 4.1 4.1 5.0 3.7 4.4 4.5 6.0
Melbourne
0.6 0.7 0.8 0.8 1.5 0.7 0.9 0.9 1.9
Mildura
10.4 11.2 12.7 12.8 16.9 11.7 13.5 13.6 20.1
Sale
1.1 1.2 1.5 1.5 2.1 1.3 1.7 1.7 2.6
Hobart
0.3 0.3 0.3 0.3 0.2 0.3 0.3 0.3 0.3
Launceston
0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1
24
3.3 Grassland fire danger index
Appendix 4 shows monthly-average GFDI and daily-average frequency distributions of GFDI at
each site. The GFDI has five intensity categories: low (less than 2.5), moderate (2.5-7.5), high (7.5-
20), very high (20-50) and extreme (50-200) (Cheney, 1997). Most sites currently have highest fire
danger in spring and summer. The magnitude of the GFDI is always higher than the FFDI since the
GFDI is more strongly influenced by wind-speed and we have assumed a worst-case scenario of
100% curing. In 2020 and 2050, the GFDI curves move upward, indicating higher grassfire danger,
particularly in spring, summer and autumn. There is an increase in the frequency of very high and
extreme days by 2020 and 2050 (Table 4). By 2020, the increases are generally 0-15% for CCAM
(Mark2) and 5-20% for CCAM (Mark3). By 2050, the increases are generally 5-30% for CCAM
(Mark2) and 15-40% for CCAM (Mark3).
Changes in extreme days are generally largest inland, e.g. at Bourke, Cobar, Mildura and Wagga
(Table 5). By 2020, the increases are generally 5-20% for CCAM (Mark2) and 10-30% for CCAM
(Mark3). By 2050, the increases are generally 10-30% for CCAM (Mark2) and 30-80% for CCAM
(Mark3). At many sites, there is a doubling (or greater) of the number of days classified as extreme
by 2050 for the high scenario.
Table 4: Average number of days when the GFDI rating is “very high” or “extreme” under present
conditions (1974-2003). Results are also shown for the years 2020 and 2050, for two climate models
(CCAM Mark2 and CCAM Mark3), and two rates of global warming (low and high).
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low
2020
high
2050
low
2050
high
2020
low
2020
high
2050
low
2050
high
Canberra
96.8 100.3 103.7 104.0 113.1 103.5 110.3 110.6 129.0
Bourke
90.6 97.5 102.9 103.3 117.9 97.7 102.7 103.0 117.0
Cabramurra
11.6 11.6 11.8 11.8 12.6 12.5 13.8 13.9 18.6
Cobar
112.8 124.1 129.0 129.4 146.6 124.0 129.5 130.1 148.1
Coffs Harbour
86.4 99.9 101.8 101.8 109.1 101.5 105.2 105.6 117.7
Nowra
71.7 80.3 81.7 81.8 86.3 83.5 88.5 88.9 104.0
Richmond
40.4 44.1 44.8 44.8 47.1 45.3 47.4 47.5 55.1
Sydney
116.2 117.6 120.0 120.1 126.8 122.1 129.3 129.7 153.5
Wagga
104.6 110.7 114.4 114.4 123.5 112.5 118.7 119.0 134.2
Williamtown
123.1 132.2 134.9 135.1 144.1 135.0 141.8 142.5 162.9
Bendigo
61.1 63.6 65.8 65.9 72.4 65.0 69.5 69.7 81.7
Laverton
110.1 109.4 111.7 111.9 118.6 111.8 117.4 117.9 131.7
Melbourne
38.7 41.2 41.2 42.2 45.7 42.3 45.0 45.2 54.5
Mildura
146.7 149.1 153.6 153.9 165.6 150.6 157.6 157.0 174.6
Sale
95.4 102.5 104.0 104.1 109.3 104.9 110.2 110.3 124.2
Hobart
67.5 67.5 67.2 67.2 66.1 68.1 68.8 69.0 71.5
Launceston
73.3 73.4 72.3 72.3 69.4 78.5 85.0 85.5 102.8
25
Table 5: Average number of days when the GFDI rating is “extreme” under present conditions (1974-
2003). Results are also shown for the years 2020 and 2050, for two climate models (CCAM Mark2 and
CCAM Mark3), and two rates of global warming (low and high).
Site Present CCAM (Mark2) CCAM (Mark3)
2020
low
2020
high
2050
low
2050
high
2020
low
2020
high
2050
low
2050
high
Canberra
18.8 19.3 20.2 20.3 23.7 20.8 23.6 23.9 32.8
Bourke
17.2 19.4 21.9 22.2 29.0 19.6 22.6 22.8 30.6
Cabramurra
0.8 0.8 0.8 0.8 0.9 0.8 1.0 1.0 1.8
Cobar
21.8 26.2 28.6 28.7 36.7 26.8 29.8 30.0 40.6
Coffs Harbour
7.8 10.6 10.9 10.9 11.9 11.1 12.8 12.8 17.8
Nowra
18.3 20.9 21.1 21.1 21.8 21.8 23.5 23.6 29.8
Richmond
7.5 8.7 8.8 8.9 9.6 9.2 10.0 10.0 13.5
Sydney
19.8 20.0 20.4 20.4 21.5 21.6 24.9 25.3 34.4
Wagga
25 27.7 29.8 30.0 36.4 28.5 31.4 31.6 42.7
Williamtown
27.9 30.4 30.7 30.8 32.4 31.5 34.6 34.7 43.6
Bendigo
9.5 9.9 10.4 10.4 11.9 10.4 11.8 11.9 16.3
Laverton
29.6 29.6 30.2 30.2 31.6 30.5 32.2 32.4 38.9
Melbourne
5.8 6.3 6.4 6.4 7.8 6.6 7.6 7.6 10.2
Mildura
30.1 31.0 32.9 33.0 39.0 32.4 36.1 36.4 48.3
Sale
17.3 19.2 19.8 19.8 21.2 20.7 23.4 23.6 30.9
Hobart
9.1 9.0 8.9 8.9 8.7 9.3 9.4 9.4 10.0
Launceston
4.6 4.6 4.3 4.3 3.7 5.3 6.0 6.1 9.6
3.4 Synthesis of results for each site
This section summarises FFDI and GFDI results for each site, based on climate change simulations
from two climate models (CCAM Mark2 and CCAM Mark3). Results from other models may be
slightly different, but quantification of that uncertainty is beyond the scope of this report. The word
“could “ is used below to emphasize that the projections are not predictions. The monthly-average
FFDI threshold selected for each site is arbitrary, and is not intended to define a fire season. It
simply provides a reference period between spring and autumn against which changes in the
duration of fire-weather risk can be compared.
Canberra
The average annual accumulated FFDI is currently 2913. This could increase 4-10% by 2020 and
10-29% by 2050. The monthly-average FFDI currently exceeds 10 from mid-November to mid-
March. This could extend from early November to mid-March by 2020, and from mid-October to
early April by 2050. On average, there are currently 23.1 days when the FFDI rating is very high or
extreme. This could increase to 25.6-28.6 days by 2020 and 27.9-38.3 days by 2050. There are
currently 96.8 days when the GFDI rating is very high or extreme. This could increase to 100.3-
110.3 days by 2020 and 104.0-129.0 days by 2050.
Bourke
The average annual accumulated FFDI is currently 5869. This could increase 4-9% by 2020 and
9-25% by 2050. The monthly-average FFDI currently exceeds 20 from late October to late
February. This could extend from early October to mid-March by 2020, and from early September
to late March by 2050. On average, there are currently 69.5 days when the FFDI rating is very high
or extreme. This could increase to 73.9-83.3 days by 2020 and 80.6-106.5 days by 2050. There are
currently 90.6 days when the GFDI rating is very high or extreme. This could increase to 97.5-102.9
days by 2020 and 103.0-117.9 days by 2050.
26
Cabramurra
The average annual accumulated FFDI is currently 501. This could increase 5-14% by 2020 and
10-40% by 2050. The monthly-average FFDI currently exceeds 2 from early December to mid-
March. This could extend from mid November to mid-March by 2020, and from mid-October to
early April by 2050. On average, there are currently 0.3 days when the FFDI rating is very high or
extreme. This could increase to 0.3-0.4 days by 2020 and 0.4-1.0 days by 2050. There are currently
11.6 days when the GFDI rating is very high or extreme. This could increase to 11.6-13.8 days by
2020 and 11.8-18.6 days by 2050.
Cobar
The average annual accumulated FFDI is currently 5818. This could increase 4-10% by 2020 and
10-26% by 2050. The monthly-average FFDI currently exceeds 20 from early November to mid-
March. This could extend from mid-October to mid-March by 2020, and from mid-September to
late March by 2050. On average, there are currently 81.8 days when the FFDI rating is very high or
extreme. This could increase to 86.6-96.2 days by 2020 and 93.0-118.3 days by 2050. There are
currently 112.8 days when the GFDI rating is very high or extreme. This could increase to 124.0-
129.5 days by 2020 and 129.4-148.1 days by 2050.
Coffs Harbour
The average annual accumulated FFDI is currently 2002. This could increase 2-6% by 2020 and
5-15% by 2050. The monthly-average FFDI currently exceeds 8 from mid-August to mid-October.
This could extend one week earlier (CCAM Mark2) or two weeks later (CCAM Mark3) by 2020,
and two weeks earlier (CCAM Mark2) or three weeks later (CCAM Mark3) by 2050. On average,
there are currently 4.4 days when the FFDI is very high or extreme. This could increase to 4.7-5.6
days by 2020 and 5.1-7.6 days by 2050. There are currently 86.4 days when the GFDI rating is very
high or extreme. This could increase to 99.9-105.2 days by 2020 and 101.8-117.7 days by 2050.
Nowra
The average annual accumulated FFDI is currently 2507. This could increase 1-6% by 2020 and
4-18% by 2050. The monthly-average FFDI currently exceeds 8 from early September to mid-
January. This could extend from late August to mid-January by 2020, and from early August to
early February by 2050. On average, there are currently 13.4 days when the FFDI rating is very
high or extreme. This could increase to 13.9-15.6 days by 2020 and 14.8-19.9 days by 2050. There
are currently 71.7 days when the GFDI rating is very high or extreme. This could increase to 80.3-
88.5 days by 2020 and 81.8-104.0 days by 2050.
Richmond
The average annual accumulated FFDI is currently 3049. This could increase 4-8% by 2020 and
8-21% by 2050. The monthly-average FFDI currently exceeds 10 from early September to mid-
January. This could extend from mid-August to late January by 2020, and from early August to
early February by 2050. On average, there are currently 11.5 days when the FFDI rating is very
high or extreme. This could increase to 12.9-14.3 days by 2020 and 14.1-19.1 days by 2050. There
are currently 40.4 days when the GFDI rating is very high or extreme. This could increase to 44.1-
47.4 days by 2020 and 44.8-55.1 days by 2050.
27
Sydney
The average annual accumulated FFDI is currently 2158. This could increase 2-7% by 2020 and
5-19% by 2050. The monthly-average FFDI currently exceeds 6 from early August to late
December. This could extend from early August to mid-January by 2020, and from late July to mid-
February by 2050. On average, there are currently 8.7 days when the FFDI rating is very high or
extreme. This could increase to 9.2-11.1 days by 2020 and 9.8-15.2 days by 2050. There are
currently 116.2 days when the GFDI rating is very high or extreme. This could increase to 117.6-
129.3 days by 2020 and 120.1-153.5 days by 2050.
Wagga
The average annual accumulated FFDI is currently 4047. This could increase 4-9% by 2020 and
9-25% by 2050. The monthly-average FFDI currently exceeds 20 from early December to late
February. This could extend from late November to early March by 2020, and from mid-November
to early March by 2050. On average, there are currently 49.6 days when the FFDI rating is very
high or extreme. This could increase to 52.7-57.4 days by 2020 and 57.6-71.9 days by 2050. There
are currently 104.6 days when the GFDI rating is very high or extreme. This could increase to
110.7-118.7 days by 2020 and 114.4-134.2 days by 2050.
Williamtown
The average annual accumulated FFDI is currently 2641. This could increase 2-7% by 2020 and
5-18% by 2050. The monthly-average FFDI currently exceeds 8 from late August to late January.
This could extend from mid-August to late January by 2020, and from early August to early
February to early April by 2050. On average, there are currently 16.4 days when the FFDI rating is
very high or extreme. This could increase to 17.2-19.4 days by 2020 and 18.4-23.6 days by 2050.
There are currently 123.1 days when the GFDI rating is very high or extreme. This could increase to
132.2-141.8 days by 2020 and 135.1-162.9 days by 2050.
Bendigo
The average annual accumulated FFDI is currently 2854. This could increase 3-8% by 2020 and
8-23% by 2050. The monthly-average FFDI currently exceeds 10 from late November to late
March. This could extend from early November to late March by 2020, and from late October to
early April by 2050. On average, there are currently 17.8 days when the FFDI rating is very high or
extreme. This could increase to 19.5-21.9 days by 2020 and 21.4-29.8 days by 2050. There are
currently 61.1 days when the GFDI rating is very high or extreme. This could increase to 63.6-69.5
days by 2020 and 65.9-81.7 days by 2050.
Laverton
The average annual accumulated FFDI is currently 2913. This could increase 3-9% by 2020 and
8-24% by 2050. The monthly-average FFDI currently exceeds 10 from mid-December to early
March. This could extend from early December to mid-March by 2020, and from early November
to late March by 2050. On average, there are currently 15.5 days when the FFDI rating is very high
or extreme. This could increase to 16.4-17.8 days by 2020 and 17.3-22.3 days by 2050. There are
currently 110.1 days when the GFDI rating is very high or extreme. This could increase to 109.4-
117.4 days by 2020 and 111.9-131.7 days by 2050.
28
Melbourne
The average annual accumulated FFDI is currently 2121. This could increase 3-8% by 2020 and
8-22% by 2050. The monthly-average FFDI currently exceeds 8 from mid-December to mid-March.
This could extend from early December to late March by 2020, and from early November to late
April by 2050. On average, there are currently 9.0 days when the FFDI rating is very high or
extreme. This could increase to 9.8-11.1 days by 2020 and 10.8-14.7 days by 2050. There are
currently 38.7 days when the GFDI rating is very high or extreme. This could increase to 41.2-45.0
days by 2020 and 42.2-54.5 days by 2050.
Mildura
The average annual accumulated FFDI is currently 5898. This could increase 3-8% by 2020 and
7-21% by 2050. The monthly-average FFDI currently exceeds 20 from late October to mid-March.
This could extend from mid-October to mid-March by 2020, and from early-October to late March
by 2050. On average, there are currently 79.5 days when the FFDI rating is very high or extreme.
This could increase to 83.9-90.7 days by 2020 and 89.9-107.3 days by 2050. There are currently
146.7 days when the GFDI rating is very high or extreme. This could increase to 149.1-157.6 days
by 2020 and 153.9-174.6 days by 2050.
Sale
The average annual accumulated FFDI is currently 2207. This could increase 3-8% by 2020 and
8-23% by 2050. The monthly-average FFDI currently exceeds 8 from early December to mid-
March. This could extend from late November to mid-March by 2020, and from late October to late
March by 2050. On average, there are currently 8.7 days when the FFDI rating is very high or
extreme. This could increase to 9.3-10.7 days by 2020 and 10.1-14.0 days by 2050. There are
currently 95.4 days when the GFDI rating is very high or extreme. This could increase to 102.5-
110.2 days by 2020 and 104.1-124.2 days by 2050.
Hobart
The average annual accumulated FFDI is currently 1723. This is unlikely to change by more than
1 or 2% over the next 50 years since projected increases in temperature are offset by increases in
rainfall and humidity. The monthly-average FFDI currently exceeds 6 from early December to mid-
March and shows little change by 2050. On average, there are currently 3.4 days when the FFDI
rating is very high or extreme. This is unlikely to change over the next 50 years. There are currently
67.5 days when the GFDI rating is very high or extreme. This could increase to 67.5-68.8 days by
2020 and 67.2-71.5 days by 2050.
Launceston
The average annual accumulated FFDI is currently 1677. This could increase 1-6% by 2020 and
3-17% by 2050. The monthly-average FFDI currently exceeds 10 from late November to late
March. This could extend from mid-November to late March by 2020, and from early November to
early April by 2050. On average, there are currently 1.5 days when the FFDI rating is very high or
extreme. This could increase to 1.5-1.9 days by 2020 and 1.6-3.1 days by 2050. There are currently
73.3 days when the GFDI rating is very high or extreme. This could increase to 73.4-85.0 days by
2020 and 72.3-102.8 days by 2050.
29
4. Discussion
Following the widespread fires in December 2002 and January 2003, a number of inquiries were
undertaken. For the ACT, the McLeod Inquiry Report (2003) recommended a range of fire
mitigation activities to be undertaken prior to, and during, the 2003-04 bushfire season, with an
additional $1.684 million being sought for that purpose, adding to the $0.5 million provided in the 2003-
04 budget.
The COAG (2004) Report of the National Inquiry on Bushfire Mitigation and Management stated
“Climate change is likely to increase the frequency, intensity and size of bushfires in much of
Australia in the future”. It is possible that changes in the FFDI and other indices will require
prescribed burning to take place a little earlier in spring and a little later in autumn, prolonging the
effective fire season, increasing the personal and employer cost for volunteers, and increasing the
cost of fire fighters. Climate change impacts would be seen in potentially prolonged fire danger
periods, increased numbers of total fire ban days, increased community based educational and
organizational programs such as Community Fire Guard (2005), and in increased reliance on the
good will of employers or volunteers. The summary concluded that “more research is needed on
building design and materials, climate and climate change, fire behaviour and ecological responses,
individual and community psychology and social processes, and Indigenous Australians’
knowledge and use of fire”. It also concluded that “long-term strategic research, planning and
investment are necessary if the Australian Government and state and territory governments are to
prepare for the changes to bushfire regimes and events that will be caused by climate change”.
The Report of the Inquiry into the 2002–2003 Victorian Bushfires (Esplin et al, 2003) noted that
“The weather leading up to a fire season is not the only aspect of climate that influences the severity
of a fire event. The weather at the time of a fire has a major impact on fire behaviour and on the
ease of suppression. In relation to the 2002–03 fire season, the Bureau of Meteorology stated: The
very dry conditions leading into the 2002/03 fire season do not in themselves fully explain the
intensity and longevity of the fire episodes. A significant contributor to the long period for which
the 2003 bushfires remained active was the absence of any significant rain for several weeks after”.
It also stated “A prolonged and severe drought, especially throughout much of the southern half of
Australia, is the stand-out climatic feature of the 2002–03 fire season. Fire agencies need to be
responsive to macro indicators of this kind, using them to assist with annual planning and
preparation activities, as well as to match their response capacity to daily weather conditions.
Operational responses during drought periods should reflect the ‘worst case’ scenario and include
optimum available resourcing. Although the full extent of the fire threat may not be realised,
operational planning must take account of this possibility”.
The results of this study provide scenarios that reconfirm the findings of these inquiries. The
impacts of climate change are likely to pose a number of challenges for natural and human systems.
However, few impact assessments have been done. It is likely that an increase in the frequency and
intensity of fire-weather would:
• alter the distribution and composition of ecosystems (Cary, 2002)
• lower the yield and quality of water from fire-affected catchments (Lavoral and Steffen,
2004)
• threaten the security of plantation forests
• increase smoke-related respiratory illness
• increase emissions of greenhouse gases to the atmosphere
• increase damage to property, livestock and crops
• increase the exposure of insurance companies to loss (Coleman et al,, 2004)
• increase the risk of injury, trauma and death to humans (BTE, 2001).
30
5. Gaps in knowledge and research priorities
This study has quantified present average fire-weather risk at 17 sites in southeast Australia and
potential changes for the years 2020 and 2050. A number of knowledge gaps remain:
• Quality of daily wind data at most sites in Australia
• Quality of daily humidity data at sites outside southeast Australia
• The effect of scenarios based on other climate models
• Future changes in intervals between rainfall events during the fire season
• Future changes in ignition (natural and anthropogenic)
• Future changes in fuel load, allowing for carbon dioxide fertilization of vegetation.
• Potential impacts on biodiversity, water yield and quality from fire affected catchments,
forestry, greenhouse gas emissions, emergency management and insurance.
Priorities for further research are outlined in Figure 7, including:
• Testing and rehabilitation of observed humidity and wind data (underway within the Bureau
of Meteorology, supported by the Bushfire CRC).
• Deriving better regional and local predictions of fire weather (especially extreme events).
This could involve creation of climate change scenarios from other models (monthly output
from a new suite of 23 models was made available by the IPCC in mid-2005), and finer
resolution daily data using “downscaling” methods.
• Extending the analysis to other regions, e.g. Queensland, South Australia, Western Australia
and the Northern Territory.
• Modelling of changed vegetation growth and fuel dynamics under climate change. This
requires incorporation of the effects of changed atmospheric composition and climate into
interactive models of the Australian biosphere.
• Modelling that integrates changing climate and fuels with landscape features to predict the
nature and extent of fire under a range of fire management scenarios, including prescribed
burning and suppression.
• Hydrological and ecological modelling to assess impacts on water and biodiversity.
• Assessment of potential impacts on community safety and insurance liabilities.
• Using satellite remote sensing (Sentinel fire mapping based on MODIS, NCAS land cover
change based on Landsat, and other data) to monitor the extent and nature of fire, recovery
of vegetation after fire, and greenhouse gas emissions from fire.
Climate
change
Vegetation / Fuels
Fire Weather
(extreme events)
Nature / extent of Fire
Greenhouse Impacts
– non-CO
2
emissions
– charcoal sink
– C stock in veg. / soil
Other Socio-economic
Impacts
– community safety
– water, biodiversity
– insurance
Risk assessment, fire management, and GHG accounting under current
(variable) Australian climate will inform policy
Figure 7. Proposed agenda for research to address knowledge gaps and inform policy.
31
Broader participation in this research is required in order to engage relevant groups in government,
industry and the community. Outcomes of discussions might include identification of:
• Regional fire management issues affected by climate variability and climate change in each
State/Territory
• Other information available for assessing current fire vulnerability and potential changes due to
greenhouse warming
• Important biophysical and behavioural/management thresholds
• Technical adaptation options
• Institutional processes that influence adaptation
• Information required for future planning (e.g. potential change in seasonal average fire risk,
frequency of extreme fire-risk days, interval between fires, fire intensity, fuel load, etc.)
• Information about fire damage from insurance companies and government sources in each
State/Territory.
6. References
Anderson, S.A.J. and Pearce H.G. (2003). Improved Methods for the Assessment of Grassland
Curing. Conference Proceedings, 3
rd
International Wildland Fire Conference, 3-6 October
2003, Sydney, Australia, 10 pp, CD-ROM.
Beer, T. and Williams, A. (1995). Estimating Australian forest fire danger under conditions of
doubled carbon dioxide concentrations. Climatic Change, 29, 169-188.
Beer, T., Gill, A.M. and Moore, P.H.R. (1988). Australian bushfire danger under changing climatic
regimes. In Greenhouse, Planning for Climate Change. Published by CSIRO, Melbourne,
421-427.
Bradstock, R.A. and Gill, A.M. (2001). Living with fire and biodiversity at the urban edge: in
search of a sustainable solution to the human protection problem, Journal of Mediterranean
Ecology, 2, 179-195.
Brown, T.J., Hall, B.L. and Westerling, A.L. (2004). The impacts of 21
st
century climate change on
wildland fire danger in the western United States: an application perspective. Climatic
Change, 62, 365-388.
BTE (2001). Economic costs of natural disasters in Australia. Bureau of Transport Economics,
Canberra.
http://www.bte.gov.au/docs/r103/contents.htm#Top.
Cary, G.J. (2002). Importance of a changing climate for fire regimes in Australia, in Bradstock,
R.A. Williams, J.E. and Gill, A.M., “Flammable Australia: Fire Regimes and Biodiversity of a
Continent”, Cambridge University Press, Melbourne, pp 26-49.
Cheney, P. (1997). Grassfires: fuel, weather and fire behaviour. CSIRO Publishing, Collingwood,
Vic, 100 pp.
COAG (2004). Report of the National Inquiry on Bushfire Mitigation and Management,
http://www.coagbushfireenquiry.gov.au/findings.htm
Coleman, T., Hoegh-Guldberg, O., Karoly, D., Lowe, I., McMichael, T., Mitchell, C.D., Pearman,
G.I., Scaife, P. and Reynolds, A. (2004). Climate change: solutions for Australia. The
Australian Climate Group, 35 pp.
CSIRO (2001). Climate change projections for Australia. CSIRO Atmospheric Research brochure,
8 pp. http://www.dar.csiro.au/publications/projections2001.pdf
Esplin, B., Gill, A.M. and Enright, N. (2003). Report of the Inquiry into the 2002–2003 Victorian
Bushfires, State Government of Victoria, Melbourne.
Goldammer, J.G. and Price, C. (1998). Potential impacts of climate change on fire regimes in the
tropics based on MAGICC and a GISS GCM-derived lightning model. Climatic Change, 39,
273-296.
Griffiths, D. (1999). Improved formula for the drought factor in McArthurs Forest Fire Danger
Meter. Australian Forestry, 62(2), 202-206.
32
Haylock, M. and Nicholls, N. (2000). Trends in extreme rainfall indices for an updated high quality
data set for Australia, 1910-1998. International Journal of Climatology, 20, 1533-1541.
Hennessy, K., McInnes, K., Abbs, D., Jones, R., Bathols, J., Suppiah, R., Ricketts, J., Rafter, T.,
Collins D. and Jones, D. (2004). Climate Change in New South Wales, Part 2: Projected
changes in climate extremes. Consultancy report for the New South Wales Greenhouse
Office by the Climate Impact Group of CSIRO Atmospheric Research and the National
Climate Centre of the Australian Government Bureau of Meteorology, 79 pp.
IPCC (2001). Summary for Policymakers, in Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M.,
Van Der Linden, P.J. and Xioaosu, D (eds) Climate Change 2001: The Scientific Basis,
Contribution of Working Group I to the Third Assessment Report of the Intergovernmental
Panel on Climate Change, Cambridge University Press, Cambridge, 944 pp.
Keetch, J.J. and Byram, G.M. (1968). A drought index for forest fire control. USDA Forest Service
Research Paper SE-38, Asheville, North Carolina.
Lavorel, S. and Steffen, W. (2004). Cascading impacts of land use through time: the Canberra
bushfire disaster. In: Steffen W, Sanderson A, Tyson P, Jäger J, Matson P, Moore III B,
Oldfield F, Richardson K, Schellnhuber H-J, Turner II BL, Wasson R. Global Change and
the Earth System: A Planet Under Pressure. IGBP Global Change Series. Springer-Verlag,
Berlin Heidelburg New York, 186-188.
McGregor, J.L. and Dix, M.R. (2001). The CSIRO conformal-cubic atmospheric GCM. In IUTAM
Symposium on Advances in Mathematical Modelling of Atmosphere and Ocean Dynamics, P.
F. Hodnett (Ed. ), Kluwer, Dordrecht, 197-202.
McInnes, K.L., Bathols, J.M., Page, C.M., Suppiah, R. and Whetton, P.H. (2004). Climate change
in Tasmania: a report undertaken for Hydro Tasmania by the Climate Impact Group, C/0919.
CSIRO Atmospheric Research, Aspendale, Vic., 49 pp.
McInnes, K.L., Abbs, D.J. and Bathols, J.M. (2005). Climate Change in Eastern Victoria Stage 1
Report: The effect of climate change on coastal wind and weather patterns. CSIRO Marine
and Atmospheric Research consultancy report for the Gippsland Coastal Board.
http://www.dar.csiro.au/publications/mcinnes_2005a.pdf, 26 pp.
McLeod Inquiry Report (2003).
http://www.cmd.act.gov.au/mcleod_inquiry/Documents/Tabling.pdf
McMichael, A.J., Woodruff, R.E., Whetton, P., Hennessy, K., Nicholls, N., Hales, S., Woodward,
A. and Kjellstrom, T. (2003). Human health and climate change in Oceania: a risk
assessment. Commonwealth Department of Health and Ageing, 116 pp.
Mitchell, T.D. (2003). Pattern scaling – an examination of the accuracy of the technique for
describing future climates. Climatic Change, 60, 217-242.
Mount, A.B. (1972). The derivation and testing of a soil dryness index using run-off data.
Tasmanian Forestry Commission, Bulletin No. 4, Hobart.
Nicholls, N. (2003). Continued anomalous warming in Australia. Geophysical Research Letters,
30(7), article 23.
Noble, I.R., Bary, G.A.V. and Gill, A.M. (1980). McArthur’s fire-danger meters expressed as
equations. Australian Journal of Ecology, 5, 201-203.
Pearce, G., Mullan, A.B., Salinger, M.J., Opperman, T.W., Woods, D. and Moore, J.R. (2005).
Impact of climate variability and change on long-term fire danger. Report to the New
Zealand Fire Service Commission, 75 pp.
Purton, C.M. (1982). Equations for the McArthur Mk 4 Grassland fire danger meter. Bureau of
Meteorology Meteorological Note no. 147, 12 pp.
SRES (2000). Special Report on Emission Scenarios: Summary for Policymakers. A Special Report
of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge
University Press, Cambridge, UK, http://www.ipcc.ch/pub/sres-e.pdf, 27 pp.
Stocks, B.J., Fosberg, M.A., Lynham, T.J., Mearns, L., Wooton, B.M. et al (1998). Climate change
and forest fire potential in Russian and Canadian boreal forests. Climatic Change, 38, 1-13.
33
Suppiah, R., Whetton, P. H., and Watterson, I. G. (2004). Climate change in Victoria: assessment of
climate change for Victoria: 2001-2002. Consultancy report for Victorian Department of
Sustainability and Environment. CSIRO Atmospheric Research, Aspendale, Vic, 33 pp.
Whetton, P.H. (2001). Methods used to prepare the ranges of projected future change in Australian
region temperature and precipitation. CSIRO Technical Report.
http://www.dar.csiro.au/impacts/docs/how.pdf
Williams, A., Karoly, D.J. and Tapper, N. (2001). The sensitivity of Australian fire danger to
climate change. Climatic Change, 49, 171-191.
Wotton, B.M., Martell, D.L. and Logan, K.A. (2003). Climate change and people-caused fire
occurrence in Ontario. Climatic Change, 60, 275-295.
34
Appendix 1 IPCC scenarios of global warming
The IPCC (2001) attributes most of the global warming observed over the last 50 years to
greenhouse gases released by human activities. To estimate future climate change, the IPCC (SRES,
2000) prepared forty greenhouse gas and sulfate aerosol emission scenarios for the 21
st
century that
combine a variety of assumptions about demographic, economic and technological driving forces
likely to influence such emissions in the future. They do not include the effects of measures to
reduce greenhouse gas emissions, such as the Kyoto Protocol.
Each scenario represents a variation within one of four 'storylines': A1, A2, B1 and B2. The experts
who created the storylines (described below) were unable to arrive at a most likely scenario, and
probabilities were not assigned to the storylines.
A1 describes a world of very rapid economic growth in which the population peaks around 2050
and declines thereafter and there is rapid introduction of new and more efficient technologies. The
three sub-groups of A1 are fossil fuel intensive (A1FI), non-fossil fuel using (A1T), and balanced
across all energy sources (A1B).
The A2 storyline depicts a world of regional self-reliance and preservation of local culture. In A2,
fertility patterns across regions converge slowly, leading to a steadily increasing population and per
capita economic growth and technological change is slower and more fragmented than for the other
storylines.
The B1 storyline describes a convergent world with the same population as in A1, but with an
emphasis on global solutions to economic, social and environmental sustainability, including the
introduction of clean, efficient technologies.
The B2 storyline places emphasis on local solutions to economic, social and environmental
sustainability. The population increases more slowly than that in A2. Compared with A1 and B1,
economic development is intermediate and less rapid, and technological change is more diverse.
The projected carbon dioxide and sulfate aerosol emissions, and carbon dioxide concentrations, are
shown in Figure 1-A1 (a, b, c). Emissions of other gases and other aerosols were included in the
scenarios but are not shown in the figure. By incorporating these scenarios into computer models of
the climate system, the IPCC (2001) estimated a global-average warming of 0.7 to 2.5
o
C by the year
2050 and 1.4 to 5.8
o
C by the year 2100 (Figure 2-A1d). The analysis allowed for both uncertainty in
projecting future greenhouse gas and aerosol concentrations (behavioural uncertainty) and
uncertainty due to differences between models in their response to atmospheric changes (scientific
uncertainty). Projected sea-level rise is shown in Figure 2-A1e.
The range of uncertainty in projections of global warming increases with time. Half of this range is
due to uncertainty about human socio-economic behaviour, and consequent emissions of
greenhouse gases and sulfate aerosols. The other half of the range is due to different climate model
responses to these scenarios of greenhouse gases and sulfate aerosols. Each of the models is
considered equally reliable.
Climate simulations indicate that warming will be greater near the poles and over the land, and that
global-average rainfall will increase. More rainfall is likely nearer the poles and in the tropics, and
less rainfall is expected in the middle latitudes such as southern Australia.
35
Figure 1-A1: (a) carbon dioxide (CO
2
) emissions for the six illustrative SRES (2000) scenarios, and
the superseded IS92a scenario, (b) CO
2
concentrations, (c) anthropogenic sulphur dioxide (SO
2
)
emissions, (d) and (e) show the projected temperature and sea level responses, respectively. Source:
IPCC (2001).
It is important to note that at present, it is not possible to assign probabilities to values within these
ranges. However, the IPCC (2001b) defined confidence levels that represent “the degree of belief
among the authors in the validity of a conclusion, based on their collective expert judgment of
observational evidence, modelling results and theory that they have examined”. The confidence
levels are:
• Very high (95% or greater);
• High (67-94%);
• Medium (33-66%);
• Low (5-32%);
• Very low (4% or less).
For the global warming data in Figure 1-A1, we have very high confidence that the lower warming
limits will be exceeded and that the higher limits will not be exceeded.
36
Appendix 2: Soil Dryness Index
Bendigo
Bourke
37
Cabramurra
Canberra
38
Cobar
]
Coffs Harbour
39
Hobart
Launceston Airport
40
Laverton
Melbourne
41
Mildura
Nowra
42
Richmond NSW
Sale
43
Sydney
Wagga
44
Williamtown
45
Appendix 3: Keetch Byram Drought Index Index
Bendigo
Bourke
46
Cabramurra
Canberra
47
Cobar
Coffs Harbour
48
Hobart
Launceston Airport
49
Laverton
Melbourne
50
Mildura
Nowra
51
Richmond NSW
Sale
52
Sydney
Wagga
53
Williamtown
54
Appendix 4: Forest Fire Danger Index
Bendigo
55
Bourke
56
Cabramurra
57
Canberra
58
Cobar
59
Coffs Harbour
60
Hobart
61
Launceston Airport
62
Laverton
63
Melbourne
64
Mildura
65
Nowra
66
Richmond NSW
67
Sale
68
Sydney
69
Wagga
70
Williamtown
71
Appendix 5: Grassland Fire Danger Index
Bendigo
72
Bourke
73
Cabramurra
74
Canberra
75
Cobar
76
Coffs Harbour
77
Hobart
78
Launceston Airport
79
Laverton
80
Melbourne
81
Mildura
82
Nowra
83
Richmond NSW
84
Sale
85
Sydney
86
Wagga
87
Williamtown
88
Climate change impacts on
fi re-weather in south-east
Australia
K. Hennessy, C. Lucas* N. Nicholls* J. Bathols, R. Suppiah
and J. Ricketts
CSIRO Marine and Atmospheric Research
* Bushfi re CRC and Australian Bureau of Meteorology
December 2005
www.csiro.au
ACT Government
2
Enquiries should be addressed to:
Kevin Hennessy
CSIRO Marine and Atmospheric Research
PMB No 1, Aspendale, Victoria, 3195
Telephone (03) 9239 4536
Fax (03) 9239 4444
E-mail Kevin.Hennessy@csiro.au
Important Notice
© Copyright Commonwealth Scientific and Industrial Research Organisation
(‘CSIRO’) Australia December 2005
All rights are reserved and no part of this publication covered by copyright may be reproduced or copied in
any form or by any means except with the written permission of CSIRO.
The results and analyses contained in this Report are based on a number of technical, circumstantial or
otherwise specified assumptions and parameters. The user must make its own assessment of the suitability
for its use of the information or material contained in or generated from the Report. To the extent permitted by
law, CSIRO excludes all liability to any party for expenses, losses, damages and costs arising directly or
indirectly from using this Report.
Use of this Report
The use of this Report is subject to the terms on which it was prepared by CSIRO. In particular, the Report may only be
used for the following purposes.
x this Report may be copied for distribution within the Client’s organisation;
x the information in this Report may be used by the entity for which it was prepared (“the Client”), or by
the Client’s contractors and agents, for the Client’s internal business operations (but not licensing to
third parties);
x extracts of the Report distributed for these purposes must clearly note that the extract is part of a larger
Report prepared by CSIRO for the Client.
The Report must not be used as a means of endorsement without the prior written consent of CSIRO.
The name, trade mark or logo of CSIRO must not be used without the prior written consent of CSIRO.
ISBN 1 921061 10 3
For more information about climate change, see http://www.cmar.csiro.au/impacts/index.html and
http://www.bom.gov.au/climate/change
Climate Change Cover.indd 2Climate Change Cover.indd 2 25/1/06 9:09:52 AM25/1/06 9:09:52 AM
Climate change impacts on
fi re-weather in south-east
Australia
K. Hennessy, C. Lucas* N. Nicholls* J. Bathols, R. Suppiah
and J. Ricketts
CSIRO Marine and Atmospheric Research
* Bushfi re CRC and Australian Bureau of Meteorology
www.csiro.au
ACT Government
Climate Change Cover.indd 1Climate Change Cover.indd 1 25/1/06 9:09:48 AM25/1/06 9:09:48 AM