Climate change and fire-weather risk in south-eastern Australia

1
Climate change and fire-weather risk in
south-eastern Australia

• Kevin Hennessy

2
Observed changes in Australia
Most of the top 20 insured losses have been due
to extreme weather events
Black Saturday fires (7 Feb 2009) 173 dead, 2029
properties destroyed (30 uninsured), 7000 people
displaced, 61 businesses destroyed, over 8600
livestock dead (4000 sheep, 4200 beef cattle, 427
dairy cows)
Insurance Council of Australia (2007)
3
Fire risk in Australia

• Fire risk is influenced by a number of factors
  including weather, fuels, ignition, terrain, land
  management and suppression
• The CSIRO-BoM study assesses potential changes to
  one of these factors, fire-weather risk,
  associated with climate change
• The most important weather variables are
  temperature, humidity, wind-speed and rainfall

4
Seasonal pattern of fire danger
5
Fire danger index

• 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 oC
• V average wind-speed 10 metres above the
  ground, in metres per second
• D drought factor in the range 0-10

6
FFDI categories
Categories of Fire Danger Rating (FDR). Taken
from Vercoe 2003.
Fire Danger Rating FFDI range Difficulty of suppression
Low 0-5 Fires easily suppressed with hand tools.
Moderate 5-12 Fire usually suppressed with hand tools and easily suppressed with bulldozers. Generally the upper limit for prescribed burning.
High 12-25 Fire generally controlled with bulldozers working along the flanks to pinch the head out under favourable conditions. Back burning may fail due to spotting.
Very High 25-50 Initial attack generally fails but may succeed in some circumstances. Back burning will fail due to spotting. Burning-out should be avoided.
Extreme 50 Fire suppression virtually impossible on any part of the fire line due to the potential for extreme and sudden changes in fire behaviour. Any suppression actions such as burning out will only increase fire behaviour and the area burnt.
7
Daily peak FFDIs for large fires
Date FFDI
13 Jan 1939 (SE Aus) 100 24 Jan 1961
(Dwellingup, WA) 110 7 Feb 1967 (Hobart,
Tas) 78 16 Feb 1983 (Deans Marsh, Vic) 100
16 Feb 1983 (Trentham, Vic) 60 16 Feb
1983 (Adelaide to Warrnambool) gt 100 8 Jan
1994 (Bankstown, NSW) 88 18 Jan 2003
(Canberra) 115 7 Feb 2009 (Vic)
100-192
8
Observed changes in FFDI
Large daily, seasonal and annual variability
9
Fire intensity, fuel load and FFDI
To get a fire of specified intensity, you need
more fuel if the fire-weather index is lower
Effect of Fuel Load on FFDI value for a fire with
an intensity of 3500 kW m-1, the threshold for
uncontrollable fires. Adapted from data
provided by Incoll 1994.
10
Report on climate change and fire-weather
Bushfire Weather in Southeast Australia Recent
Trends and Projected Climate Change Impacts C.
Lucas, K. Hennessy, G. Mills and J.
Bathols Bushfire CRC and Australian Bureau of
Meteorology CSIRO Marine and Atmospheric
Research September 2007 Consultancy Report
prepared for The Climate Institute of
Australia www.bushfirecrc.com/research/downloads/
climate-institute-report-september-2007.pdf
11
Changes in fire weather risk in south-east
Australia were assessed for the period 1973-2007
at 26 sites
Site selection was limited by availability of
daily temperature, rainfall, humidity and wind
data
12
Observed changes in FFDI

• The annual cumulative FFDI displays a rapid
  increase in the late-90s to early-00s at many
  locations.
• Increases of 10-40 between 1980-2000 and
  2001-2007 are evident at most sites.
• The increases are associated with a jump in the
  number of very high and extreme fire danger days.
  

AP Photo
13
Observed changes in FFDI
The trend (red number at top left) is 51 FFDI
units per year
14
Observed changes in FFDI
Melbourne airport
The trend (red number at top left) is 23 FFDI
units per year
15
Observed changes in FFDI
The trend (red number at top left) is 51 FFDI
units per year
16
Observed changes in FFDI
The trend (red number at top left) is 24 FFDI
units per year
17
Climate change projections

• Rather than simply extrapolating observed trends,
  we use computer models of the climate system
  driven by scenarios of greenhouse gas and aerosol
  emissions, and ozone depletion
• The IPCC emission scenarios have various
  assumptions about demographic, economic and
  technological change

Currently tracking the high scenario
18
Projected global warming
For the IPCC emission scenarios, 23 climate
models simulate a warming of 0.5 to 1.6C by
2030, rising to 1.1 to 6.4C by 2100
A1FI (estimated by CSIRO)
IPCC 2007
19
Observed trends in temperature

• Global average temperature is tracking the upper
  end of the IPCC projections
• The rate of warming since 1975 is almost 0.2oC
  per decade

Observed
Rahmstorf et al 2007
20
Climate change scenarios for 26 sites were
generated using 2 CSIRO climate models
Model selection criteria good simulation of
1961-1990 mean temperature, rainfall and MSLP
availability of daily data at fine resolution,
e.g. 50 km Two climate change simulations were
suitable CCAM driven by the CSIRO Mark2 GCM, and
CCAM driven by the CSIRO Mark3 GCM The mean
warming is 0.5-1.5oC by 2020 and 1.5-3.0oC by
2050 CCAM Mark2 rainfall decreases except in
autumn in northern Vic and southern NSW, humidity
decreases in spring and summer and increases in
autumn and winter, wind-speed decreases CCAM
Mark3 rainfall decreases in spring-summer and
increases in autumn-winter, humidity decreases,
wind-speed increases
21
Simulated changes in mean and variability
This study included simulated changes in the mean
and variability of daily temperature, rainfall,
humidity and wind-speed Changes in daily decile
values for each calendar month were applied to
observed daily data from 1974-2007 Observed daily
temperature and rainfall data were considered
high quality Observed daily humidity data were
acceptable at most sites Observed wind data were
not homogenised, so there were some jumps and
missing data
22
change in average no. of days with very high
and extreme forest fire danger in 2020 and 2050
Extreme fire-weather days (high global
warming) 2020 2050 Bendigo 53-65 135-230
Melbourne 26-38 81-136 Mildura 25-38 76-
120 Sale 15-45 80-215
All changes are relative to a period centred on
1990
23
FFDI gt 100
We defined a catastrophic fire-weather category
for FFDI gt 100. Only 12 of the 26 sites have
recorded catastrophic fire-weather days since
1973. The 2020 low scenarios indicate little or
no change, except for a halving of the return
period (doubling frequency) at Bourke. The 2020
high scenarios show catastrophic days occurring
at 20 sites, 10 of which have return periods of
around 16 years or less. By 2050, the low
scenarios are similar to those for the 2020 high
scenarios. The 2050 high scenarios show
catastrophic days occurring at 22 sites, 19 of
which have return periods of around 8 years or
less, while 7 sites have return periods of 3
years or less.
24
Longer fire seasons
Median FFDI for summer, autumn, winter and spring
for now (1973-2007), 2020 and 2050 at Melbourne
and Canberra The median FFDI increases in all
seasons (mostly spring and summer), implying
longer fire seasons
25
Adaptation challenges

• Longer and more intense fire seasons present
  significant challenges for adaptation
• Greater social, environmental and economic
  costs
• More control-burning, smoke pollution and
  respiratory illness
• More volunteer fire-fighters, more pressure on
  families/employers
• Better weather forecasts and early warning
  systems
• New technology for fighting fires
• More disaster relief payments counselling
  services
• Better protection of water catchments
  plantation forests
• Better emergency plans (stay or go)
• Planning guidelines for the urban/rural fringe
• Insurance premium incentives for those that
  reduce their fire risk
• Building codes/designs with reduced
  flammability

26
Uncertainties

• Quality of observed daily wind and humidity data
  at most sites in Australia
• The effect of scenarios based on other climate
  models
• Assessment of daily-annual variability in FFDI,
  not just annual averages
• Changes in ignition (natural and human-induced)
• 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

AP Photo The Age
27
Research priorities

• Testing and rehabilitation of observed humidity
  and wind data (supported by Bushfire CRC and BoM)
• Creation of regional climate change scenarios
  from other models (underway)
• Assess fire-weather risk over the whole of
  Australia (underway)
• Fine scale fire modelling that captures
  vegetation and terrain features and fire
  management, e.g. using FIRESCAPE (Sydney basin
  project)
• Hydrological and ecological modelling to assess
  impacts on water and biodiversity