ESCI Technical Report - Standardised Method - 1

Extreme temperature, wind and bushfire weather
projections using a standardised method
Andrew Dowdy1, Andrew Brown1, Acacia Pepler1, Marcus Thatcher2, Tony Rafter2, Jason
Evans3, Hua Ye1, Chun-Hsu Su1, Samuel Bell1 and Christian Stassen1
1
Bureau of Meteorology, Docklands, Australia
2
CSIRO Oceans and Atmosphere, Aspendale, Australia
3
University of New South Wales, Sydney, Australia
July 2021
Bureau Research Report - 055

EXTREME TEMPERATURE, WIND AND BUSHFIRE WEATHER PROJECTIONS USING A STANDARDISED METHOD
EXTREME TEMPERATURE, WIND AND BUSHFIRE WEATHER USING A STANDARDISED METHOD
Extreme temperature, wind and bushfire weather

projections using a standardised method
Andrew Dowdy1, Andrew Brown1, Acacia Pepler1, Marcus Thatcher2, Tony Rafter2,
Jason Evans3, Hua Ye1, Chun-Hsu Su1, Samuel Bell1 and Christian Stassen1
Bureau of Meteorology, Docklands, Australia
1
CSIRO Oceans and Atmosphere, Aspendale, Australia

2
3
University of New South Wales, Sydney, Australia
Bureau Research Report No. 055
July 2021
National Library of Australia Cataloguing-in-Publication entry
Authors: Andrew Dowdy, Andrew Brown, Acacia Pepler, Marcus Thatcher, Tony Rafter, Jason Evans,
Hua Ye, Chun-Hsu Su, Samuel Bell and Christian Stassen
Title: Extreme temperature, wind and bushfire weather projections using a standardised method
ISBN: 978-1-925738-32-2
ISSN: 2206-3366
Series: Bureau Research Report – BRR055
i
Enquiries should be addressed to:
Lead Author: Andrew Dowdy
Bureau of Meteorology
GPO Box 1289, Melbourne
Victoria 3001, Australia
andrew.dowdy@bom.gov.au
Copyright and Disclaimer
© 2021 Bureau of Meteorology. To the extent permitted by law, 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 the Bureau of Meteorology.
The Bureau of Meteorology advise that the information contained in this publication comprises general
statements based on scientific research. The reader is advised and needs to be aware that such
information may be incomplete or unable to be used in any specific situation. No reliance or actions
must therefore be made on that information without seeking prior expert professional, scientific and
technical advice. To the extent permitted by law and the Bureau of Meteorology (including each of its
employees and consultants) excludes all liability to any person for any consequences, including but not
limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly
from using this publication (in part or in whole) and any information or material contained in it.
ii
Contents
Abstract ...................................................................................................................... 1
1. Introduction and overview ................................................................................ 2
2. Methodology ...................................................................................................... 4
2.1 Standardised method for projections information ..................................................... 4
2.2 Examples of method outputs being used ................................................................. 6
3. Extreme temperature projections .................................................................... 8
3.1 Introduction ............................................................................................................... 8
3.2 Summaries for physical processes ........................................................................... 8
3.3 Summaries for historical climate ............................................................................. 15
3.4 Summaries for future climate .................................................................................. 15
3.5 Lines of evidence table ........................................................................................... 17
3.6 Projections and confidence information .................................................................. 18
4. Extreme wind projections ............................................................................... 20
4.1 Introduction ............................................................................................................. 20
4.2 Summaries for physical processes ......................................................................... 20
5. Extreme fire weather projections ................................................................... 34
5.1 Introduction ............................................................................................................. 34
5.2 Summaries for physical processes ......................................................................... 35
6. Conclusion ...................................................................................................... 45
Acknowledgements.................................................................................................. 45
References ................................................................................................................ 45
iii
ABSTRACT
The influence of anthropogenic climate change on extreme temperatures, winds and bushfire
weather in Australia is assessed here using a standardised method for projections information.
These assessments consider a comprehensive range of factors based on observations, modelling
and physical process understanding. Those factors are reviewed using a standardised method to
collate lines of evidence and then guide the production of projections data and confidence
assessments. Projections are produced based on global climate model data as well as dynamical
downscaling data using three regional climate modelling approaches (CCAM, BARPA and
NARCliM/WRF), with environmental diagnostics also used for severe convective winds from
thunderstorms. The projections data are all calibrated using quantile matching methods trained
on observations-based data, with a particular focus on the accurate representation of extremes.
The resultant projections data include nationally consistent maps corresponding to the 10-year
average recurrence interval (i.e., return period) around the middle of this century, with a focus of
the discussion on regions around southern and eastern Australia during summer as needed for
some risk assessment applications. The projections data are also available for other seasons and
time periods throughout this century, as well as for other metrics of extreme or average conditions.
The results for southern and eastern Australia during summer show more extreme temperatures
(very high confidence), more severe winds (low confidence) and more dangerous bushfire
conditions (high confidence in southern Australia; medium confidence in eastern Australia)
attributable to increasing greenhouse gas emissions.
1
1. INTRODUCTION AND OVERVIEW

This document presents climate change projections for extreme temperature, wind and
fire weather conditions based on applying a standardised method. This method uses a
comprehensive range of lines of evidence from physical process understanding, observations,
reanalyses and climate modelling. It is designed to be beneficial particularly in cases with many
contributing factors and uncertainties (such as for some extremes and for mean rainfall, wind,
flood, etc.), including for the selection of projections methods and datasets as well as for
confidence assessments. For example, this method can help provide guidance when producing
the projections data products on whether a particular modelling approach could be useful to
include or not (or perhaps weighted differently within a broader ensemble of datasets). That type
of targeted guidance can be used along with other more general sources of guidance relevant to
projections data, such as based on broader assessments of models and methods relating to climate
change projections (CSIRO and BoM 2015; Thatcher et al. 2021). This standardised method is
used here together with a new set of calibrated climate projections for Australia, including for the
first time using three regional modelling approaches for dynamical downscaling, with the aim of
providing the best-available projections information for extreme temperatures, winds and bushfire
danger due to increasing greenhouse gas emissions (as detailed in Sections 3–5).
For some planning and design activities relating to future climate change, decisions will
often need to be made regardless of whether highly confident projections are available or not.
Consequently, there may be benefits in scientists providing information on projections even if
those projections are not highly confident, as long as the degree of uncertainty is assessed and
communicated when those projections are provided. The results presented here are intended to
help underpin such decisions, based on considering a broad range of lines of evidence.
The standardised method for projections information used here can be applied for an
individual weather variable and region, or for a multivariate/compound event (e.g., relating to
bushfire risk factors based on considering a range of different processes). A previous study
provides examples of how this type of method can be applied for one weather variable (mean
rainfall) in four individual seasons (Dowdy et al. 2015). The method is applied here in this study
for extreme values of the following three variables with a focus on summer (December to
February), selected based on discussions with stakeholders on key needs for climate risk
assessments in the energy sector (also noting these extremes are relevant for many other sectors
and purposes for society and environment throughout Australia):
o extreme temperatures, based on daily maximum air temperature at a height of 2 m (Section 3);
o extreme winds, based on 3 second average wind gust speed at a height of 10 m (Section 4);
o extreme bushfire weather, including based on a compound event type of framework and
considering a range of different risk factors (Section 5).
The lines of evidence are also used when producing confidence assessment information
in the projections. The confidence assessment information can include measures such as ranges
of change (e.g., probabilistic estimates of likely ranges that may be above, or below, the most
likely estimate for the projected change) and other approaches such as descriptive terms for
communicating the degree of confidence (e.g., words used here with quantitative probabilities
associated with them such as those used for IPCC (Mastrandrea et al. 2011)).
For information on the likelihood of projected future changes in these weather-related
variables, stakeholder codesign activities recommended a focus on extremes corresponding to the
10-year average recurrence interval (ARI), representing an event with a return period of 10 years
on average (noting that the return period is equal to the reciprocal of the annual probability of
2
exceedance). This included maps of the most likely future projected change in values
corresponding to the 10-year ARI, together with estimates of the 10th and 90th percentile range of
plausible 10-year ARI values as a confidence assessment measure. National maps of those
quantities are presented here, together with confidence assessment information, based on
considering various lines of evidence. The resultant maps and data layers (with supporting
confidence assessment information) are intended for a broad range of user groups including in
sectors for which extreme temperatures, winds or bushfires are relevant. The projections are
presented for the future climate around the middle of this century, as well as for the historical
climate, but are also available for other time periods throughout this century as well as for other
metrics of extreme or average conditions throughout Australia (with data available on request).
The following Section 2 describes the steps for applying the standardised method, then
provides some examples of practical uses of the outputs for improved resilience to climate
hazards. Sections 3-5 document the application of the method for extreme temperature, wind and
fire weather, respectively.
3
2. METHODOLOGY
2.1 Standardised method for projections information

The standardised method consists of two steps, referred to here as producing the Lines of
Evidence Table (Step 1) and then producing the Projections Likelihood Information (Step 2). The
Lines of Evidence Tables are provided to document the supporting science details, as well as to
help guide the production of the Projections Likelihood Information including the confidence
assessment.
Examples of applying these steps are provided in Sections 3-5. For the purposes of this
study, the Projections Likelihood Information is shown as maps for the most probable change in
values corresponding to the 10-year ARI, together with estimates of the 10th and 90th percentile
range of plausible change in those 10-year ARI values (as one measure for providing confidence
information).
Step 1 – Produce the Lines of Evidence Table

 Collect a wide range of information on climate change that could be of relevance to consider
when populating the Lines of Evidence Table. This information could be obtained from new
analyses as well as from a review of existing literature, considering aspects such as
observations, reanalyses, model data and physical process understanding. For example,
relevant aspects to consider could potentially include analysis of long-term observed trends,
model simulations of future climate, uncertainties in observations, uncertainties relating to a
modelling approach’s ability to simulate physical processes and observed features (such as
the seasonal cycle or spatial detail of extremes), as well as the influence of large-scale drivers
(e.g., ENSO, IOD and SAM) in the historical and future projected climates.
 Collate that information into short text summaries on each aspect being considered, with
accompanying figures and references provided to support those summaries, aiming for a
general balance of evidence based on the available science. The summaries can be grouped
into broader categories (e.g., physical processes, historical climate and future climate).
 Use those short text summaries to populate the Lines of Evidence Table. This table contains
a different row for each of the different aspects being considered. Key details can be listed on
each row including the degree of influence that this aspect has on the variable in the region
being considered, as well as what this implies for the direction of projected future change
(colour-coded to show either an increase, decrease, little change or increased uncertainty).
This is intended as a standardised way to help collate and synthesise a broad range of
information.
Step 2 – Produce the Projections Likelihood Information

 For the projected change of interest (e.g., a change from the historical period to a future period
in values corresponding to the 10-year ARI), use the Lines of Evidence Table to determine
the best available data and methods for estimating a given likelihood measure. For example,
likelihood measures could include the most probable projected change, together with
estimates of the 10th and 90th percentile range of plausible change. The method to determine
the best available estimate for a given likelihood measure may vary between different weather
variables of interest (e.g., depending on the degree of confidence in models to simulate
relevant physical processes). For example, this variation could include the selection of
different datasets and methods (e.g., the use of direct model output or statistical diagnostic
methods) or scaling some data differently in a model ensemble.
4
 For quantities that have a reasonably robust range of evidence, with good agreement between
those different lines of evidence (e.g., about two thirds of the Lines of Evidence Table having
a consistent sign of future change), then model output may be the best option for producing
the Projections Likelihood Information, while still considering the various uncertainties and
strengths/weaknesses of different modelling approaches for helping to guide the production
of the data products. For quantities with lower confidence (i.e., more limited evidence and/or
lower agreement between lines of evidence), then a more qualitative best estimate could be
appropriate. For example, in some cases with very high uncertainty the best estimate for the
Projections Likelihood Information might simply be ‘an increase is more likely than a
decrease’ for a particular region, if that is the best information that can be provided based on
the balance of available knowledge from the Lines of Evidence Table. It is acknowledged
that given the broad range of different information sources and data types (e.g., direct model
output or statistical diagnostic methods) this step of the process may require some degree of
expert judgement to be used.
 The Projection Likelihood Information can include confidence assessment information, such
as based on the degree of evidence and agreement from the Lines of Evidence table. For
example, estimates of the 10th and 90th percentile range of plausible change is one measure
that could be used to help indicate the degree of confidence in a projected future change, as
well as noting various other approaches that could be used for some applications, including
the framework shown in Table 2.1 together with various words that have a range of
quantitative probabilities associated with them to accompany the provision of projections.
Additional details on the method

Depending on the intended purpose, the method can be applied for mean values or a
magnitude range of a particular variable of interest (e.g., the likelihood of occurrence for wind
speeds in the range 20–30 m.s-1, and/or > 30 m.s-1, etc.). Similarly, it can be applied for a region
or for individual locations, as well as applied individually for each variable of interest (such as
for extreme rainfall, wind speed, etc.) or used to examine compound events based on multiple
variables in combination with each other. It can also be applied for a particular time period and
greenhouse gas emissions pathway of interest, to help understand the strengths and limitations of
projections information for specific variables and regions in a future projected climate.
The standardised method can enable a likelihood measure (i.e., probability of occurrence)
to be assigned to projections based on considering a comprehensive range of information. This
can be done for different projected values (or ranges) within the full distribution of plausible
change, noting that the total sum of the percent likelihood measures should equal 100%. The
number of different projection ranges selected can be varied depending on the specific application
intended, noting that it will always include at least two ranges (e.g., a projected increase in
temperature with a likelihood estimate of 99% also implies a 1% likelihood estimate of little
change or decrease).
To determine the Projections Likelihood Information for each quantity of interest, model
output is considered together with the other information provided in the Lines of Evidence Table
(i.e., the observations and physical process understanding). The Lines of Evidence Table can help
guide the expert judgement that may be required to produce the projections information. For
example, this could include a greater reliance on direct model output for variables such as extreme
temperatures for which there is typically higher confidence than for variables such as extreme
winds for which a greater reliance on physical process understanding and other lines of evidence
may be practical (e.g., statistical diagnostic methods calibrated to observations data, rather than
the use of direct model output).
5
For this study, the method is applied for the projected change in climate from the time
period 1986–2005 (i.e., a commonly used historical reference period for CMIP5 data (CSIRO &
BoM 2015)) to the time period 2040–2059 (i.e., a time period centred on the middle of this century
around 2050 as requested by energy sector stakeholders). The information collected here for Step
1 for the Lines of Evidence Table is intended to be relevant for the National Energy Market
(NEM) region around southeast Australia, including listing any regional variations that might be
important to consider.
The RCP8.5 scenario, representing a high emissions pathway for anthropogenic
greenhouse gases, is used for the future projections for a number of reasons. Of the set of modelled
greenhouse gas emission pathways provided in CMIP5 (which start to deviate from each other
after 2005), observed climate change trends for temperature indicate that the high emissions
pathway RCP8.5 has been followed more closely than low emissions pathways (e.g., RCP2.6)
(IPCC 2013; Schwalm et al. 2020). Additionally, although there is potential for reductions in
greenhouse gas emissions and the associated rate of temperature increase later this century,
RCP8.5 is used here for the application of this method given that it takes many years after changes
in emissions for an emergent change in a climate trend, noting the focus for this application on
the period from now until around the middle of this century. However, for applications in which
projections are needed based on lower emissions pathways than RCP8.5, methods could be used
for scaling these projected changes according to the global warming magnitude for a particular
time period or emissions pathway, such as has been recently demonstrated (NESP 2020).
Table 2.1: Confidence can be assessed based on the degree of evidence and agreement,
consistent with IPCC guidelines. The degree of confidence can then be used together with the
projections data to help provide likelihood estimates (i.e., probability of occurrence) consistent
with Mastrandrea et al. (2011).
Limited evidence Medium evidence Robust evidence
High agreement Medium Medium-high High confidence

confidence confidence
Medium Low-medium Medium Medium-high
agreement confidence confidence confidence
Low agreement Low confidence Low-medium Medium
confidence confidence
2.2 Examples of method outputs being used

The outputs from applying this standardised method, including the calibrated projections
and confidence assessments, are being used in energy sector applications such as listed below.
The outputs are also intended for use in other sectors, given the relevance of temperature, wind
and fire weather projections to many aspects of society and environment. This includes for
applications such as improved planning and helping to build resilience in relation to the influence
of anthropogenic climate change on future hazards in Australia.
 Enhanced design and planning: The probabilistic projections information for extremes from
the method outputs are being used to help understand the future risk of failure for various
types of infrastructure (e.g., electricity transmission towers), providing important knowledge
for the design and planning of individual components in the NEM and other energy sector
6
applications in Australia. Another example is the inclusion of the10-yr ARI maps for
temperature in AEMO's 2020 Integrated System Plan (ISP).
 Enhanced resilience modelling: Randomised failures are currently used as synthetic input to
energy sector modelling for matching supply and demand, including for assessing future
changes in the resilience of the NEM. As suggested by energy sector groups, the outputs of
the standardised method can be used to refine these failure rates, to help design and plan for
a network that is more resilient to future climate change based on considering a
comprehensive range of evidence.
 Enhanced reliability modelling: The outputs can be used for providing guidance to
accompany the projections data provided as input for the NEM reliability modelling,
including insight on whether some datasets might be preferentially weighted over others for
some variables/regions.
 Enhanced guidance for stakeholders on climate risk and hazard scenarios including
compound events: The comprehensive review and synthesis framework of the standardised
method is being used to help examine some details for compound event scenarios, intended
for use in subsequent risk assessment applications and 'stress testing' activities on climate
hazards.
 Broader applicability: Although the results presented here are primarily intended to meet the
needs of the electricity sector in Australia, they are also intended to have broader benefits
including for other sectors, given the relevance of extreme temperatures, winds and bushfire
conditions to other sectors. These projections and confidence assessments are also providing
a foundation for the initial stages of the Australian Climate Service (ACS) recently
established for producing and providing climate information in Australia.
7
3. EXTREME TEMPERATURE PROJECTIONS
3.1 Introduction
The standardised method for projections information is applied here for extreme values
of daily maximum temperature at a height of 2 m during summer, with a focus on the 10-yr ARI
values for regions around southern and eastern Australia. The application of this method follows
the two steps described in Section 2.1.
For Step 1 of the method, short summaries are presented below (not listed in order of
importance) for different aspects relating to future changes in extreme temperature during
summer, with regional variations noted where relevant. The summaries are then used to populate
the Lines of Evidence Table (Table 3.1), with key details from the summaries noted succinctly in
the rows of that table, including the degree of influence that this aspect has on extreme
temperature and its implied direction of projected change (either an increase, decrease, little
change or increased uncertainty).
For Step 2 of the method, the results from the Lines of Evidence Table are used for
guidance in producing the Projections Likelihood Information. For this study, this includes the
best estimate of the most probable projections for extreme temperature (presented here as maps
showing the 10-yr ARI values) as well as estimates of the 10th and 90th percentile range of
plausible change in the 10-yr ARI values (as a measure for indicating the degree of confidence in
the projections). As discussed in Section 2.1, the RCP8.5 emissions pathway from CMIP5 is
considered relevant for use in providing projections towards the middle of this century, with a
focus here on a historical reference period 1986–2005 and a projected future period 2040–2059.
3.2 Summaries for physical processes

Soil moisture
Through its control on the exchange of water and energy between the land and the
atmosphere, near-surface soil moisture plays a key role in determining air temperature. For
example, drier soils can increase the likelihood of extreme temperatures including as has been
documented for eastern Australia (Perkins et al., 2015; Herold et al., 2016) and northern Australia
(Hirsch et al. 2019). Soil moisture also plays an important role in developing and maintaining
extreme heat as documented for Australian heatwaves (Perkins et al., 2016; Wehrli et al., 2019).
On a daily timescale, soil moisture is highly variable in time and space and depends on a
range of factors such as recent rainfall, vegetation water use and evaporation (Jovanovic et al.
2008; Ukkola et al. 2019). Soil moisture also varies seasonally and can depend on the previous
season's weather conditions and climate states such as large-scale modes of atmospheric and
oceanic variability (e.g., ENSO) and associated weather variations as well as longer-term drought
conditions.
The high level of natural variability of soil moisture in both time and space, as well as
the broad range of factors that can influence soil moisture, makes it challenging to determine
future changes in these quantities based on model simulations. In the coming decades, soil
moisture is projected to decrease on average in many regions of Australia, including in the
southeast where mean rainfall is expected to decrease (particularly during the cooler months of
the year) and atmospheric evaporative demand is expected to increase (CSIRO & BOM 2015;
Berg et al., 2017). For southern and eastern Australia, more frequent periods of dry soil are
projected to occur in the future with a reasonably high degree of confidence, mostly in winter and
8
spring but also summer (CSIRO & BoM 2015; Ukkola et al. 2020), with this higher frequency of
drier soils expected due to higher rates of atmospheric evaporative demand and increased periods
of drought (including meteorological drought defined based on rainfall deficit measures). It is
also noted that there are considerable uncertainties around climate models simulations of how soil
moisture can influence temperature through land-atmosphere coupling processes. For example,
climate models may overestimate the coupling between soil moisture and extreme temperatures
in wet areas of the globe, so potentially overestimate this aspect to some degree relating to
increases in extreme temperatures in some cases (Ukkola et al., 2018). There are also uncertainties
in the influence of climate change on the direction and magnitude of soil moisture change,
including relating to uncertainties in changes to rainfall, potential evaporation and the use of soil
water by vegetation under increasing levels of CO2 (Jovanovic et al. 2008; Ukkola et al. 2020).
In summary, soil moisture can be an important influence on temperature extremes, while
noting some uncertainties in the ability of climate models to simulate some processes that are
relevant for soil moisture. Projections indicate more frequent periods of dry soil moisture on
average in the future during summer in southern and eastern Australia, which will act to increase
the risk of extreme temperatures, with medium confidence.
Cloud cover and solar radiation

Extremely high surface temperatures require strong solar radiation (e.g., downwelling
shortwave radiation near the surface) which can occur during periods of reduced cloud cover.
Conversely, cloud cover can reduce the chance of extreme temperatures. For example, in
California, coastal low clouds have been found to moderate heatwaves, particularly the likelihood
of a heatwave to extend to the coast (Clemesha et al. 2018).
There is a large degree of natural variability in cloud cover and solar radiation, which
makes it challenging to determine long-term changes in these quantities (Jovanovic et al. 2011).
Projections based on global climate models (GCMs) indicate little change or a small increase in
solar radiation in southern and eastern Australia during summer but with considerable variability
between different models (CSIRO & BoM 2015). However, the presence of clouds is a major
area of uncertainty in climate models, including in terms of limitations in accurately simulating
clouds and for the interaction between clouds and other variables like temperature and
atmospheric circulation (Grise & Polvani 2014; Myers & Norris 2015; Voigt et al. 2020).
Additionally, clear skies (i.e., reduced cloud cover) can be associated with the subtropical ridge,
noting that the subtropical ridge is projected to intensify in the future (as discussed in the
following section).
In summary, cloud cover and solar radiation are important influences on the occurrence
of extreme temperature. Future changes for regions around southern and eastern Australia during
summer indicate little change or a small increase, but with low confidence due to high natural
variability and considerable variation between models as well as the limitations of climate models
in being able to accurately simulate clouds. Regional climate models (RCMs) may provide
improvements over GCMs in relation to this aspect, although evidence in the literature is sparse.
Subtropical ridge
An intense subtropical ridge (STR) of mean sea-level pressure is associated with an
increase in the mean maximum temperature and the frequency of days above the 90th percentile
in southern Australia in all seasons (Pepler et al. 2018). This relationship is strongest in winter
and spring, including in southern regions such as Victoria and Tasmania. During summer, an
intense STR is associated with more hot days in Tasmania but fewer hot days on the east coast
including Brisbane. Observations and reanalysis data show the STR has grown more intense in
recent decades, which has contributed to observed declines in southeast Australian rainfall
9
(Timbal & Drosdowsky, 2013), but it is unknown whether the intensification of the STR has also
contributed to past changes in maximum temperature or hot days.
The STR seasonal cycle is relatively well simulated in CMIP5 and is projected, with high
confidence, to intensify in the future (CSIRO & BoM, 2015). In this regard, CMIP5 models
represent a significant improvement over CMIP3 models. Despite confidence in the projection of
STR intensification, it is uncertain how this future intensification will impact future extreme
temperatures in Australia, while noting that the subtropical ridge is typically associated with
descending air and relatively clear skies (i.e., reduced cloudiness) that could potentially be one
contributing factor for increased temperature extremes. Although CMIP5 models have limited
ability to replicate the STR influence on Australian rainfall (CSIRO & BoM, 2015), the STR
relationships with temperature are mostly independent of the STR-rainfall relationships (Pepler
et al. 2018) and it is a current knowledge gap in the literature as to how well the CMIP5 models
replicate the STR relationship with extreme temperatures.
In the Southern Hemisphere, the STR intensification and other measures of tropical
expansion have been linked to climate change (Nguyen et al. 2015; Grise et al. 2019) with some
contribution from ozone depletion in the summer months as well as natural variability (Garfinkel
et al., 2015; Waugh et al., 2015). Climate models consistently project a future southward shift and
intensification of the subtropical ridge (Kent et al. 2013; Grose et al. 2015). However, this may
be masked by the influence of ozone hole recovery during the summer months in coming decades
to some degree (IPCC 2013).
In summary, the STR has historically had a significant influence on the occurrence of
extreme temperatures, with more intense STR associated with hotter summer temperatures
particularly in southern Australia. Although CMIP5 models do a reasonable job of simulating the
STR, including an increase in intensity being likely in the future, the impact of the STR on future
extreme temperatures is somewhat uncertain. As STR is a large-scale feature with links to
broader-scale processes such as tropical expansion, RCMs may offer relatively limited
improvement over GCMs in representing the STR. However, RCMs may be better able to
simulate the impacts of the STR on local climate extremes, due to better simulation of interactions
between the large scale and local factors such as cloud cover.
Cold Fronts
Frontal systems are major drivers of extreme temperature events in southern Australia.
Strong northwesterly winds prior to cold fronts can enhance the advection of extreme heat from
inland Australia towards the southeast regions during summer. Some studies suggest relatively
little change in the frequency of fronts in southeast Australia and a slight decrease in their mean
intensity over recent decades (Rudeva & Simmonds, 2015), while some studies also indicate the
frequency of fronts has decreased in some regions of southeast Australia such as for the eastern
seaboard (Pepler et al. 2021).
Climate models are generally able to simulate the average annual frequency of fronts in
the Australian region during winter, but relatively few studies have examined this during summer
(Catto et al. 2015; Blázquez & Solman 2017). Climate model projections have a weak increase in
the frequency and intensity of fronts in southern Australia, but the available projections do not
distinguish the cold fronts associated with northwesterly winds from warm fronts and stationary
fronts (Catto et al. 2014; Blázquez & Solman 2019). Using the older CMIP3 climate models, a
simple temperature-based proxy for very extreme cold fronts associated with summer temperature
extremes and bushfires indicated a likely future increase in the frequency of frontal systems
(under both medium and high emissions scenarios), increasing from ~0.5 events per year in the
current climate to 1-2 events per year by the end of the 21st century (Hasson et al. 2009).
Considering studies such as these, considerable uncertainties remain in relation to extreme
10
temperature events associated with fronts in the during summer and how these events could
potentially change in the future.
In summary, observations indicate frontal activity has undergone little change in southern
and eastern Australia during summer. A future projected increase appears more likely than a
decrease in the number of fronts that occur in southeast Australia, but there is considerable
uncertainty for future projected changes in frontal systems and their impact on extreme
temperatures based on the current knowledge. Given that fronts are synoptic-scale systems which
GCMs can simulate reasonably well, there may not be a large benefit from using RCMs to
examine future frontal system activity compared to other factors like clouds and solar radiation.
However, RCMs could potentially provide value for some aspects relating to fronts such as their
interaction with terrain and associated extreme weather impacts for localised regions in some
cases.
Blocking / high pressure systems

For southeast Australia, anticyclones (high pressure systems) are typically associated
with cool southerly winds to the east of the high-pressure system and warm northerly winds to
the west. A persistent and slow-moving (‘quasi-stationary’) high pressure system in the Tasman
Sea is often referred to as a blocking high and can cause extreme heat in southeast Australia
(Marshall et al. 2013; Boschat et al. 2015; Gibson et al. 2017).
GCMs are generally able to simulate anticyclones, as they tend to be large-scale systems,
but may potentially underestimate their persistence and the frequency of long periods of ‘quasi-
stationary’ blocking (Woollings et al. 2018). Anticyclones tend to be stronger and slightly further
south in CMIP5 projections of future climate, but with a future weakening of the overall pressure
couplet that can lead to heat waves in southern Australia (Purich et al. 2014). Patterson et al.
(2019) reported no significant change in blocking in the future projected climate for Australia.
In summary, blocking / high pressure systems, particular in the Tasman Sea region, can
influence the occurrence of extreme heat events in regions around southern and eastern Australia
during summer. There are considerable uncertainties around how these systems might change in
the future and what effect that might have on extreme heat events, with little change projected in
general based on recent studies. GCMs can provide a reasonable representation of some of the
larger-scale pressure features relevant to the advection of hot air from further inland over the
continent, while noting some blocking events can be better represented by finer resolution models
(Dawson et al. 2012).
Tropical cyclones
The occurrence of tropical cyclones (TCs) in northern Australia has been linked with the
intensification of heat extremes in southern Australia, including in southeast Australia during
summer (Parker et al. 2013; Quinting & Reeder 2017; Quinting et al. 2018). For example, the
extreme heat experienced around the time of the Black Saturday fires in 2009, which set new
temperature records for daily maximum air temperature for Melbourne and surrounding locations,
was associated with the presence of a TC (Parker et al., 2013). Observations indicate a decrease
in occurrence frequency of TCs for the Australian region over recent decades (Dowdy 2014;
Chand et al. 2019).
Future projections of TCs during summer for the Australian region indicate a small
decrease in their frequency (medium confidence) (Bell et al. 2019). However, the frequency of
intense category 4 and 5 TCs may not change or increase slightly, along with some poleward
migration (low confidence) (CSIRO & BoM 2015; Knutson et al. 2020; NESP 2020). In general,
GCMs have insufficient spatial and temporal resolution to adequately simulate tropical cyclones.
11
RCMs generally have finer resolution and better resolve tropical cyclones, although RCMs still
do not fully capture all relevant processes. For this reason, additional methods for cyclone
projections can also be useful to consider, such as synthetic cyclone tracks, in addition to dynamic
modelling.
Modes of variability - ENSO

The relationship between the El Niño-Southern Oscillation (ENSO) and temperature
extremes is complex. El Niño years are also associated with reduced cloud cover leading to higher
temperatures and an increase in the temperature of the hottest day of the year across most of
Australia (Arblaster & Alexander, 2012). Across most of northern and eastern Australia, the
frequency, duration and amplitude of heatwaves increases during El Niño years (Perkins et al.
2015; Loughran et al. 2019). However, in parts of the southeast including Victoria, there are
weaker relationships between ENSO and heatwaves (Parker et al. 2014; Perkins et al. 2015).
Although correlations between mean temperature and ENSO conditions have been
examined in numerous previous studies, this has not been examined in much detail for more
extreme measures of temperature. To help address that knowledge gap, correlations are presented
here in Fig. 3.1 for ENSO, as well as for SAM and IOD (relating to subsequent sections below).
The general patterns of correlation (indicating the strength of the relationship with ENSO) are
broadly consistent for mean and extreme temperatures, indicative of higher temperatures in
general occurring for El Niño than La Niña conditions.
There are considerable uncertainties around how ENSO conditions (including extreme
ENSO events) may change later this century based on GCMs (CSIRO & BoM 2015). Projections
of an increase in frequency of ENSO events being sensitive to the model used (Freund et al 2020)
and frequency of extreme ENSO events sensitive to the definition used (Marjani et al. 2019). As
the teleconnections between ENSO and Australian rainfall and temperatures have varied over
time (Power et al. 1999), the strength of these relationships may also change in the future (Fasullo
et al. 2018). However, some studies have suggested there might be an increase in the number of
strong El Niño and La Niña events in future (Cai et al. 2018a). As modes of variability such as
ENSO are phenomena generated in association with very large-scale atmosphere-ocean
interactions, RCMs do not provide benefits over GCMs in simulation how the modes of variability
may change in the future. However, RCMs may provide further detail on how modes of variability
influence local and regional climate, including cloud cover. In fact, RCMs have been shown to
capture the historical teleconnection between ENSO and Australian maximum temperatures quite
well (Fita et al. 2016).
In summary, the influence of ENSO on future extreme temperature events involves
considerable uncertainties. The uncertainties in ENSO simulation are not able to be resolved
through the use of currently available RCM data (e.g., no coupled RCM simulation has been
performed to date over a domain large enough to encompass the processes leading to ENSO) but
RCMs may help in simulating local responses to large-scale drivers such as ENSO.
12
Figure 3.1: Correlations between temperature and climate measures. This is presented in the
upper row of panels for daily maximum temperature (using average summer values for the
months December to February: DJF) and measures representing different modes of variability
including ENSO (using the NINO3.4 index), SAM (using the SAM index) and IOD (using the
DMI index). Similar correlations are also shown in the lower row of panels, but for the number
of days with temperature above the 99.5th percentile during summer. These correlations are all
based on the period from 1979 to 2019, using one value for each summer period (DJF).
NINO3.4 and DMI data are attained from the NASA ESRL
(https://psl.noaa.gov/gcos_wgsp/Timeseries/) while SAM data are from
https://legacy.bas.ac.uk/met/gjma/sam.html. Pearson's correlation coefficient, r, is shown with
stippling corresponding to statistically significant values at the 95% confidence level (2-tailed).
Modes of variability – IOD

The Indian Ocean Dipole (IOD) mode of variability mostly influences Australian weather
during the winter and spring, so it has little relationship with extreme heat during the summer
months in general (Perkins et al. 2015), as well as noting interactions between the IOD and ENSO
(Cai et al. 2019). The relationship between IOD and average values of daily maximum
temperature is broadly similar to that for the more extreme values of daily maximum temperature,
with positive correlations through southern and eastern Australia in general (Fig. 3.1). There is
some indication that extreme positive IOD events may become more frequent in the future (Cai
et al. 2018b) but there is considerable uncertainty in the ability of climate models to simulate such
events (CSIRO & BoM 2015).
Modes of variability - SAM

The Southern Annular Mode (SAM) is a large-scale alternation of atmospheric mass
between the middle and high latitudes. The positive phase is associated with a higher-than-normal
mean sea level pressure in middle latitudes and lower pressure in high latitudes. During a positive
phase of the SAM there is a southward shift for the belt of westerly winds that circles Antarctica,
while the opposite occurs during the negative phase. The La Niña phase of ENSO increases global
mean temperature and can contribute to a negative shift in the SAM (Wang & Cai 2013).
13
Positive SAM is associated with a decreased likelihood of extreme heat during the spring,
but correlations are more mixed during the summer months (Hendon et al. 2007; Marshall et al.
2013; Perkins et al. 2015). The relationship between SAM and average values of daily maximum
temperature during summer is broadly similar in spatial patterns (e.g., sign of correlation, from
Fig. 3.1) to the case for the relationship between SAM and the occurrence of more extreme values
of daily maximum temperature, with generally weak correlations or a negative correlation in
central eastern regions (particularly for mean temperature). A strong negative SAM is also
associated with sudden stratospheric warmings (as occurred in the 2019 Austral spring), which
can cause extreme heat during spring and early summer (Lim et al. 2019), potentially associated
with some of the negative correlations apparent in Fig. 3.1 for the central east region.
SAM has been becoming more positive in recent decades, particularly during the summer
months (Marshall, 2003), which has been linked to a combination of increased greenhouse gases
as well as ozone depletion and natural variability (Garfinkel et al. 2015; Waugh et al. 2015).
CMIP5 models project a robust shift towards more positive values of SAM in all seasons during
the 21st century (Lim et al. 2016), although this may be masked to some degree by the influence
of ozone hole recovery during the summer months in coming decades (Banerjee et al. 2020). In
summary, climate models can simulate SAM well, but projections of a positive trend in SAM
would likely cause little change in the risk of heat extremes during summer apart from potentially
central east (noting a negative correlation with temperature as well as links with sudden
stratospheric warmings (Lim et al. 2019) for which future projected changes are not currently
known).
Modes of variability – MJO

The Madden-Julian Oscillation (MJO) is the dominant mode of atmospheric intra-
seasonal variability and the cornerstone for sub-seasonal prediction of extreme weather events
(Wang et al. 2019). Extreme heat in south-eastern Australia is more common during MJO phases
2 and 3 in spring and phases 3-6 in summer (Marshall et al. 2013; Parker et al. 2014). The
influence of climate change on the MJO is uncertain, with less confidence in changes in MJO-
related wind and circulation anomalies than for rainfall (Maloney et al. 2019), noting that CMIP5
GCMs are not able to provide a good representation of the MJO (CSIRO & BoM 2015).
Consequently, this remains an uncertain factor in relation to extreme summer heat in the future
including for southern and eastern Australia.
Urban effects including urban heat island

The temperatures in urban environments are typically warmer than the surrounding rural
areas, particularly at night. This is a consequence of changes to many surface properties which
alter the surface energy budget, in addition to the presence of additional sources of anthropogenic
heat. The additional overnight heat can contribute to enhanced heat stress on urban populations,
although this may be partially counteracted by lower humidity (Fischer et al. 2012; Williams et
al. 2012). While some studies have suggested that the urban heat island (UHI) is more intense
during hotter conditions, this varies between studies and between different areas of the world
(Scott et al. 2018, Zhao et al. 2018, Chew et al. 2020). Due to the small spatial scale of cities and
the complexity of their terrain, these are typically only well simulated in high resolution regional
downscaled simulations, not coarse GCMs (Argueso et al. 2015; Wouters et al. 2017).
The UHI effect adds a few degrees to temperatures over urban environments (Gartland
2011). This has been shown over the largest cities in Australia including Sydney (Argueso et al.
2014), Melbourne (Imran et al. 2019), Brisbane (Chapman et al. 2019) and Adelaide (Guan et al.
2016). The UHI has been found to exacerbate temperature extremes at night during heatwaves in
these cities (Argueso et al. 2015; Imran et al. 2019; Rogers et al. 2019). Daytime maximum
14
temperatures during heatwaves reflect the standard UHI addition to the temperature of the
surrounding areas. The increased night-time temperatures mean that systems have less
opportunity to cool overnight which poses a hazard for some systems including human health.
Cities will likely experience similar temperature increases due to global warming as their
surrounding regions but will remain warmer due to the UHI. It is uncertain whether the intensity
of the UHI will change as the planet warms, with any changes sensitive to changes in other factors
such as green space (i.e., vegetated areas including tree cover), soil moisture and circulation
(Fischer et al. 2012; Zhao et al. 2018). However, in regions which are currently on the urban
fringe, future population growth and urban expansion is expected to result in additional increases
in hot extremes beyond that expected from climate change alone (Argueso et al. 2015; Wouters
et al. 2017). In summary, the UHI effect means that extreme heat events are more severe in urban
regions, regardless of climate change, and urban areas are often not well simulated in coarse
resolution GCMs, although this can be better resolved in RCMs with dedicated urban
parameterisations. It is unclear if this effect will change in the future, but future warming is
expected to be larger in areas which are also experiencing urbanisation.
3.3 Summaries for historical climate

Observed trends
Extreme temperature events have been steadily increasing in frequency and intensity
throughout Australia, due to increases in atmospheric concentrations of greenhouse gases,
including shifting the full frequency distribution for temperatures towards higher values (CSIRO
& BoM 2015; BoM & CSIRO 2020). The number of extreme heat records in Australia has
outnumbered extreme cool records by about 3 to 1 since 2001 for daily maximum temperatures
(BoM & CSIRO 2020), characteristic of a shift in the full distribution of temperature values due
to anthropogenic global warming. In parts of southeast Australia, the hottest summer days have
increased by a larger degree than expected from the change in mean temperatures alone (Gross et
al. 2019). Heatwave events have also increased in intensity, frequency and duration across
Australia in recent decades (Perkins-Kirkpatrick et al. 2016). The 2019 year was Australia’s
hottest on record, as well as having 42 days when the Australian area-averaged daily mean
temperature was above the 99th percentile (which also set a new record for that measure of extreme
temperatures for individual days).
Model assessment for historical climate

The ability of climate models to simulate aspects such as the seasonal cycle, observed
trends, spatial detail and extremes is important for helping to understand the degree of
confidence in future projected changes based on these models. Assessments as presented in
CSIRO & BoM (2015) indicate that global models provide a reasonably good representation of
these aspects, including regional and seasonal temperature variations through Australia as well
as the observed trends.
3.4 Summaries for future climate

Several datasets are available for future projections of values corresponding to 10-year
ARI of daily temperature. These datasets have all been calibrated using the quantile matching for
extremes (QME) method described in Dowdy (2020b). The datasets provide a 16-member
ensemble comprising of the following:
15
- Dynamical downscaling using the CCAM modelling approach (conformal cubic atmospheric
model; Thatcher and McGregor (2011)) applied to 5 GCMs (ACCESS1-0, CanESM2, GFDL-
ESM2M, MIROC5 and NorESM1-M);
- Dynamical downscaling using the recently developed BARPA modelling approach (Bureau
of Meteorology Atmospheric Regional Projections for Australia) applied here to one GCM
(ACCESS1-0 GCM for eastern Australia);
- Dynamical downscaling using the NARCliM modelling approach (NSW and ACT Regional
Climate Model; Evans et al. 2014) applied to 3 GCMs (ACCESS1-0, ACCESS1-3 and
CanESM2) with 2 configurations of each (providing 6 different ensemble members);
- Calibrated data based on the QME method applied to four GCMs (ACCESS1-0, CNRM-
CM5, GFDL-ESM2M and MIROC5 GCMs).
For further details on the selection and assessment of these models see CSIRO & BoM
(2015) and Thatcher et al. (2021). It is generally recommended to consider results from a broad
range of modelling approaches (rather than only relying on a single method) when trying to
sample the uncertainty space for plausible future changes, such that a focus on this report is on
the combined results from this 16-member ensemble of calibrated projections datasets (i.e., 5
from CCAM, 1 from BARPA, 6 from NARCliM and 4 from GCMs).
To calculate the values corresponding to the 10-year ARI, a Generalised Extreme Value
(GEV) approach was used. This is based on 20-year time slices: using 1986–2005 for the historic
period and 2040–2059 for the future climate projection for the RCP8.5 emission pathway (noting
that these projections data are also available for other time periods throughout this century and
historical periods, as well as for RCP4.5).
The projections from these different modelling approached are presented in Fig. 3.2 (i.e.,
based on the GCMs, CCAM, BARPA and NARCliM ensembles, all with QME calibration
applied). The projections for each of these modelling approaches show clear increases in extreme
temperatures for the future climate.
In addition to these results based on CMIP5, some results have recently been published
based on some CMIP6 projections (Grose et al. 2020). Those results show broadly similar changes
for temperature extremes in Australia to those based on CMIP5 projections, noting that
subsequent studies will continue to examine this further including based on a larger set of CMIP6
models than was available for that study.
16
Figure 3.2: Projected change in values corresponding to the 10-year ARI for daily maximum
temperature at a height of 2 m. This is shown based on GCMs (left panels), CCAM (second to
left panels), BARPA (second to right panels) and NARCliM (right panels), all calibrated using
the QME method. Maps are shown for Australia based on the model ensemble average in each
case. This is presented for the historical climate based on 1986–2005 (upper panels) and future
simulated climate based on 2040–2059 under a high emissions pathway RCP8.5 from CMIP5
(lower panels).
3.5 Lines of evidence table

Table 3.1: Lines of Evidence Table for extreme daily maximum temperature at a height of 2 m,
with a focus on summer around southern and eastern Australia. The degree of influence is listed
in black, followed by whether this information implies an increase (red), decrease (blue) or little
change (black) in extreme temperature, as well as by increased uncertainty (purple) in the
direction of change. The rows of information are not in order of importance.
Physical processes
Soil moisture Moderate influence. More frequent dry soil with medium confidence.
Influence on temperature potentially overestimated. Regional models
likely to add value.
Cloud cover and Moderate influence. Low confidence in little change or a small
solar radiation increase. Regional models likely to add value.
Subtropical ridge Moderate influence, primarily in southern Australia. Potential increase

with low confidence in future influence on extreme temperature.
Fronts Moderate influence. Future change uncertain.
Blocking / High Moderate influence. Future change uncertain.

pressure systems
17
Tropical cyclones Small influence. Fewer in the future (medium confidence); regional
models likely to add value.
ENSO Small to moderate influence. Uncertain future change; potentially more

frequent strong El Niño events (low-medium confidence).
IOD Small to moderate influence. Uncertain future change; potentially more

frequent strong IOD events (low-medium confidence).
SAM Small to moderate influence. Positive trend in SAM relevant for

northeast region temperatures (medium confidence).
MJO Small influence. Uncertain future change.
Urban effects Important for local heat extremes. Urban heat island adds a few degrees
and stays reasonably consistent in future (high confidence); increased
temperature extremes in areas of future urban growth.
Historical climate
Seasonal cycle Models reproduce the seasonal cycle and spatial variability (high
confidence).
Historical trend Strong increase from observations (high confidence). Models reproduce
the trend well (high confidence).
Future climate
GCMs: CMIP5 Strong increase (high confidence).

and CMIP6
RCM: CCAM Strong increase (high confidence).
RCM: NARCliM Strong increase (high confidence).
RCM: BARPA Strong increase (high confidence). Based on one model to date.
Convection- Uncertain future change due to lack of available data and analysis.
permitting models
3.6 Projections and confidence information

The Lines of Evidence Table (Table 3.1) shows considerable agreement on increased
extreme temperatures in a warming climate, including 10-year ARI daily maximum temperatures
in regions around southern and eastern Australia during summer as is a key focus here. Although
there are some physical processes noted that add uncertainties (purple text in Table 3.1),
particularly based on GCM projections data, the RCM approaches (CCAM, BARPA and
18
NARCliM) can help with the simulation of some of these processes. Therefore, the relatively high
level of agreement between RCM approaches helps add some confidence for projected future
increases. Based on this overall assessment considering this wide range of factors, there is very
high confidence in the projected direction of change, with a future increase in 10-year ARI
temperatures being very likely (i.e., 90-100% probability).
Based on the above points and details in the Lines of Evidence Table, projected changes
for 10-year ARI temperatures for the 2050 climate are considered here based on the 16-member
ensemble of calibrated projections datasets, combined based on equally weighting each member
of this ensemble. The ensemble median is used as a central estimate of the most probable
projected change (Fig. 3.3). As an estimate of the range of plausible values from the 16 ensemble
members, the second lowest value from the ensemble is used for the 10th percentile and the second
highest value from the ensemble is used for the 90th percentile. These values are calculated
individually at each grid cell location for the median and percentile estimates.
The results show that the future projected temperatures are higher than for the historical
period, including for the lower estimate corresponding to the 10th percentile of the model
ensemble in the future, as well as for the median and upper estimate (90th percentile). This
highlights the considerable degree of agreement between these diverse modelling approaches.
Figure 3.3: Projected values corresponding to the 10-year ARI for daily maximum temperature
at a height of 2 m, based on a 16-member ensemble of calibrated model projections. Maps are
shown through Australia for the historical period (based on 1986-2005; upper panel), as well as
for the future simulated climate (based on 2040-2059 under a high emissions pathway RCP8.5:
lower panels) including a central estimate with lower and upper estimates also provided.
19
4. EXTREME WIND PROJECTIONS
4.1 Introduction
The standardised method for projections information is applied in this section with a
focus on extreme winds during summer (DJF) in regions around southern and eastern Australia.
Destructive winds in Australia can be caused by severe thunderstorms (mesoscale weather
systems characterised by strong and deep moist convection) and by larger-scale synoptic systems
such as tropical cyclones or extratropical cyclones (including east coast lows) and associated
frontal systems. In particular, severe thunderstorms have been responsible for most of the surface
wind gusts which exceed the 10-year ARI near the major population centres including in southern
and eastern Australia (Holmes 2002), such that the focus here is on severe convective wind gusts,
with other synoptic-scale phenomena also considered here in some sections for completeness.
For Australia, wind gusts are defined by a 3-second average wind speed. Severe
convective wind gusts (SCWs) are considered for the purposes of this study as exceeding 25 m.s-
1
, at a height of 10 meters above ground level, caused by thunderstorm outflow. This threshold
(equivalent to exceeding 90 km.hr-1) is consistent with the threshold used for severe weather
forecasting and operational warnings produced by the Australian Bureau of Meteorology. While
gusts of around 25 m.s-1 may not always be destructive, it is noted that this definition is based on
exceeding that value and therefore also includes higher wind speeds (e.g., around 45 m.s-1) which
have a higher chance of causing property damage. This covers a range of ARI values consistent
with wind speeds such as provided in current Australian standards, including spanning a range
broadly similar to the values for the 10-yr ARI in southern and eastern Australia assuming flat,
open terrain (Holmes, 2002). The atmospheric environments which produce this range of wind
gusts (roughly around 25-45 m.s-1) are typically characterised by unstable atmospheric conditions
(i.e., conducive for convection) as well as likely to include conditions favourable for convective
organisation which can lead to increased severity of hazards (such as can be associated with strong
wind shear between vertical levels (Taszarek et al. 2017)). Tornadoes are a special class of severe
convective winds that are not considered here, including due to their very rare occurrence at a
given location and their very small spatial scale, as well as noting that the design standards widely
used in Australia do not intend structures to withstand the occurrence of a tornado.

Thunderstorm environments
Environments conducive for thunderstorm occurrence are often associated with unstable
atmospheric conditions (based on the vertical profile of temperature and moisture), while severe
thunderstorms may also require other contributing factors such as vertical wind shear (that is
when the wind changes in speed and/or direction with height) which can sometimes help organise
the structure of a severe thunderstorm (Brooks et al. 2003; Taszarek et al. 2017). Globally, the
vertical temperature lapse rate (the rate of temperature decrease with height) is predicted to
decrease/stabilise (increase/destabilise) into the future in the extratropics (tropics) due to different
rates of warming in the lower atmosphere compared to the upper atmosphere (Bony et al. 2006),
while atmospheric moisture content is predicted to increase by about 7% per degree of warming
based on the Clausius-Clapeyron relation (IPCC 2013). Vertical wind shear is predicted to
decrease in the global mid-latitudes due to reduced zonal surface temperature gradients via the
thermal wind relation (IPCC 2013; CSIRO & BoM 2015).
20
Combining these factors through the use of environmental thunderstorm diagnostics

applied to model data, the frequency of thunderstorm environments has been projected to increase
during the coming century in the United States (Trapp et al. 2007; Diffenbaugh et al. 2013;
Gensini et al. 2014; Seeley & Romps 2015) and Europe (Púčik et al. 2017), likely driven by
increases in atmospheric moisture content resulting in increases to convective available potential
energy. This is similar to results for eastern Australia during the warm season (Allen et al. 2014),
noting various model uncertainties remain unquantified for the Australian region, as well as a
need for additional studies using a broader range of models and methods.
Historical increases in the frequency of thunderstorm environments have been indicated
by reanalysis data for some near-coastal parts of southeast Australia, but with decreasing
frequency overall for most regions of Australia (Dowdy 2020a), while noting those results were
for thunderstorm activity in general rather than specifically focused on severe thunderstorms that
can cause SCWs. Historical increases in thunderstorm environments have been reported for
Europe (Rädler et al. 2018), although trends are less certain in North America, which may
partially be due to increasing convective inhibition (CIN) offsetting increases in convective
instability (Taszarek et al. 2020) as a factor which limits thunderstorm development.
A recent study indicates CIN projected to increase over most land areas in the future
(Chen et al. 2020). Some regional projections studies in the United States have also noted that
CIN is likely to increase in a future climate, which could contribute to offsetting increases in
available convective energy as discussed above (Hoogewind et al. 2017; Rasmussen et al. 2017).
However, CIN could potentially decrease on days with high amounts of instability (Diffenbaugh
et al. 2013) as well as noting that CIN tends to be poorly resolved in large-scale dynamical models
due to issues in representing fine-scale features of the vertical temperature profile (King &
Kennedy 2019), with future changes in CIN representing one of key uncertainties in thunderstorm
projections.
Overall, there is low confidence in an increasing frequency of favourable environments
for severe thunderstorms during summer in Australia, including based on results from other
regions and the work of Allen et al. (2014) for projections of future changes in Australia (while
noting that is based on a relatively limited range of modelling approaches). There are considerable
uncertainties around this such as discussed in the examples above, including around the role of
individual environmental conditions in a changing climate (e.g., CIN). Additionally, favourable
environmental factors are necessary but not sufficient for thunderstorm occurrence, given that
additional factors are required for SCW occurrence (such as also depending on initiating
mechanisms, microphysical processes, etc.).
Severe convective wind environments

In addition to the thunderstorm environmental factors mentioned above, there are
additional environmental factors which can be conducive to SCW production. SCWs can be
formed due to intense downdrafts within thunderstorms, with the downdrafts initiated due to the
evaporative cooling of precipitation which causes cold, dense air to accelerate downwards, also
aided by the weight of the precipitation itself. Downdrafts which reach the surface will transfer
momentum into the horizontal, causing severe wind gusts. This process can be associated with a
range of environmental factors including a relatively dry lower atmosphere combined with a steep
temperature lapse rate as well as strong environmental wind speeds (Proctor 1989; Kuchera &
Parker 2006; Brown & Dowdy 2021), although the relative importance of these may vary with
convective mode (Doswell & Evans 2003; Smith et al. 2012). It follows that the variability of
SCWs on climate timescales may be different to thunderstorms in general (Brooks 2013).
The impact of climate change on individual convective hazards, such as severe surface
winds, is highly uncertain (Allen 2018). However, recent work in Australia has suggested the
21
potential for increases in the frequency of severe convective wind environments into the late
century (Spassiani 2020), which is similar to historical findings for Europe (Rädler et al. 2018).
There have also been future projections of severe convective wind speeds for Tasmania (Cechet
et al. 2012), applying a severe thunderstorm diagnostic to historical observed wind speeds. In
addition to considering such studies, here we also present projections of SCW environments,
following the method of Brown & Dowdy (2021), as detailed in Section 4.4.
Thunderstorm initiation
Given an environment favourable for severe convection (i.e., thermodynamically
unstable conditions, including such as described in the above subsections), thunderstorm initiation
can be triggered in various ways including associated with the occurrence of synoptic systems
(cyclones, fronts and jet streams), atmospheric waves, orographic influences (sea-breezes and
mountains) and small-scale moisture fluctuations (Weckwerth 2000). There are considerable
uncertainties for the influence of climate change on triggering mechanisms such as these for
thunderstorm initiation, with some details as follows.
Projection studies tend to indicate that changes in synoptic initiation mechanisms such as
mid-latitude extratropical cyclones (including east coast lows: ECLs) are not clear for Australia
during the summer months (Catto et al. 2014; Pepler et al. 2016; Dowdy et al. 2019a). Cyclone-
related convection is sensitive to coastal sea surface temperatures (SSTs) which can be a source
of warm and moist air relevant for thunderstorm occurrence (Chambers et al. 2015), noting that
the Tasman Sea east of Australia is a region of accelerated ocean warming including with the
strengthening of the East Australia Current (EAC) (CSIRO and BoM 2015) and that this region
has shown an increase in thunderstorm activity based on environmental conditions over recent
decades from reanalysis data (Dowdy 2020a). Projections related to fronts were discussed in
Section 3.2, indicating considerable uncertainty, with little or no change being the most plausible
outcome.
There is relatively little information on long-term changes to orographic flows such as
sea breezes; however, the strength of the sea breeze is strongly related to the land-sea temperature
contrast, which is expected to increase into the future. One study found an increase in the
frequency and intensity of sea breezes in Adelaide between 1955-2007 (Masouleh et al. 2019).
Regional model simulations at 20 km resolution have been shown to provide a reasonable
simulation of the sea breeze in the Mediterranean region (Drobinski et al. 2018), although
convective parameterisations are less skilful in simulating sea breeze-related CIN (Birch et al.
2015).
In summary, there are a range of processes that are important for thunderstorm initiation.
However, the influence of climate changes on those processes is highly uncertain, including
during summer in southern and eastern Australia. SST increases might potentially help provide
enhanced moisture sources (noting the strengthening EAC due to climate change), although there
is a need for further research to understand how relevant that association might be between the
EAC thunderstorm activity.
Modes of variability – ENSO, IOD and SAM

Details on modes of variability were provided in Section 3, including in relation to the
potential influence of climate change on ENSO, IOD and SAM. Building on that information,
aspects specifically relating to SCWs are summarised here.
Thunderstorm environments are not significantly related to ENSO conditions in general
for Australia, apart from in northern Cape York Peninsula where they are more likely during La
Niña than El Niño conditions (Allen & Karoly 2014; Dowdy 2016, 2020a). However, it is still
22
feasible that ENSO may potentially modulate convective initiation mechanisms in some regions.
For example, reduced cloud cover and enhanced sea-breeze circulation in south-east Queensland
during El Niño conditions might potentially increase the frequency of severe thunderstorm events
(Soderholm et al. 2017). It is likely that there is not a strong relationship between ENSO and
synoptic-scale initiation mechanisms, including little or no relationship found between ENSO and
fronts in southern Australia or between ENSO and ECLs in eastern Australia (Rudeva &
Simmonds; Power and Callaghan 2016; Dowdy et al. 2019a). The relationship between ENSO
and SCW environments is shown here in Fig. 4.1a, suggesting very little relationship with ENSO
in eastern Australia during the summer (e.g., only very small regions with significant correlations,
noting that 5% of the region would have a significant correlation on average due to random chance
alone given the use of the 95% confidence level). In summary, the influence of ENSO on SCWs
appears to be relatively weak while noting considerable uncertainties based on limited data and
analysis to date. This is also the case for the relationship between ENSO and severe thunderstorm
occurrence, as well as between ENSO and synoptic initiation mechanisms (including fronts and
cyclones in southern and eastern Australia during summer).
The IOD does not appear to have a strong influence on thunderstorm activity in Australia
during summer (including in southern and eastern Australia) as detailed in Dowdy (2020a), while
noting that study was not specifically focussed on severe thunderstorms which could potentially
have different characteristics to thunderstorms in general. The influence of the IOD on severe
thunderstorms in Australia is currently uncertain based on a lack of previous analyses, although
the IOD may relate to extreme wind gust variability in general, with potential for higher
occurrence frequencies during negative IOD phases (Azorin-Molina et al. 2021). The influence
of the IOD on SCW environments is not significant during the summer in southeastern Australia
but a significant negative correlation is show in Fig. 4.1c in northeast regions.
Similar to the IOD and ENSO, the influence of SAM on severe thunderstorms in Australia
is largely uncertain. No consistent relationship has been found previously with thunderstorm
environments (Dowdy 2020a). From Fig. 4.1b it appears that the negative phase of SAM is more
conducive than the positive phase for SCW environments in eastern Australia. In addition,
enhanced westerlies and associated cold fronts during the negative phase of SAM (Rudeva &
Simmonds, 2015) may increase the frequency of synoptic initiation mechanisms in some southern
regions, and extreme wind gusts from station data have also been shown to be more frequent in
this phase (Azorin-Molina et al., 2021).
Figure 4.1: Correlations for summer between the number of days with a favourable severe
convective wind environment and seasonally-averaged indicators of a) ENSO (Niño3.4 index)
23
b) IOD (Dipole Mode Index) and c) SAM (Marshall Index) for 1979-2018. The thunderstorm
environments are calculated from the ERA5 reanalysis (Hersbach et al. 2020) based on the
method of Brown & Dowdy (2021). Hatched regions indicate a significant relationship at the
95% confidence level (e.g., about 5% of the region could be expected hatched on average due to
random chance alone). NINO3.4 and DMI data are attained from the NASA ESRL
https://legacy.bas.ac.uk/met/gjma/sam.html.
Other phenomena that can cause severe wind gusts

Phenomena other than thunderstorms can produce severe wind gust speeds in some cases,
including TCs in the more northern regions of Australia (with relatively little influence on central-
east regions of Australia), as well as ECLs in near-coastal regions in the southeast and central-
east regions of Australia while noting that the most damaging ECLs typically occur during the
cooler months of the year (which reduces their relevance to this study's application here for
summer). Long-term climate trends in the occurrence of TCs and ECLs and associated severe
wind gusts during the summer months are briefly discussed here, including in relation to a
changing climate, while noting that the primary focus of the analysis here is on severe
thunderstorms for the purposes of this study.
Fewer ECLs are projected in a warming world, but with higher confidence during the
cooler months of year and more uncertain changes projected in the future occurrence of ECLs
during summer (Dowdy et al. 2019a). This includes large uncertainties around the projected
change in the intensity of intense ECLs during summer (i.e., those with extreme wind speeds).
There has been a significant downward trend in the occurrence frequency of TCs
observed for the Australian region as a whole (Dowdy 2014; Chand et al. 2019). For the east coast
of Australia, there has been no change in severe landfalling TCs (Chand et al. 2019), with an
increase suggested by Holmes (2020) primarily since 2011 and mostly evident between
Townsville and Rockhampton (noting that this is a relatively short time period for climatological
assessments of rare events with large interannual variability such as these). Wang et al. (2013)
reported that structures along the north-east coast of Australia may already be subject to higher
gust speeds than the current design standard permits, with projected changes in severe wind gust
speeds being sensitive to TC frequency and intensity change, particularly between Cairns and
Townsville.
Future projections based on global models indicate a downward trend in the occurrence
frequency of TCs in the Australia region (Bell et al. 2019). However, the currently available range
of climate models have large uncertainties in their simulations to identify the more intense and
damaging TCs (e.g., Category 4-5) such that there is considerable uncertainty in future changes
in damaging wind speeds associated with them (Knutson et al. 2020). A recent review that
considered observations and future projections concluded that the frequency of Category 4 and 5
TCs may not change or increase slightly along with some poleward migration or little change in
their spatial extent being plausible future outcomes, but with considerable uncertainties, as
detailed in NESP (2020).
To summarise for TCs, the rareness of landfalling category 4-5 TC events and relatively
short historical time period for high-quality observations, as well as the limited ability of climate
models to simulate such systems, means that there is considerable uncertainty around the
influence of climate change on extreme wind gusts from TCs in eastern Australia. However, based
on the available information including from modelling and observations, as well as considering
summaries from other review studies (Knutson et al. 2020; NESP 2020), it can be said with low-
medium confidence that little change or an increase are more likely than a decrease in the
24
occurrence frequency of Category 4-5 TCs in the future for Australia, including for the east coast
during summer.

Observed trends
Because of observational constraints, historical trends in the frequency and intensity of
convective winds in Australia are unknown (Walsh et al. 2016; Brown & Dowdy 2019). This is
largely due to spatio-temporal inhomogeneities in severe weather reports (Allen et al. 2011) and
wind observations (Jakob 2010). It is also noted that convective phenomena occur on small spatial
scales which are often missed by the observational network and make the detection of trends
difficult. However, observed lightning activity, indicative of convective activity, shows a
potential long-term decrease in occurrence frequency during winter in southern Australia with
little change during summer (Bates et al. 2015).
More broadly, strong winds from station data (defined as the 90th percentile of daily
maximum observations and including all wind-producing phenomena) have shown long-term
decreases in frequency in Australia (Azorin-Molina et al. 2021), consistent with decreases in
average wind gust magnitude (McVicar et al. 2008). These changes may be partly attributable to
environmental factors such as vertical wind shear and thermal instability, although the exact
causes are unknown.
Further details on trends are also provided in subsequent sections below. This includes
results based on SCW environments.
Model assessment
GCMs, reanalyses and commonly used downscaling approaches available for Australia
are unable to resolve the small spatial scales required for simulation of SCWs. Therefore, models
are assessed here in terms of their ability to correctly represent the environments which are
favourable for SCW occurrence, as well as the spatial and temporal variability of these
environments. The ability of environmental model diagnostics to represent the variability of
observed events is also discussed. Additionally, some details on fine-scale (convection-
permitting) modelling are also provided in Section 4.4.
For Australia, GCMs are generally able to represent the spatial distribution of severe
thunderstorm environments, although significant biases may exist for individual models in the
seasonal and diurnal cycle, related to the representation of near-surface moisture (Allen et al.
2014). In other regions, climate model representation of thunderstorm environments has been
shown to vary greatly with individual models (Seeley & Romps 2015), while some models have
been shown to replicate historical trends in environments for sufficiently large climate signals
(Pistotnik et al. 2016). Individual model biases for severe thunderstorm environments may be
addressed to some extent using a multi-model ensemble with bias correction.
Reanalysis models used for historical analyses can reliably represent atmospheric
environments based on observed sounding data (Brown & Dowdy 2021), although some key
elements such as CIN may remained unresolved due to insufficient vertical resolution (King &
Kennedy 2019). SCW diagnostics from these models can broadly represent the seasonal and
diurnal cycle of measured wind events in Australia (Brown & Dowdy 2021). Diagnostics have
also been shown to have a statistically significant correlation with the observed inter-annual
variability of SCW events, which has also been found for other small-scale convective hazards in
other regions, such as tornado events in the United States (Gensini & Brooks 2018). In addition,
25
environmental model diagnostics have been shown to explain most of the variability in
convection-permitting modelled thunderstorms (Hoogewind et al. 2017).
In summary, reanalysis data can provide a good representation of thunderstorm
environments as well as SCW environments, such as broadly representing features of the
variability of observed events. Significant biases exist in the representation of these environments
within individual climate models, although biases may be somewhat addressed using multi-model
ensembles with bias correction.
Trends in severe convective winds based on diagnostic methods

Historical trends in the frequency of atmospheric environments favourable for SCWs are
assessed here using the ERA5 reanalysis (Hersbach et al. 2020). Fig. 4.2 presents historical
summertime trends from 1979-2018, using four different diagnostics for environment
identification. This includes one method which has been developed by Brown & Dowdy (2021)
using a statistical diagnostic (referred to herein as BDSD), as well as three other diagnostics that
have been used in a range of previous studies and for severe weather forecasting purposes.
The BDSD was shown to provide a good representation of spatial and temporal variability
in observed convective wind events as compared to other commonly used environmental
diagnostics for severe thunderstorm environments. The BDSD is specifically tailored to SCW
environments and designed to represent a broad range of relevant physical processes (e.g., a
broader range of processes than is the case for the other diagnostics shown in Fig. 4.2). However,
the other diagnostics are also considered in this analysis for general completeness, as well as
noting the considerable uncertainties around the use of any single method for analysis of long-
term climate trends in SCWs based on currently available knowledge. Further details on these
diagnostics and analysis available in Brown & Dowdy (2021).
The BDSD indicates little or no long-term trend in occurrence frequency for southeast
Australia (Fig. 4.2). There are areas of decreasing frequency over some inland regions, broadly
similar to previous results for the state of South Australia based on somewhat different diagnostics
(Brown and Dowdy 2019) and noting some fine-scale regional variations. In addition to BDSD,
the other three diagnostics shown in Fig 4.2 indicate decreases around some part of eastern
Australia, particularly in the more inland regions, as well as in northern Australia. These
alternative diagnostics also indicate some areas of increasing frequency around the southeast
coast. These reanalysis-based trends are somewhat similar to previous analysis of thunderstorm
environments indicating positive trends in this far-southeast region with negative trends in general
for other regions including northern Australia (Dowdy 2020a), once again noting the increases in
SSTs in this region associated with a strengthening EAC as one plausible contributing factor that
might provide a source of enhanced warm and moist air to aid convective processes (as discussed
in Section 4.2).
In summary, this trend analysis based on reanalysis data indicates relatively little change
throughout most of southeast Australia based on the BDSD statistical method. Decreases are
indicated for most northern and central-eastern regions, including from the full set of diagnostics
more broadly. Increases are indicated for some near-coastal regions in the far southeast, as well
as for some southwest regions of Australia. Although there is relatively low confidence in these
trends in general, the use of these multiple methods as presented in Fig. 4.2 provides considerable
new insight from what was previously available, with further details on these methods and
findings available in Brown and Dowdy (2021).
26
Figure 4.2: Long-term changes in the frequency of days with favourable SCW environments
during the summer, based on ERA5 reanalysis data. Changes are based on four diagnostics, (a)
BDSD, (b) total totals (T-Totals), (c) severe hazards in reduced buoyancy environments
(SHERBE) and (d) the derecho composite parameter (DCP). The change in the mean number of
days per season is shown, calculated as the difference from the period 1979:1998 to the period
1999:2018. Significant changes are represented by hatching based on Student’s t-test with a
90% confidence level (two-tailed).

Global climate models
As discussed in sections above, there is very limited information available on projections
of SCWs in Australia. Here we use various environmental diagnostics (as used to assess historical
trends in Section 4.3) applied to future projections data from a bias-corrected 12-member CMIP5
ensemble (Taylor et al. 2012).
Future changes in the frequency of environments are presented for four diagnostics
relevant for convective winds between 1979-2005 and 2081-2100, presented for the summer
months DJF (Fig. 4.3). These diagnostics are the same as those used in Section 4.3, again noting
that the BDSD (Fig. 4.3a) is potentially most suitable based on representing the variability of
historical events (Brown & Dowdy 2021).
The projections for BDSD generally indicate future increases in the frequency of
environments across Australia, although little or no change may be more plausible for some near-
coastal regions in eastern Australia and Tasmania. Increases are also generally indicated for two
of the other three diagnostics (SHERBE and DCP), while decreases are indicated by the total
totals diagnostic. The diagnostics which indicate increasing frequency in environments are largely
driven by increasing moisture content in the lower atmosphere, while the decrease for total totals
is driven by a stabilisation of the temperature lapse rate. Increasing moisture and decreasing lapse
rate are expected in the future (see Section 4.2) and have opposite effects on the potential for
convection to occur. These competing factors introduce uncertainty for future projections of SCW
environments as represented by these diagnostics, as it is unclear whether changes to the
atmospheric lapse rate or moisture will be more influential.
27
Figure 4.3: Projected future changes in the frequency of favourable SCW environments during
summer shown as (a-d) a change in the number of days per season and (f-i) percentage changes.
The changes are calculated from the period 1979:2005 to the period 2081:2100 based on a high
emissions pathway (RCP8.5) using an ensemble of 12 GCMs. The ensemble median response is
shown. Changes where at least 10 (out of 12) models agree on the sign of change, as well as
where the seasonal mean number of environments in the historical period is greater than one,
are shown with hatching. These results are intended for broad-scale guidance on some of the
plausible changes that could occur for SCW occurrence in a warmer world, including on
direction of change and estimated range of potential future change as represented by these
metrics.
Convection-permitting modelling
Convection-permitting (a.k.a. convection-allowing) modelling has been used in a
relatively limited number of studies as an alternative to the large-scale environmental approaches
commonly used for projections of severe thunderstorms and associated hazards. Although being
very computationally expensive, this type of modelling can have the advantage of simulating
some factors which are more challenging to represent in environmental approaches. This can
potentially include better simulation of CIN and some triggering mechanisms such as the
influence of localised orographic features, as well as potential for improved representation of
some other aspects of thunderstorm characteristics (e.g., potentially providing some estimates of
intensity and morphology in some cases).
28
Leslie et al. (2008) used a convection-permitting model to dynamically downscale

climate model data in order to study potential future changes to hailstorms in Sydney, with results
suggesting an increase in the number of large hail events but with little change to the total number
of hail events. Modelling in the United States has found similar increases for large hail with little
change or decreases for moderate- and smaller-sized hail (Trapp et al. 2019; Raupach et al. 2021).
There have also been modelled increases for the frequency of hazardous convective events in
general without being specific on the type of hazard (Hoogewind et al. 2017). Elsewhere,
convection-permitting modelling in the United Kingdom has suggested an increase in the
intensity and frequency of convective rainfall (Kendon et al. 2017). However, more modelling at
these fine scales, including with a greater number of driving GCMs and covering longer periods
needs to be done to build on these results, including with a focus on severe thunderstorms in
Australia's changing climate.
A limited amount of convection-permitting modelling was produced for this study by
applying the BARPA modelling framework using around 4 km horizontal grid spacings, covering
a reduced mid-latitude domain including the capital cities of Sydney, Adelaide, Melbourne and
Hobart (as well as noting the availability of BARPAC-T using a 2 km grid spacing for a region
around the tropical east coast of Australia). Initial results suggest that this convection-permitting
approach which includes downscaling the ACCESS1-0 GCM (BARPAC-M) can provide a better
representation of severe wind gusts relative to its host model: the convection-parameterising
BARPA configuration (BARPA-R) that has a 12-km horizontal grid spacing (using BARPA-R
downscaling from the ACCESS1-0 GCM from CMIP5 for the RCP8.5 emissions pathway). For
example, analyses of BARPAC-M and BARPA-R data are presented here and compared with
daily maximum wind gust observations from station data at 12 locations (Fig. 4.4a,b), indicating
broadly similar results for BARPAC-M to those based on observations with somewhat lower wind
speeds for the upper tail in BARPA-R. These 12 locations are from observation stations in the
BARPAC-M region that have a reasonable quality and length of wind data suitable for climate
analysis, such as discussed in Brown & Dowdy (2021).
Results also suggest that the BARPAC-M model under a future climate scenario (2039-
2059) produces stronger 20-year maximum wind gusts when considering all land points in the
domain relative to the historical run (1985-2005; Fig. 4.4c). These results for future changes may
not be statistically significant due to the small sample size of extreme gusts and noting various
uncertainties from the modelling approaches (including potential variation between different host
models, time periods, emission pathways, etc.), the gust origins (i.e., synoptic or convective, as
well as potential for different types of convective modes) or spatial variations. However, they
demonstrate that convection-permitting approaches may provide additional insight into future
projected changes in extreme events such as severe thunderstorm hazards such as SCWs. In
particular, these initial results indicate that increased intensity of SCWs in the future is a plausible
outcome, while noting the considerable uncertainties discussed above and the limited data
currently availability for convection-permitting modelling of future simulated climates.
29
Figure 4.4: Modelled wind gust speed vs observed wind gust speed, presented for different
quantiles of daily maximum wind gusts at 12 locations. Results are presented from the
convection-permitting mid-latitude model run of BARPAC-M (a) as well as from its host model
for the regional configuration of BARPA-R (b). As one example of projected future changes
based on BARPAC-M, the 20-year maximum wind gust is shown under historical (1985-2005)
and future (2039-2058) conditions, with the blue line representing the quantile-matching of
wind speeds between those two periods using data for individual grid points (land only). The
dotted line represents no change from historical to future, with values below and above that line
representing decreases and increases, respectively, in the occurrence frequencies of wind speeds
in the ranges shown.
30

Table 4.1: Lines of Evidence Table for severe convective winds (SCWs), with a focus on
summer in regions around southern and eastern Australia. The degree of influence is listed in
black, followed by whether this information implies an increase (red), decrease (blue) or little
change (black) in the occurrence of SCWs, as well as by increased uncertainty (purple) in the
direction of change. The rows of information are not in order of importance.
Physical processes
Thunderstorm Moderate-strong influence. More favourable environments in

environments (not parts of southeast (low confidence) with increasing moisture
specific only to content (high confidence), as well as decreasing atmospheric
SCWs) lapse rate (medium-high confidence) and vertical wind shear
(medium confidence).
SCW Strong influence. Many uncertainties and few studies to date.

environments
Thunderstorm Strong influence. Uncertain changes (relating to extratropical

initiation cyclones, fronts, jet-streams, atmospheric waves, orographic
flows and convective inhibition), with increasing SSTs.
ENSO Small influence. Uncertain future change.
IOD Small influence (moderate in northeast). Uncertain future change.
SAM Moderate influence in east. Projected shift towards positive

SAM.
Additional factors Moderate influence of TCs in subtropics, as well as ECLs in
including coastal east and southeast, for damaging winds in summer.
phenomena such Uncertain expansion of TC range. Likely to be fewer TCs, but
as cyclones more intense on average. Uncertain projections for summer
ECLs, including their intensity and associated extreme winds.
Historical climate
Historical trend in Uncertain due to observational constraints.

observed SCWs
Model assessment Many uncertainties and limitations for current modelling.

However, SCW environments can be simulated reasonably well
by calibrated climate model ensembles.
Historical trend in Little change or fewer through inland eastern Australia with
SCW small region of potential increase in southeast (low confidence,
environments with uncertainty in modelling methods and limited observations).
Future climate
31
GCMs: CMIP5 More SCW environments for southern and eastern Australia (low
confidence due to uncertainty in model diagnostics).
RCMs and Some indication of a potential increase, but with very limited
convection- available data and analysis to date (highlighting a need for more
permitting models research).

The lines of evidence table (Table 4.1) shows high uncertainty in observed trends and
projected future changes for extreme winds during summer in regions around southern and eastern
Australia. Uncertainty arises from numerous sources including the lack of suitably homogenous
observations for long-term climate trend analysis. Uncertainty also arises from modelling
limitations, including due to the small spatial scales associated with the physical processes that
lead to the occurrence of severe thunderstorms and the SCWs they can cause, as well as
uncertainties around future projected changes in other phenomena that can cause extreme winds
(including landfalling severe TCs and intense summer ECLs). Potential improvements might be
obtained from convection-permitting modelling, while noting very little analysis on that to date
(including as presented in Section 4.4).
Insight on plausible future changes is provided by the environmental diagnostic approach
(i.e., large-scale diagnostics). Calibrated model projections from an ensemble of GCMs indicate
a range of changes (including increases and decreases) in the frequency of days with favourable
conditions for SCWs, with ensemble median changes of 7% and 8% increased frequency for
southern Australia and eastern Australia, respectively (based on those two supercluster regions as
defined in CSIRO & BoM (2015)). Confidence in this result is relatively low (i.e., much lower
than for extreme temperature projections from the previous section) and spans a wide range of
plausible change indicated by the different diagnostics and individual GCMs: 10th and 90th
percentile estimates based on a 48-member model-diagnostic ensemble are provided in Table 4.2.
Increasing environmental frequency based on median estimates agrees with expected changes to
thunderstorm environments in Australia based on physical process understanding (Low
Confidence), including increased atmospheric moisture content in a warmer world, although
decreases are also plausible due to decreasing atmospheric lapse rates (noting the future decreases
projected by one of the diagnostics in Fig. 4.4: Total-Totals). Additional uncertainties also relate
to factors not included in these environmental diagnostics (such as initiation mechanisms,
convective inhibition, microphysical processes, etc.).
The results based on environmental diagnostics are broadly similar to the initial results
from the convection-permitting model runs of BARPAC-M (used to dynamically downscale the
BARPA-R downscaling projections), which indicated a small increase in the upper tail of wind
gust speeds in the future. However, further research is required to examine how well these
extremes can be simulated in the fine-scale model data provided by convection-permitting
modelling approaches.
Based on this overall assessment considering this wide range of factors, there is Low
Confidence in the projected direction of change, with a future increase in 10-year ARI
temperatures being more likely than not (i.e., > 50% probability). In particular, an increased
occurrence frequency of extreme winds is indicated in southern and eastern Australia during
summer (Low Confidence), noting that both increases and decreases are plausible outcomes based
on the full range of lines of evidence considered here. The estimated range from the environmental
32
modelling (Table 4.2) is intended to be useful for some planning and risk management purposes.
The central estimates of the model ensemble could also be useful in some cases, showing that the
most likely projections for the future is little change or a small increase in frequency.
Table 4.2: Projected percentage changes in severe convective wind environment frequency
(days per season) during summer, based on 12 CMIP5 GCMs, as well as using four diagnostics
(Brown & Dowdy 2021). This results in a 48-member ensemble, with the median, 10th and 90th
percentile changes shown. The changes are calculated from the period 1979:2005 to the period
2081:2100 based on a high emissions pathway (RCP8.5), averaged over Eastern and Southern
Australia (using the regions defined in CSIRO & BoM (2015)).
Region Median change 10th percentile 90th percentile
Eastern Australia 8% -56% 33%
Southern Australia 7% -49% 45%
33
5. EXTREME FIRE WEATHER PROJECTIONS
5.1 Introduction
Bushfires can be considered as a form of compound event given the range of factors that
influence their occurrence, including based on the combined influence of various weather factors
(from various near-surface conditions to higher-level atmospheric processes including
convection). Bushfire occurrence is also influenced by various other factors including vegetation
conditions (such as relating to fuel load and type) and ignition sources (such as associated with
human activities or with lightning), some of which can be challenging to model (given current
climate modelling capabilities). The primary focus of the analysis presented here is on dangerous
weather conditions for bushfires, with other factors also considered to some degree (i.e., relating
to vegetation conditions and ignition sources).
Bushfire weather conditions are often represented by indices as a useful way of
combining various weather conditions known to influence fire behaviour (e.g., near-surface
humidity, wind speed, temperature and rainfall). Examples of such indices include the Forest Fire
Danger Index (FFDI) commonly used in Australia (McArthur 1967) as well as the Fire Weather
Index (FWI) originally developed in Canada but now widely used throughout the world (Van
Wagner 1987; Field et al. 2017). The FFDI and FWI are both based on near-surface measures of
humidity, wind speed, temperature and rainfall, with a broadly similar order of sensitivity to these
four individual weather conditions (Dowdy et al. 2009). Indices have also been developed for
grass fires, while noting that grass fires were not identified as a key hazard of interest by
stakeholders for this research. Indices are also available for various other fuel types including a
multi-index system currently in development for Australia (known as the Australia Fire Danger
Rating System: AFDRS).
Indices such as the Continuous-Haines index (C-Haines) are based on conditions at
higher levels of the atmosphere and can be useful for indicating risk factors associated with the
occurrence of extreme fire events (including very dangerous fires that generate thunderstorms in
their fire plumes known as pyrocumulonimbus or pyroCb clouds) (Mills & McCaw 2010; Dowdy
et al. 2019b). Many of the more disastrous fire events in recent decades have been associated with
the occurrence of pyroCbs, including for the Canberra fires in 2003 and the Black Saturday fires
in 2009 as well as during the 2019/2020 Black Summer fires (Fromm et al. 2006; Cruz et al. 2012;
McRae et al. 2013; Dowdy et al. 2017; Australian Government 2020). Such events are often
associated with extreme fire weather conditions occurring simultaneously at near-surface levels
(e.g., as indicated by the FFDI) as well as at higher levels (e.g., as indicated by the C-Haines
index) (Dowdy and Pepler 2018; Di Virgilio et al. 2019).
The standardised method for projections information is applied here for extremely
dangerous fire weather conditions during summer in regions around southern and eastern
Australia. The combined influence of multiple different weather conditions known to influence
fire behaviour is considered here. Factors considered include near-surface weather variables such
as humidity, wind speed, temperature and drought measures relating to fuel availability, as well
as other atmospheric phenomena such as the influence of synoptic systems, mesoscale convective
processes as well as large-scale atmospheric and oceanic modes of variability. Although the focus
here is on fire weather, other factors relating to bushfire occurrence are also discussed including
ignition and fuel conditions.
34

Individual weather factors
Weather conditions such as humidity, wind speed and temperature can influence fire
behaviour in Australia (e.g., McArthur (1967)). These conditions can change as our climate
warms (CSIRO & BoM 2015; BoM & CSIRO 2020) as examined in this section.
Climate change is increasing the frequency and severity of extreme heat events (very high
confidence), including for individual days as well as for more prolonged events (e.g., heatwaves).
This is based on many lines of evidence including from observations, modelling and physical
processes understanding. For details, see Section 3.
Increased temperatures lead to an increase in the moisture holding capacity of the
atmosphere (of about 6-7% per degree of warming based on the Clausius-Clapeyron relation),
which results in increased water vapour pressure in general (i.e., increased specific humidity).
Observed climate trends in humidity are not well documented for Australia, but most sites show
long-term increases in atmospheric water vapour concentrations (i.e., including measures of this
such as dewpoint temperature and specific humidity), with the largest increases in the interior of
the continent and some eastern regions (Lucas 2010). However, it is relative humidity or related
measures such as vapour pressure deficit that are important to consider for fire behaviour,
including due to influencing fuel moisture. As some regions warm faster than others (e.g., land
regions warm more than ocean in general) there can be differences in the relative humidity for a
given change in water vapour content. In general for Australia, a decrease in relative humidity is
projected to occur, including during summer with CSIRO & BoM (2015) listing medium
confidence for this (as compared to high confidence for winter and spring). It is also noted that
some finer-scale modelling from RCMs indicates little change in some regions (Clarke & Evans
2019).
A small decrease in wind speed has been observed for Australia in general, while noting
considerable uncertainties relating to data availability and homogenisation (Azorin-Molina et al.
2021). There are also considerable uncertainties around model data for wind speed, including due
to significant negative bias in modelled wind speed during high wind conditions (in general for
most models). Many factors such as boundary layer mixing, form drag for sub-grid orography
and surface properties can influence wind estimation over land. The representation of the stable
boundary layer remains challenging due to the multiplicity of physical processes (including
turbulence, radiation, land surface coupling and heterogeneity, turbulent orographic form drag)
involved and their complex interactions, such that models typically suffer biases in wind speed
under such conditions. Projections for Australia indicate little change or a small decrease during
summer in mean wind speed, with considerable variation between different models: some show
increases and others show decreases, typically within about +/-5% in magnitude (CSIRO & BoM
2015). Further details on processes that can cause strong winds are provided below in this section
(in relation to synoptic-scale phenomena such as fronts).
Drought and fuel moisture

Drought conditions can lead to low moisture content in vegetation that increases the
availability of fuel for bushfires. Climate change is expected to increase the intensity, frequency
and duration of meteorological drought (i.e., a measure of drought based only on rainfall deficit),
including based on longer periods with little rainfall as detailed in CSIRO & BoM (2015). It is
also noted that there are various other ways that drought conditions can be defined including
agricultural drought measures that can include the influence of other weather conditions (e.g.,
temperature, humidity, wind as well as evapotranspiration) in addition to rainfall.
35
Fire weather indices such as the FFDI and FWI include drought measures in their
formulation that are more similar to measures of agricultural drought than meteorological drought
in that they include the influence of other weather conditions in addition to rainfall. For example,
temperature is used together with rainfall as input to the Keetch-Byram Drought Index (KBDI)
(Keetch & Byram 1967) as often used as an input for the Drought Factor used in the FFDI (used
to indicate a proxy estimate of fuel availability based on moisture content). In contrast, relative
humidity, temperature and wind speed are used for the multiple different fuel moisture measures
that the formulation of the FWI System includes (Van Wager et al. 1974).
As noted in the section above on individual weather factors, mean temperatures as well
as the frequency of extreme temperature events are projected to increase in the future with high
confidence, together with a general decrease in relative humidity, as well as little change or a
small decrease in wind speed. Considering these factors together with the projected increase in
meteorological drought (including increased frequency, intensity and duration) suggests a likely
increase in the frequency of very dry fuel conditions. However, there are considerable
uncertainties around projected changes in different types of drought as well as fuel moisture
responses to climate change, including as noted in Section 3 in relation to soil moisture
projections. Regional models may add value for some of these factors (e.g., more detail on land
surface processes, rainfall and orographic dependencies).
Combined weather conditions

Fire weather indices provide a useful way to combine a range of weather conditions
known to influence fire danger. The index values are typically calculated for each individual time
step (e.g., day) using data for each weather factor obtained from a single model (as is the case
throughout this report). This ensures the coherence of these individual weather factors when
applied for individual time steps from a single model. After the fire weather index values have
been calculated for each model, the ensemble statistics and other derived products can then be
produced, rather than using ensemble average values of individual weather conditions as input to
calculate the fire weather indices as that will lose the coherence of individual weather factors
(including noting the importance of this for representing extremes of the fire weather index
values). Similarly, the weather data should be calibrated prior to using those data for calculating
the fire weather indices, rather than calibrating the resultant index values, to keep the relative
balance of each weather factor correct for the index formulation.
The FFDI is commonly used in Australia as a general indicator of regional weather
features associated with dangerous fire conditions. It shows broad similarities to some other fire
weather indices used around the world such as the FWI including for its sensitivity to different
input ingredients (including being most sensitive to wind speed followed by humidity and then
temperature) (Dowdy & Mills 2012). Observational studies have identified an increase in both
the average FFDI and the frequency of high FFDI days over much of southern Australia,
particularly during the spring months, contributing to a lengthening of the fire season (Dowdy
2018, Harris & Lucas 2019). These trends are attributable at least in part to anthropogenic climate
change, including as they combine several different weather variables of which some
(temperature) can be more easily attributed to climate change than others (humidity and wind).
Although a significant climate change signal is able to be demonstrated already based on
observations (Dowdy 2018; Harris & Lucas 2019), the attribution of individual fire events to
climate change is more challenging while noting one recent study that has done this for the Black
Summer of 2019/2020 (van Oldenborgh et al. 2021).
Projected changes in extreme daily FFDI were recently produced for Australia drawing
on a comprehensive range of modelling techniques, comprising an ensemble of projections based
on GCM output as well as two ensembles of projections based on dynamical downscaling using
36
regional model approaches (Dowdy et al. 2019b). Those projections indicate an increase in the
number of days with very high fire weather conditions (based on FFDI above 25) as well as an
increase in the number of days with FFDI above the 95th percentile for 1990-2009), noting lower
agreement between models in some parts of eastern Australia. Similarly, future increases were
also projected for the number of days with FFDI above 50 and for the number of days with FFDI
above the 99th percentile for 1990-2009 (Dowdy 2020b). In addition to the projections presented
in those studies, plausible variation above and below such values is indicated from previous
studies based on different metrics and different modelling approaches using FFDI. For example,
relatively large increases have been derived using monthly mean climate changes from 3 GCMs
to scale observations and calculate changes in severe fire weather days with FFDI > 50 (CSIRO
& BoM 2015), as well as other studies that indicate less confidence in large increases in FFDI in
the future (Clarke et al. 2016).
Projections of future climate have also been produced based on other fire weather indices,
including a global study that used the FWI (Abatzoglou et al. 2019) and reported no emergent
climate change signal in general for Australia based on the methods they presented. Although
increases were projected in some regions they were not statistically significant at a high
confidence level noting the high interannual variability that can occur in weather and climate
conditions in Australia (such as due to the influence of large-scale modes of variability including
ENSO, discussed in sections below). Examples such as that based on FWI with little change
indicated, together with the range of FFDI projections from various studies noted above, show
that considerable differences can occur between different studies and highlight the benefit of
considering results from a broad variety of datasets and methods (as is a goal of this standardised
method).
Very dangerous types of fire events have also been examined in relation to climate
change, including extreme pyro-convection conditions (i.e., associated with thunderstorms that
form in fire plumes: pyroCbs). PyroCbs occurred for the Black Saturday fires in 2009 and the
Canberra fires in 2003 fires as well as many examples during the 2019/20 Black Summer fires
(Fromm et al. 2006; McRae et al. 2013; Dowdy et al. 2017; Australian Government 2020).
Significant trends have been found for extreme pyro-convection risk factors including based on
historical data (Dowdy & Pepler 2018) and future projections (Di Virgilio et al. 2019; Dowdy et
al. 2019b). These studies indicate increased risk factors for parts of southern and southeast
Australia as well as decreases in some cases for other regions including in parts of eastern
Australia. However, a range of uncertainties around future changes in convective systems is also
noted, such as the contrasting roles of increasing water vapour content and decreasing lapse rates
that can have various influences on risk factors associated with fire behaviour and/or potential for
convective systems to develop (with details also available in Section 4 around uncertainties in
future projected changes for convective systems).
Subtropical ridge; Blocking / high pressure systems; Cold fronts

Details on various phenomena including the subtropical ridge, blocking highs and cold
fronts were provided previously (see Section 3), including observed and projected changes during
summer, as well as strengths and limitations of different modelling approaches. Building on that
information for those phenomena, details specific to fire weather conditions are provided in this
section.
The projected increase in the strength of the subtropical ridge could potentially act to
exacerbate the severity of some fire weather events in the future, especially in parts of southern
Australia. For example, the high-pressure systems that characterise the subtropical ridge can lead
to descending dry air and clear skies associated with hot and dry conditions. High pressure
systems can also circulate air around inland Australia in some cases, as a dynamical mechanism
37
contributing to the build-up of extremely hot and dry air, while noting it is not currently known
if this process would change in the future. Blocking (quasi-stationary) highs over the Tasman Sea
can advect hot and dry air from inland regions towards the more densely populated regions closer
to the south and east coasts, as well as interact with approaching cold fronts from the south to
produce strong northwesterly winds, corresponding to a very dangerous set of fire weather
conditions for southern and eastern Australia (Hasson et al. 2009; Reeder et al. 2015; Dowdy et
al. 2017).
The passage of the front (or pre-frontal trough) comprises shifts in wind direction which
can change the direction of fire movement, i.e., the northern flank can become the new head fire
leading to rapid increases in the rate of area burnt. This can cause significant challenges for
firefighters (Cruz et al. 2012). While future projections of cold fronts are generally uncertain, as
detailed in Section 3, one study based on the older generation of climate models (CMIP3) found
a projected increase in frequency of such extreme events from 0.5 to 1-2 per year by the end of
the 21st century (Hasson et al. 2009).
Modes of variability – ENSO, IOD and SAM

Details on modes of variability including ENSO, IOD and SAM in a changing climate
were provided in Section 3. Building on that information, aspects relating to fire weather
conditions are summarised here.
A recent paper summarised the seasonal influences of these three modes of variability on
average fire weather conditions in Australia (Harris & Lucas 2019), finding a strong influence
from ENSO during spring and summer in the east, from IOD during spring in the southeast and
east and from SAM during spring and summer in the east. This is broadly similar to various other
studies that have also examined some of those aspects (Dowdy 2018; Abram et al. 2021), as well
as studies considering individual fire seasons (e.g., of 21 significant bushfire seasons since 1950
in south-east Australia, 11 were preceded by a positive IOD (Cai et al. 2009)). In southeast
Australia, a positive IOD during spring is typically associated with lower rainfall and higher
temperatures, exacerbating dry conditions and increasing the fuel availability leading into
summer.
Sudden stratospheric warmings can also influence fire weather conditions in Australia,
including hotter and drier conditions for parts of eastern Australia during spring and early summer
which could also influence fuel moisture content during summer to some degree, noting that the
influence of such events can also be indicated through the SAM index (given the association
between polar stratospheric vortex conditions and measures of the Southern Annular Mode) (Lim
et al. 2019; 2021). The influence of climate change on sudden stratospheric warming events is
currently unknown.
Although the relationships between fire weather and modes of variability (including
ENSO, IOD and SAM conditions) have been examined in numerous previous studies (such as
those discussed in this section), this has not previously been examined in detail for more extreme
measures of fire weather, such that some new analysis on that is shown in Fig. 5.1. Correlations
are presented between the number of days with FFDI > 99.5th percentile and various modes of
variability (using indices representing ENSO, SAM and IOD) showing broadly similar features
to those for average values of fire weather measures as described based on previous studies
mentioned above. In particular, fire weather conditions in the southeast and east of Australia
during summer show significant relationships with ENSO and IOD (positive correlations), with
SAM having some influence in central east regions (positive correlation) but to a lesser degree
than ENSO and IOD. There are some regions of negative correlation for the SAM results in the
more inland parts around central-east and southeast Australia, but those correlations are not
statistically significant. It is also noted that the influence of sudden stratospheric warmings
38
(relating to negative SAM conditions to some degree) can be associated with more severe fire
weather conditions in central eastern Australia during spring (Lim et al. 2019), with this not
expected to be represented in these results focussed on summer in Fig. 5.1.
Figure 5.1: Correlations for the number of days with FFDI > 99.5th percentile during summer
and measures representing different modes of variability including ENSO (using the NINO3.4
index), SAM (using the SAM index) and IOD (using the DMI index). These correlations are all
based on the period from 1979 to 2019, using one value for each summer period (DJF).
NINO3.4 and DMI data are attained from the NASA ESRL
https://legacy.bas.ac.uk/met/gjma/sam.html. Pearson's correlation coefficient, r, is shown with
stippling corresponding to statistically significant values at the 95% confidence level (2-tailed).
Additional factors- lightning ignitions as well as fuel load and type

Although the focus of this analysis is on extreme fire weather conditions, a brief summary
is provided here to note some of the other conditions that are important for the occurrence of
bushfires. This includes ignition sources as well as vegetation-related factors such as fuel load
and type.
Lightning was the ignition source for many of the largest and most damaging fires during
the 2019/2020 summer fire season in southeast Australia (Australian Government 2020). In
addition to individual summers, lightning has been found to cause most of the total area burnt
when averaged over many fire seasons in southeast Australia (Dowdy & Mills 2012). Human-
caused ignitions are also a key cause of fires in Australia, noting that projected future changes in
that are highly uncertain. Given the occurrence of lightning, the chance that it will cause a
sustained ignition and develop into a bushfire is strongly dependent on the amount of rainfall that
accompanies it, leading to the concept of 'dry lightning' as an important natural ignition source
for bushfires (i.e., lightning that occurs without significant rainfall). There is some indication of
an increased frequency of dry-lightning in some parts of southeast Australia in recent decades as
well as decreases in some other regions more broadly for Australia (Dowdy 2020a). However,
projections of future changes in the occurrence of dry-lightning is a key knowledge gap in general
for Australia, affecting our understanding of potential future changes to bushfire ignition (and
therefore also bushfire occurrence) throughout Australia. In addition to the rainfall that
accompanies the lighting, the pre-existing moisture content of the fuel (i.e., vegetation) is also a
factor that influences the chance that a fire will occur (given the occurrence of lighting).
Consequently, the increased frequency projected for dry vegetation conditions in the future (see
39
Section 3.2) suggests that an increase in lightning ignitions might be more likely than a decrease
in the future climate, while noting considerable uncertainties around this.
Changes in vegetation characteristics including amount (fuel load) and type can also
influence fire hazards throughout Australia, noting that this is particularly important for grassfires
in the more northern and central regions of Australia (McKeon et al. 2009). There are potential
increases in fuel loads for various vegetation types associated with projected increases in carbon
dioxide concentrations, often referred to as the 'fertilisation effect' (Clarke et al. 2016), where
higher concentrations of atmospheric carbon dioxide promote vegetation growth (Donohue et al.,
2013). Global drylands have generally been greening over recent decades and the fertilisation
effect has been identified as a causal factor in this greening (Burrell et al. 2020). Consequently,
an increase in some fuel-related fire risk factors may be considered more likely than a decrease,
while noting considerable uncertainties given the relatively limited ability of current climate
models to accurately simulate future changes in some risk factors relating to fuel characteristics.
Similarly, there are also large uncertainties around potential future changes in fuel type, such as
whether or not vegetation may shift to types that tend to burn more frequently during this
transition period to a warmer world, with no studies currently available on this topic for Australia.

Observed trends
Early studies on fire weather trends in Australia based on FFDI were not able to separate
the influences, if any, of climate change as different to natural variability, as was concluded by
Clarke et al. (2013). Using a longer time period, different methods and a gridded analysis based
on observations, a statistically significant increase in FFDI has since been documented,
particularly during spring and summer in many parts of southern and eastern Australia, with this
being attributable at least in part to human-caused climate change (Dowdy 2018). That trend
towards more dangerous weather conditions for bushfires is due to increased temperatures and
associated changes in relative humidity and fuel availability indicators. Similar results were also
reported based on station data for individual locations, finding that significant increases in FFDI
have already occurred during spring and summer different to what can likely be explained based
on natural variability alone (Harris & Lucas 2019).
Studies using observations-based data and reanalysis have also examined other fire
weather indices in Australia, including the C-Haines index over the period back to 1979 (Dowdy
& Pepler 2018), finding that statistically significant increases have already occurred including
during summer in some parts of southern Australia, including for simultaneous occurrences of
dangerous near-surface and upper-level conditions (based on FFDI and C-Haines).
Trends toward more dangerous weather conditions for bushfire have been further
confirmed in other recent climate change studies considering a range of other factors and analysis
methods. This includes some analysis over palaeontological time scales around how climate
change can influence large-scale modes of variability (e.g., extremes for ENSO and IOD
conditions) that can then lead to more dangerous fire weather conditions (Abram et al. 2021).
Model assessment
The ability of climate models to simulate aspects such as the seasonal cycle, observed
trends, spatial detail and extremes is important for helping to understand the degree of confidence
in future projected changes based on these models. Assessments presented in various studies
(CSIRO & BoM 2015; Di Virgilio et al. 2019; Dowdy et al. 2019b) indicate that global models
40
as well as downscaling approaches provide a reasonably good representation of these aspects,

including seasonal and regional variations through Australia as well as the observed trends.

As discussed in sections above, previous studies have examined projected future changes
in measures of extreme such as FFDI exceeding 25 or 50 as well as FFDI exceeding its historical
95th or 99th percentile. Here we examine projections of the 10-yr ARI of daily FFDI from the
available modelling approaches based on GCMs, CCAM, BARPA and NARCliM (described in
Section 3.4). These datasets all have QME calibration applied to the input variables for each
individual model prior to calculating the FFDI, with the ARI values then calculated from the FFDI
data using a GEV approach (as was the case for temperature extremes in Section 3).
The results show increases in the severity of fire weather conditions projected from the
historical climate to the future projected climate during summer (i.e., DJF), as represented by the
10-yr ARI value of daily FFDI. Some variation is apparent between the different model ensembles
in the magnitude of the increases, with somewhat larger increases for NARCliM in some regions,
but with general agreement over these modelling approaches on a projected future increase in
these values corresponding to the 10-year ARI.
Figure 5.2: Projections for values corresponding to the 10-year ARI for daily fire weather
conditions as represented by the FFDI (with the FFDI intended as a useful means of combining
different weather factors known to influence fire behaviour in Australia). This is shown based
on GCMs (left panels), CCAM (second to left panels), BARPA (second to right panels) and
NARCliM (right panels), all calibrated using the QME method. Maps are shown through
Australia based on the model ensemble average in each case, presented for the historical period
(based on 1986–2005; upper panels) as well as for the future simulated climate (based on 2040–
2059 under a high emissions pathway RCP8.5 from CMIP5; lower panels).
41

Table 5.1: Lines of Evidence Table for extreme fire weather conditions, with a focus on
summer in regions around southern and eastern Australia. The degree of influence is listed in
black, followed by whether this information implies an increase (red), decrease (blue) or little
change (black) in the frequency and severity of extreme fire weather conditions, as well as by
increased uncertainty (purple) in the direction of change. The rows of information are not in
order of importance. Additional factors are also noted around lighting and fuel conditions.
Physical processes
Individual weather Strong influence. More extreme temperatures and heatwaves, lower
factors relative humidity; small decrease in wind speed.
Drought and fuel Strong influence. Projected increase in frequency of meteorological

moisture drought and very dry fuel conditions. Considerable uncertainties for some
factors; regional models likely to add value.
Combined near- Strong influence. Projected increase in frequency of dangerous conditions

surface weather in general based on numerous studies; poor agreement between models
conditions, FFDI near east coast.
Combined near- Strong influence. Projected increase, but not statistically significant, and
surface weather only based on one study.
conditions, FWI
Upper-level Strong influence (including extreme pyroconvection). Increased frequency

conditions, C-Haines of dangerous conditions in southeast (including simultaneous occurrence
with dangerous near-surface conditions) and decrease in northeast.
Subtropical ridge Moderate influence in southeast. Potential increase.
Blocking Moderate influence. Future change uncertain.
Fronts Moderate influence. Future change uncertain.
ENSO Strong influence. Uncertain future change; potentially more frequent

strong ENSO events (low-medium confidence).
IOD Strong influence. Uncertain future change; potentially more frequent

strong IOD events (low-medium confidence).
SAM Strong influence in central east. Positive trend in SAM reducing

dangerous fire weather in central east region (medium confidence).
Historical climate
Seasonal cycle Models reproduce the seasonal cycle and spatial variability well (high
confidence).
42
Historical trend Increase from observations (medium confidence). Models reproduce the
trend well (medium confidence).
Future climate
GCMs: CMIP5 Increase (very high confidence).
RCM: CCAM Increase (high confidence in general; medium near east coast).
RCM: NARCliM Increase (high confidence in general; medium near east coast).
RCM: BARPA Increase (high confidence in general; based on one model to date).
Additional factors
Lightning ignitions Strong influence. Influence of climate change largely uncertain but
increase more likely than decrease (low confidence).
Fuel load and type Strong influence. Influence of climate change largely uncertain but
increased fuel load more likely than decrease (low confidence).

The Lines of Evidence Table shows considerable agreement on more dangerous fire
weather conditions in a warming climate for Australia, including in relation to 10-year ARI fire
weather conditions in regions around southern and eastern Australia during summer (as is a key
focus here). Although there are some physical processes noted that add uncertainties, particularly
based on GCM projections data, the RCM approaches can help with the simulation of some of
these processes such that the considerable level of agreement between RCM approaches
(particularly in southern Australia, but somewhat less so in parts of eastern Australia) helps add
some confidence for projected future changes.
Observed trends and RCM simulations are available for near-surface and higher-level
conditions, including combining those different levels using a compound event framework
(Dowdy & Pepler 2018; Di Virgilio et al. 2019; Dowdy et al. 2019b), showing increases in
southern Australia with more variation between results in eastern Australian including decreases
being indicated in some regions. Additionally, although there is low confidence for projected
future changes in vegetation-related conditions such as fuel load and type, as well as in ignition
risk factors including the occurrence of dry lightning, there is some indication that increases may
be more likely than decreases in risk factors associated with fuel condition and ignition sources
for bushfires (while noting considerable uncertainties and more research needed on such topics).
Based on this assessment of a broad range of factors that can influence the occurrence of
extremely dangerous fire weather conditions, there is high confidence in southern Australia and
medium confidence in parts of eastern Australia for the projected direction of change, with a future
increase in 10-year ARI fire weather conditions being likely (i.e., 66-100% probability) for
southern and eastern Australia.
Considering all of the review details in the sections above, and noting the predominance
of an increase from the Lines of Evidence Table, projections for 10-year ARI extreme fire weather
43
conditions in 2050 are developed here based on combining data from various calibrated modelling
approaches including GCMs (4 ensemble members), CCAM (5 ensemble members), BARPA (1
ensemble member) and NARCliM (6 ensemble members). FFDI data are available from these
models and are the primary data source used here. The contrasting modelling approaches are
combined based on equally weighting the changes.
The ensemble median is used as a central estimate of the most probable projected change
(Fig. 5.3). As an estimate of the range of plausible values, the second lowest value from the
ensemble is used for the 10th percentile and the second highest value is used from the ensemble
is used for the 90th percentile, with these values calculated individually at each grid cell location.
However, given some of the uncertainties and variations between different modelling approaches
and studies as noted in this section (including projections based on the FWI showing smaller
changes than for FFDI), the lower bound of the range provided here has been modified to reflect
the potential for lower values. This is done based on reducing any projected increases for the 10th
percentile by a factor of two (as a qualitative estimate based on expert judgement). For example,
at a given grid-cell location, if the 10th percentile for the future period was higher by a value of 8
as compared to the 1986–2005 value, it would be changed to only be a value of 4 higher than the
1986–2005 value at that location. Projections for any regions that show decreases for the 10th
percentile are not changed. Only the 10th percentile is changed to allow for lower values, but no
lines of evidence suggest these FFDI projections data systematically underestimate future
increases such that the 90th percentile is unchanged and is considered a plausible upper estimate
for the future projected changes for these fire weather conditions.
Figure 5.3: Projected change in values corresponding to the 10-year ARI for daily fire weather
conditions during summer. Maps are shown through Australia for the historical period (based on
1986–2005; upper panel), as well as for the future simulated climate (based on 2040–2059 under
a high emissions pathway RCP8.5: lower panels) including a central estimate with lower and
upper estimates also provided. The data are based on the FFDI, with some modifications based
on considering the broader lines of evidence from Table 5.1.
44
6. CONCLUSION
The influence of anthropogenic climate change on extreme temperatures, winds and fire
weather was assessed using this standardised method for projections information. Calibrated data
from GCMs and RCMs were used for temperature and fire weather, with environmental
diagnostics also used for severe convective winds from thunderstorms. The projections presented
here are more extreme than examined in previous studies (e.g., 10-yr ARI projections for fire
weather and severe convective winds), with care taken to communicate uncertainties and
document the comprehensive lines of evidence considered here.
The nationally consistent calibrated projections presented here, including based on new
RCM data from BARPA, CCAM and NARCliM as well as GCMs, are intended to be of use for
a broad range of applications. This includes for applications such as improved planning and
helping to build resilience in relation to the influence of anthropogenic climate change on future
hazards in Australia. Data are available on request.
The resultant productions data include nationally consistent maps corresponding to the
10-year average recurrence interval (ARI) around the middle of this century, with a focus of the
discussion on regions around southern and eastern Australia during summer, as needed for some
risk assessment applications. The projections are also available for other seasons and time periods
throughout this century, as well as for other metrics of extreme or average conditions. The results
for southern and eastern Australia during summer show more extreme temperatures (very high
confidence), more severe winds (low confidence) and more dangerous bushfire conditions (high
confidence in southern Australia; medium confidence in eastern Australia) attributable to
increasing greenhouse gas emissions.
ACKNOWLEDGEMENTS
This research was funded through the Energy Sector Climate Information (ESCI) project
supported by the Australian Government, Bureau of Meteorology (BoM), CSIRO and AEMO.
The authors would like to thank Chiara Holgate (BoM) and Justin Peter (BoM) for providing
reviews on earlier versions of this report as well as others in the ESCI project team for
contributing to various aspects around the information covered in this report, including Ben
Jones (AEMO), Peter Steinle (BoM), Greg Roff (BoM), Ian Watterson (CSIRO), Chi-Hsiang
Wang (CSIRO) and Leanne Webb (CSIRO).
REFERENCES
Abatzoglou, J. T., Williams, A. P., Boschetti, L., Zubkova, M., & Kolden, C. A. (2018). Global
patterns of interannual climate–fire relationships. Global Change Biology, 24(11), 5164-
5175.
Abram, N. J., Henley, B. J., Gupta, A. S., Lippmann, T. J., Clarke, H., Dowdy, A. J., Sharples,
J. J., Nolan, R. H., Zhang, T., Wooster, M. J., & Wurtzel, J. B. (2021). Connections of
climate change and variability to large and extreme forest fires in southeast
Australia. Communications Earth & Environment, 2(1), 1-17.
Allen, J., Karoly, D., & Mills, G. (2011). A severe thunderstorm climatology for Australia and
associated thunderstorm environments. Aust. Meteorol. Oceanogr. J., 61(3), 143–158.
45
Allen, J. T., Karoly, D. J., & Walsh, K. J. (2014). Future Australian Severe Thunderstorm
Environments. Part I: A Novel Evaluation and Climatology of Convective Parameters from
Two Climate Models for the Late Twentieth Century. J. Clim., 27, 3827–3847,
doi:10.1175/JCLI-D-13-00425.1
Allen, J. T., Karoly, D. J., & Walsh, K. J. (2014). Future Australian Severe Thunderstorm
Environments. Part II: The Influence of a Strongly Warming Climate on Convective
Environments. J. Climate, 27, 3848–3868.
Allen, J. T., & Karoly, D. J. (2014). A climatology of Australian severe thunderstorm
environments 1979 – 2011: Inter-annual variability and ENSO influence. Int. J. Climatol.,
34, 81–97.
Allen, J. T., & Allen, E. R. (2016). A review of severe thunderstorms in Australia. Atmos. Res.,
178–179, 347–366, doi:10.1016/j.atmosres.2016.03.011
Allen, J. (2018). Climate Change and Severe Thunderstorms. Oxford Research Encyclopedia of
Climate Science. Retrieved 13 Jan. 2021, from
https://oxfordre.com/climatescience/view/10.1093/acrefore/9780190228620.001.0001/acref
ore-9780190228620-e-62
Arblaster, J. M., & Alexander, L. V. (2012). The impact of the El Nio-Southern Oscillation on
maximum temperature extremes. Geophysical Research Letters, 39(20), 2–6,
https://doi.org/10.1029/2012GL053409
Argüeso, D., Evans, J. P., Fita, L., & Bormann, K. J. (2014). Temperature response to future
urbanization and climate change. Climate Dynamics, 42(7–8), 2183–2199,
https://doi.org/10.1007/s00382-013-1789-6
Argüeso, D., Evans, J. P., Pitman, A. J., & Di Luca, A. (2015). Effects of city expansion on heat
stress under climate change conditions. PloS One, 10(2), e0117066,
doi:10.1371/journal.pone.0117066
Australian Government (2020). Royal Commission into National Natural Disaster
Arrangements. Australian Government, ISBN: 978-1-921091-45-2.
Azorin-Molina, C., McVicar, T. R., Guijarro, J. A., Trewin, B., Frost, A. J., Zhang, G., Minola,
L., Son, S., Deng, K., & Chen, D. (2021). A decline of observed daily peak wind gusts with
distinct seasonality in Australia, 1941-2016, Journal of Climate, 1-63.
https://doi.org/10.1175/JCLI-D-20-0590.1
Banerjee, A., Fyfe, J. C., Polvani, L. M., Waugh, D., & Chang, K. L. (2020). A pause in
Southern Hemisphere circulation trends due to the Montreal Protocol. Nature, 579(7800),
544–548. https://doi.org/10.1038/s41586-020-2120-4
Bates, B. C., Chandler, R. E., & Dowdy, A. J. (2015). Estimating trends and seasonality in
Australian monthly lightning flash counts. Journal of Geophysical Research, 120(9), 3973–
3983. https://doi.org/10.1002/2014JD023011
Bell, S. S., Chand, S. S., Tory, K. J., Dowdy, A. J., Turville, C., & Ye, H. (2019). Projections of
southern hemisphere tropical cyclone track density using CMIP5 models. Climate
Dynamics, 52(9), 6065-6079.
Berg, A., Sheffield, J., & Milly, P. C. D. (2017). Divergent surface and total soil moisture
projections under global warming. Geophysical Research Letters, 44(1), 236–244,
https://doi.org/10.1002/2016GL071921
46
Birch, C. E., Roberts, M. J., Garcia-Carreras, L., Ackerley, D., Reeder, M. J., Lock, A. P., &
Schiemann, R. (2015). Sea-Breeze Dynamics and Convection Initiation: The Influence of
Convective Parameterization in Weather and Climate Model Biases. Journal of Climate,
28(20), 8093-8108, https://doi.org/10.1175/JCLI-D-14-00850.1
Blázquez, J., & Solman, S. A. (2017). Fronts and precipitation in CMIP5 models for the austral
winter of the Southern Hemisphere. Climate Dynamics, 50, 2705,
https://doi.org/10.1007/s00382-017-3765-z
Blázquez, J., & Solman, S. A. (2019). Relationship between projected changes in precipitation
and fronts in the austral winter of the Southern Hemisphere from a suite of CMIP5 models.
Climate Dynamics, 52(9–10), 5849–5860, https://doi.org/10.1007/s00382-018-4482-y
Bony, S., et al. (2006). How well do we understand and evaluate climate change feedback
processes? Journal of Climate, 19, 3445–3482, doi:10.1175/JCLI3819.1
Boschat, G., Pezza, A., Simmonds, I., Perkins, S., Cowan, T., & Purich, A. (2015). Large scale
and sub-regional connections in the lead up to summer heat wave and extreme rainfall
events in eastern Australia. Climate Dynamics, 44(7–8), 1823–1840,
https://doi.org/10.1007/s00382-014-2214-5
Brooks, H. E., Lee, J. W., & Craven, J. P. (2003). The spatial distribution of severe
thunderstorm and tornado environments from global reanalysis data. Atmospheric Research,
67–68, 73–94, doi:10.1016/S0169-8095(03)00045-0
Brooks, H. E. (2013). Severe thunderstorms and climate change. Atmospheric Research, 123,
129–138, http://dx.doi.org/10.1016/j.atmosres.2012.04.002
Brown, A., & Dowdy, A. (2019). Extreme wind gusts and thunderstorms in South Australia
analysed from 1979-2017. Bureau Research Report No. 034. Available at:
http://www.bom.gov.au/research/publications/researchreports/BRR-034.pdf (accessed 25
November 2020).
Brown, A., & Dowdy, A. (2021). Severe convection-related winds in Australia and their
associated environments. Journal of Southern Hemisphere Earth Systems Science,
https://doi.org/10.1071/ES19052
Bureau of Meteorology (BOM), & CSIRO (2020). State of the Climate 2020. Available from:
http://www.bom.gov.au/state-of-the-climate/
Burrell, A. L., Evans, J. P., & De Kauwe, M. G. (2020). Anthropogenic climate change has
driven over 5 million km 2 of drylands towards desertification. Nature Communications,
11(1), 1-11.
Cai, W., T. Cowan & M. Raupach (2009). Positive Indian Ocean Dipole events precondition
southeast Australia bushfires. Geophysical Research Letters 23, 3357-3360.
https://doi.org/10.1029/2009GL039902
Cai, W., Wang, G., Dewitte, B., Wu, L., Santoso, A., Takahashi, K., Yang, Y., Careric, A., &
McPhaden, M. J. (2018a). Increased variability of easten Pacific El Nino under greenhouse
warming. Nature, 564, 201-206. https://doi.org/10.1038/s41586-018-0776-9
Cai, W., Wang, G., Gan, B., Wu, L., Santoso, A., Lin, X., Chen, Z., Jia, F., & Yamagata, T.
(2018b): Stabilised frequency of extreme positive Indian Ocean Dipole under 1.5C
warming. Nature Communications, 9, 1419. https://doi.org/10.1038/s41467-018-03789-6
Cai, W., Wu, L., Lengaigne, M., Li, T., McGregor, S., Kug, J.-S., et al. (2019). Pantropical
climate interactions. Science, 363(6430), eaav4236. https://doi.org/10.1126/science.aav4236
47
Catto, J. L., Nicholls, N., Jakob, C., & Shelton, K. L. (2014). Atmospheric fronts in current and
future climates. Geophysical Research Letters, 41(21), 7642–7650,
https://doi.org/10.1002/2014GL061943
Catto, J. L., Jakob, C., & Nicholls, N. (2015). Can the CMIP5 models represent winter frontal
precipitation? Geophysical Research Letters, 42(20), 2015GL066015,
https://doi.org/10.1002/2015GL066015
Cechet, R. P., Sanabria, L. A., Divi, C. B., Thomas, C., Yang, T., Arthur, W. C., Dunford, M.,
Nadimpalli, K., Power, L., White, C. J. & Bennett, J. C. (2012). Climate Futures for
Tasmania: severe wind hazard and risk technical report.
Chambers, C. R. S., Brassington, G. B., Walsh, K. et al. (2015). Sensitivity of the distribution of
thunderstorms to sea surface temperatures in four Australian east coast lows. Meteorology
and Atmospheric Physics, 127, 499–517. doi:10.1007/s00703-015-0382-4
Chand, S. S., Dowdy, A. J., Ramsay, H. A., Walsh, K. J., Tory, K. J., Power, S. B., Bell, S. S.,
Lavender, S. L., Ye, H., & Kuleshov, Y. (2019). Review of tropical cyclones in the
Australian region: Climatology, variability, predictability, and trends. Wiley
Interdisciplinary Reviews: Climate Change, 10(5), p.e602.
Chapman, S., Thatcher, M., Salazar, A., Watson, J. E. M., & McAlpine, C.A. (2019). The
impact of climate change and urban growth on urban climate and heat stress in a subtropical
city. International Journal of Climatology, 39, 3013– 3030. https://doi.org/10.1002/joc.5998
Chen, J., Dai, A., Zhang, Y., & Rasmussen, K. L. (2020). Changes in convective available
potential energy and convective inhibition under global warming. Journal of Climate, 33(6),
2025-2050.
Chew L. W., Liu X., Li X-X, & Norford, L. K. (2020). Interaction between heat wave and urban
heat island: A case study in a tropical coastal city, Singapore. Atmospheric Research, 247,
105134, https://doi.org/10.1016/j.atmosres.2020.105134
Clarke, H., Lucas, C., & Smith, P. (2013). Changes in Australian fire weather between 1973 and
2010. International Journal of Climatology, 33(4), 931-944.
Clarke, H., Pitman, A. J., Kala, J., Carouge, C., Haverd, V., & Evans, J. (2016). An
investigation of future fuel load and fire weather in Australia. Climatic Change, 139 (3),
591-605.
Clarke, H., & Evans, J. P. (2019). Exploring the future change space for fire weather in
southeast Australia. Theoretical and Applied Climatology, 136(1), 513-527.
Clemesha, R. E. S., Guirguis, K., Gershunov, A., Small, I. J., & Tardy, A. (2018). California
heat waves: their spatial evolution, variation, and coastal modulation by low clouds. Climate
Dynamics, 50, 4285–4301, https://doi.org/10.1007/s00382-017-3875-7
Cruz, M. G., Sullivan, A. L., Gould, J. S., Sims, N. C., Bannister, A. J., Hollis, J. J., & Hurley,
R. J. (2012). Anatomy of a catastrophic wildfire: the Black Saturday Kilmore East fire in
Victoria, Australia. Forest Ecology and Management, 284, 269-285.
CSIRO & Bureau of Meteorology (2015). Climate Change in Australia Information for
Australia’s Natural Resource Management Regions: Technical Report, CSIRO and Bureau
of Meteorology, Australia.
Dawson, A., Palmer, T. N., & Corti, S. (2012) Simulating regime structures in weather and
climate prediction models. Geophysical Research Letters, 39, L21805,
doi:10.1029/2012GL053284
48
Di Virgilio, G., Evans, J. P., Blake, S. A., Armstrong, M., Dowdy, A. J., Sharples, J., & McRae,
R. (2019). Climate change increases the potential for extreme wildfires. Geophysical
Research Letters, 46(14), 8517-8526.
Diffenbaugh, N. S., M. Scherer, & R. J. Trapp (2013). Robust increases in severe thunderstorm
environments in response to greenhouse forcing. Proceedings of the National Academy of
Sciences, 110, 16361–16366, doi:10.1073/pnas.1307758110
Donohue, R. J., M. L. Roderick, T. R. McVicar, & G. D. Farquhar (2013). Impact of CO2
fertilization on maximum foliage cover across the globe’s warm, arid environments,
Geophysical Research Letters, 40, 3031–3035, doi:10.1002/grl.50563
Doswell, C. A., & Evans, J. S. (2003). Proximity sounding analysis for derechos and supercells:
An assessment of similarities and differences. Atmospheric Research, 67–68, 117–133.
Dowdy, A. J., Mills, G. A., Finkele, K., & de Groot, W. (2009). Australian fire weather as
represented by the McArthur forest fire danger index and the Canadian forest fire weather
index. Centre for Australian Weather and Climate Research Tech. Rep, 10, p.91.
Dowdy, A. J., & Mills, G. A. (2012). Atmospheric and fuel moisture characteristics associated
with lightning-attributed fires. Journal of Applied Meteorology and Climatology, 51(11),
2025-2037. https://doi.org/10.1175/JAMC-D-11-0219.1
Dowdy, A. J. (2014). Long‐term changes in Australian tropical cyclone numbers. Atmospheric
Science Letters, 15(4), 292-298. https://doi.org/10.1002/asl2.502
Dowdy, A., Grose, M., Timbal, B., Moise, A., Ekstrom, M., Bhend, J., & Wilson, L. (2015).
Rainfall in Australia’s eastern seaboard: a review of confidence in projections based on
observations and physical processes. Australian Meteorological and Oceanographic
Journal, 65(1), 107–126. https://doi.org/10.22499/2.6501.008
Dowdy, A. J. (2016). Seasonal forecasting of lightning and thunderstorm activity in tropical and
temperate regions of the world. Scientific Reports, 6, 1–10, doi:10.1038/srep20874
Dowdy, A. J., & J. L. Catto (2017). Extreme weather caused by concurrent cyclone, front and
thunderstorm occurrences. Scientific Reports, 1–8, doi:10.1038/srep40359
Dowdy, A. J., Fromm, M. D., & McCarthy, N. (2017). Pyrocumulonimbus lightning and fire
ignition on Black Saturday in southeast Australia. Journal of Geophysical Research:
Atmospheres, 122(14), 7342-7354. https://doi.org/10.1002/2017JD026577
Dowdy, A. J. (2018). Climatological variability of fire weather in Australia. Journal of Applied
Meteorology and Climatology, 57(2), 221-234. https://doi.org/10.1175/JAMC-D-17-0167.1
Dowdy, A. J., & Pepler, A. (2018). Pyroconvection risk in Australia: Climatological changes in
atmospheric stability and surface fire weather conditions. Geophysical Research Letters,
45(4), 2005-2013. https://doi.org/10.1002/2017GL076654
Dowdy, A. J., Pepler, A., Di Luca, A., Cavicchia, L., Mills, G., Evans, J., et al. (2019a). Review
of Australian east coast low pressure systems and associated extremes. Climate Dynamics,
53, 4887, https://doi.org/https://doi.org/10.1007/s00382-019-04836-8
Dowdy, A. J., Ye, H., Pepler, A., Thatcher, M., Osbrough, S. L., Evans, J. P., Di Virgilio, G., &
McCarthy, N. (2019b). Future changes in extreme weather and pyroconvection risk factors
for Australian wildfires. Scientific Reports, 9(1), 1-11. https://doi.org/10.1038/s41598-019-
46362-x
Dowdy, A. J. (2020a). Climatology of thunderstorms, convective rainfall and dry lightning
environments in Australia. Climate Dynamics, doi:10.1007/s00382-020-05167-9
49
Dowdy, A. J. (2020b). Seamless climate change projections and seasonal predictions for
bushfires in Australia. Journal of Southern Hemisphere Earth Systems Science, 70(1), 120-
138. https://doi.org/10.1071/ES20001
Drobinski, P., Bastin, S., Arsouze, T. et al. (2018). North-western Mediterranean sea-breeze
circulation in a regional climate system model. Climate Dynamics, 51, 1077–1093.
https://doi.org/10.1007/s00382-017-3595-z
Evans, J. P., Ji, F., Lee, C., Smith, P., Argüeso, D., & Fita, L. (2014). Design of a regional
climate modelling projection ensemble experiment–NARCliM. Geoscientific Model
Development, 7(2), 621-629.
Fasullo, J. T., Otto-Bliesner, B. L., & Stevenson, S. (2018). ENSO’s Changing Influence on
Temperature, Precipitation, and Wildfire In a Warming Climate. Geophysical Research
Letters, 1–10, https://doi.org/10.1029/2018GL079022
Field, R. D., Spessa, A. C., Aziz, N. A., Camia, A., Cantin, A., Carr, R., De Groot, W. J.,
Dowdy, A. J., Flannigan, M. D., Manomaiphiboon, K., & Pappenberger, F. (2015).
Development of a global fire weather database. Natural Hazards and Earth System
Sciences, 15(6), 1407-1423.
Fischer, E. M., Oleson, K. W., and Lawrence, D. M. (2012). Contrasting urban and rural heat
stress responses to climate change, Geophysical Research Letters, 39, L03705,
doi:10.1029/2011GL050576
Fita L., Evans, J. P., Argüeso, D., King, A., & Liu, Y. (2016). Evaluation of the regional climate
response in Australia to large-scale climate modes in the historical NARCliM simulations.
Climate Dynamics, 1–15, https://doi.org/10.1007/s00382-016-3484-x
Freund, M. B., Brown, J. R., Henley, B. J., Karoly, D. J., & Brown J. N. (2020). Warming
patterns affect El Nino diversity in CMIP5 and CMIP6 models. Journal of Climate,
10.1175/JCLI-D-19-0890.1
Fromm, M., Tupper, A., Rosenfeld, D., Servranckx, R., & McRae, R. (2006). Violent pyro‐
convective storm devastates Australia's capital and pollutes the stratosphere. Geophysical
Research Letters, 33(5).
Garfinkel, C. I., Waugh, D. W., & Polvani, L. M. (2015). Recent Hadley cell expansion: The
role of internal atmospheric variability in reconciling modeled and observed trends.
Geophysical Research Letters, 42(24), 10824–10831,
https://doi.org/10.1002/2015GL066942
Gartland, L. (2011). Heat Islands: Understanding and mitigating heat in urban areas. Routledge.
ISBN 9781849712989
Gensini, V. A., Ramseyer, C., & Mote, T. L. (2014). Future convective environments using
NARCCAP. International Journal of Climatology, 34, 1699–1705, doi:10.1002/joc.3769
Gensini, V. A., & Brooks, H. E. (2018). Spatial trends in United States tornado frequency. npj
Climate and Atmospheric Science, 1(1), 1-5. https://doi.org/10.1038/s41612-018-0048-2
Gibson, P. B., Perkins-Kirkpatrick, S. E., Uotila, P., Pepler, A. S., & Alexander, L. V. (2017).
On the use of self-organizing maps for studying climate extremes. Journal of Geophysical
Research, 122(7), 3891–3903, https://doi.org/10.1002/2016JD026256
Grise, K. M., Davis, S. M., Simpson, I. R., Waugh, D. W., Fu, Q., Allen, R. J., et al. (2019).
Recent tropical expansion: Natural variability or forced response? Journal of Climate, 32(5),
1551–1571, https://doi.org/10.1175/JCLI-D-18-0444.1
50
Grise, K. M., & Polvani, L. M. (2014). Southern Hemisphere Cloud–Dynamics Biases in

CMIP5 Models and Their Implications for Climate Projections. Journal of Climate, 27,
6074-6092. https://doi.org/10.1175/JCLI-D-14-00113.1
Grose, M., Timbal, B., Wilson, L., Bathols, J., & Kent, D. (2015). The subtropical ridge in
CMIP5 models, and implications for projections of rainfall in southeast Australia.
Australian Meteorological and Oceanographic Journal, 65, 90–106.
https://doi.org/10.22499/2.6501.007
Gross, M. H., Donat, M. G., & Alexander, L. V. (2019). Changes in daily temperature extremes
relative to the mean in CMIP5 models and observations. International Journal of
Climatology, 1–19. https://doi.org/10.1002/joc.6138
Grose, M. R., Narsey, S., Delage, F. P., Dowdy, A. J., Bador, M., Boschat, G., Chung, C.,
Kajtar, J. B., Rauniyar, S., Freund, M. B., & Lyu, K. (2020). Insights from CMIP6 for
Australia's future climate. Earth's Future, 8(5), p.e2019EF001469.
Guan, H., Kumar, V., Clay, R., Kent, C., Bennett, J., Ewenz, C., Hopkins, G., & Simmons, C.
T. (2016). Temporal and spatial patterns of air temperature in a coastal city with a slope
base setting. Journal of Geophysical Research: Atmospheres, 121(10): 2016JD025139.
https://doi.org/10.1002/2016JD025139
Harris, S., & Lucas, C. (2019). Understanding the variability of Australian fire weather between
1973 and 2017. PloS One, 14(9), p.e0222328.
Hasson, A. E. A., Mills, G. A., Timbal, B., & Walsh, K. (2009). Assessing the impact of climate
change on extreme fire weather events over southeastern Australia. Climate Research,
39(2), 159–172. https://doi.org/10.3354/cr00817
Hendon, H. H., Thompson, D. W. J., & Wheeler, M. C. (2007). Australian rainfall and surface
temperature variations associated with the Southern Hemisphere annular mode. Journal of
Climate, 20(11), 2452–2467. https://doi.org/10.1175/JCLI4134.1
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., et al. (2020). The ERA5 global
reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), 1999-2049.
Herold, N., Kala, J., & Alexander, L. V. (2016). The influence of soil moisture deficits on
Australian heatwaves. Environmental Research Letters, 11(6), https://doi.org/10.1088/1748-
9326/11/6/064003
Hirsch, A. L., Evans, J. P., Di Virgilio, G., Perkins-Kirkpatrick, S. E., Argüeso, D., Pitman, A.
J., et al. (2019). Amplification of Australian Heatwaves via Local Land-Atmosphere
Coupling. Journal of Geophysical Research: Atmospheres, 625–647,
https://doi.org/10.1029/2019JD030665
Holmes, J. D. (2002). A Re-analysis of Recorded Extreme Wind Speeds in Region A.
Australian Journal of Structural Engineering, 4, 29–40,
doi:10.1080/13287982.2002.11464905
Holmes, J. D. (2020) Land-falling tropical cyclones on the Queensland coast and implications
of climate change for wind loads, Australian Journal of Structural Engineering, 21, 135-
142, DOI: 10.1080/13287982.2020.1717842
Hoogewind, K. A., Baldwin, M. E., & Trapp, R. J. (2017). The Impact of Climate Change on
Hazardous Convective Weather in the United States: Insight from High-Resolution
Dynamical Downscaling, Journal of Climate, 30(24), 10081-10100.
51
Imran, H. M., Kala, J., Ng, A. W. M. & Muthukumaran, S. (2019). Impacts of future urban
expansion on urban heat island effects during heatwave events in the city of Melbourne in
southeast Australia. Quarterly Journal of the Royal Meteorological Society, 145, 2586–
2602, https://doi.org/10.1002/qj.3580
IPCC (2013). Climate Change 2013: The Physical Science Basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
[Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y.
Xia, V. Bex & P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA, 1535pp.
Jakob, D. (2010). Challenges in developing a high-quality surface wind-speed data-set for
Australia. Australian Meteorological and Oceanographic Journal, 60, 227-236,
10.22499/2.6004.001.
Jovanovic, B., Jones, D. A., & Collins, D. (2008). A high-quality monthly pan evaporation
dataset for Australia. Climatic Change, 87, 517–35.
Jovanovic, B., Collins, D., Braganza, K., Jakob, D. and Jones, D.A. (2011). A high-quality
monthly total cloud amount dataset for Australia. Climatic Change, 108(3), 485-517.
Keetch, J. J., & Byram, G. M. (1968). A drought index for forest fire control. USDA Forest
Service, Research Paper SE-38.
Kendon, E. J., et al. (2017). Do convection-permitting regional climate models improve
projections of future precipitation change? Bulletin of the American Meteorological Society,
98, 79–93, doi:10.1175/BAMS-D-15-0004.1
Kent, D. M., Kirono, D. G. C., Timbal, B., & Chiew, F. H. S. (2013). Representation of the
Australian sub-tropical ridge in the CMIP3 models. International Journal of Climatology,
33(1), 48–57, https://doi.org/10.1002/joc.3406
King, A. T., & A. D. Kennedy (2019). North American supercell environments in atmospheric
reanalyses and RUC-2. Journal of Applied Meteorology and Climatology, 58, 71–92,
doi:10.1175/JAMC-D-18-0015.1
Knutson, T., Camargo, S. J., Chan, J. C., Emanuel, K., Ho, C. H., Kossin, J., Mohapatra, M.,
Satoh, M., Sugi, M., Walsh, K., & Wu, L. (2020). Tropical cyclones and climate change
assessment: Part II: Projected response to anthropogenic warming. Bulletin of the American
Meteorological Society, 101(3), E303-E322.
Kuchera, E. L. & Parker, M. D. (2006). Severe Convective Wind Environments. Weather
Forecasting, 21(4), 595–612.
Leslie, L. M., Leplastrier, M., & Buckley, B. W. (2008). Estimating future trends in severe
hailstorms over the Sydney Basin: A climate modelling study. Atmospheric Research, 87,
37–51.
Lim, E.-P., Hendon, H. H., Arblaster, J. M., Delage, F., Nguyen, H., Min, S.-K., & Wheeler, M.
C. (2016). The impact of the Southern Annular Mode on future changes in Southern
Hemisphere rainfall. Geophysical Research Letters, 2016GL069453,
https://doi.org/10.1002/2016GL069453
Lim, E.-P., Hendon, H. H., Boschat, G., Hudson, D., Thompson, D. W. J., Dowdy, A. J., &
Arblaster, J. M. (2019). Australian hot and dry extremes induced by weakenings of the
stratospheric polar vortex. Nature Geoscience, 12(11), 896–901,
https://doi.org/10.1038/s41561-019-0456-x
52
Lim, E. P., Hendon, H. H., Butler, A. H., Thompson, D. W., Lawrence, Z., Scaife, A. A.,
Shepherd, T. G., Polichtchouk, I., Nakamura, H., Kobayashi, C., & Comer, R. (2021). The
2019 Southern Hemisphere stratospheric polar vortex weakening and its impacts. Bulletin of
the American Meteorological Society, 1-50.
Loughran, T. F., Pitman, A. J., & Perkins-Kirkpatrick, S. E. (2019). The El Niño–Southern
Oscillation’s effect on summer heatwave development mechanisms in Australia. Climate
Dynamics, 52(9–10), 6279–6300. https://doi.org/10.1007/s00382-018-4511-x
Lucas, C., (2010). A High-quality Historical Humidity Database for Australia. CAWCR
Technical Report 24, Bureau of Meteorology, Docklands, Victoria, Australia.
Maloney, E. D., Adames, Á. F., & Bui, H. X. (2019). Madden–Julian oscillation changes under
anthropogenic warming. Nature Climate Change, 9(1), 26–33,
https://doi.org/10.1038/s41558-018-0331-6
Marjani, S., Alizadeh-Choobari, O., & Irannejad, P. (2019). Frequency of extreme El Niño and
La Niña events under global warming. Climate Dynamics, 53(9), 5799–5813,
https://doi.org/10.1007/s00382-019-04902-1
Marshall, A. G., Hudson, D., Wheeler, M. C., Alves, O., Hendon, H. H., Pook, M. J., & Risbey,
J. S. (2013). Intra-seasonal drivers of extreme heat over Australia in observations and
POAMA-2. Climate Dynamics, 43(7–8), 1915–1937, https://doi.org/10.1007/s00382-013-
2016-1
Marshall, G. J. (2003). Trends in the Southern Annular Mode from observations and reanalyses.
Journal of Climate, 16(24), 4134–4143, https://doi.org/10.1175/1520-
0442(2003)016<4134:TITSAM>2.0.CO;2
Masouleh, Z. P., Walker, D. J., & McCauley Crowther, J. (2019). A Long-Term Study of Sea-
Breeze Characteristics: A Case Study of the Coastal City of Adelaide, Journal of Applied
Meteorology and Climatology, 58(2), 385-400.
Mastrandrea, M. D., et al. (2011). The IPCC AR5 guidance note on consistent treatment of
uncertainties: a common approach across the working groups. Climatic Change, 108(4),
675-691.
McArthur, A. G. (1967). Fire behaviour in eucalypt forests. Forestry and Timber Bureau, No
107, Canberra, Australia.
McKeon, G. M., et al. (2009). Climate change impacts on northern Australian rangeland
livestock carrying capacity: a review of issues, Rangel. J., 31, 1–29.
McRae, R. H., Sharples, J. J., Wilkes, S. R., & Walker, A. (2013). An Australian pyro-
tornadogenesis event. Natural Hazards, 65(3), 1801-1811.
McVicar, T. R., Van Niel, T. G., Li, L. T., Roderick, M. L., Rayner, D. P., Ricciardulli, L., &
Donohue, R. J. (2008). Wind speed climatology and trends for Australia, 1975–2006:
Capturing the stilling phenomenon and comparison with near-surface reanalysis output.
Geophysical Research Letters, 35(20), L20403, https://doi.org/10.1029/2008GL035627
Mills, G. A., & McCaw, W. L. (2010). Atmospheric stability environments and fire weather in
Australia: extending the Haines Index. CAWCR Technical Report, Centre for Australia and
Weather Climate Research, Melbourne, Australia.
Myers, T. A., & Norris, J. R. (2015). On the Relationships between Subtropical Clouds and
Meteorology in Observations and CMIP3 and CMIP5 Models. Journal of Climate, 28, 8,
2945-2967, https://doi.org/10.1175/JCLI-D-14-00475.1
53
NESP (2020). Earth Systems and Climate Change Hub. 2020. Scenario analysis of climate-
related physical risk for buildings and infrastructure: climate science guidance. Technical
report by the National Environmental Science Program (NESP) Earth Systems and Climate
Change Science (ESCC) Hub for the Climate Measurement Standards Initiative, ESCC Hub
Report No.21, ISBN: 978-0-6489444-1-6.
Nguyen, H., Lucas, C., Evans, A., Timbal, B., & Hanson, L. (2015). Expansion of the Southern
Hemisphere Hadley cell in response to greenhouse gas forcing. Journal of Climate, 28(20),
8067–8077, https://doi.org/10.1175/JCLI-D-15-0139.1
Parker, T. J., Berry, G. J., & Reeder, M. J. (2013). The influence of tropical cyclones on
heatwaves in Southeastern Australia. Geophysical Research Letters,
https://doi.org/10.1002/2013GL058257
Parker, T. J., Berry, G. J., Reeder, M. J., & Nicholls, N. (2014). Modes of climate variability
and heat waves in Victoria, southeastern Australia. Geophysical Research Letters, 41(19),
6926–6934, https://doi.org/10.1002/2014GL061736
Patterson, M., Bracegirdle, T., & Woollings, T. (2019). Southern Hemisphere atmospheric
blocking in CMIP5 and future changes in the Australia‐New Zealand sector. Geophysical
Research Letters, 5, 1–10, https://doi.org/10.1029/2019gl083264
Pepler, A. S., Alexander, L. V, Evans, J. P., & Sherwood, S. C. (2015). Zonal winds and
southeast Australian rainfall in global and regional climate models. Climate Dynamics, 46,
123–133, https://doi.org/10.1007/s00382-015-2573-6
Pepler, A. S., Di Luca, A., Ji, F., Alexander, L. V., Evans, J. P., & Sherwood, S. C. (2016),
Projected changes in east Australian midlatitude cyclones during the 21st century.
Geophysical Research Letters, 43, 334– 340, doi:10.1002/2015GL067267
Pepler, A., Ashcroft, L., & Trewin, B. (2018). The relationship between the subtropical ridge
and Australian temperatures. Journal of Southern Hemisphere Earth Systems Science, 68(1),
201–214, https://doi.org/10.22499/3.6801.011
Pepler, A. S., Dowdy, A. J., & Hope, P. (2021). The differing role of weather systems in
southern Australian rainfall between 1979 – 1996 and 1997 – 2015. Climate Dynamics,
https://doi.org/10.1007/s00382-020-05588-6
Perkins, S. E., Argüeso, D., & White, C. J. (2015). Relationships between climate variability,
soil moisture, and Australian heatwaves. Journal of Geophysical Research: Atmospheres,
120(16), 2015JD023592, https://doi.org/10.1002/2015JD023592
Perkins-Kirkpatrick, S. E., White, C. J., Alexander, L. V., Argüeso, D., Boschat, G., Cowan, T.,
et al. (2016). Natural hazards in Australia: heatwaves. Climatic Change, 139(1), 101–114.
https://doi.org/10.1007/s10584-016-1650-0
Pistotnik, G., P. Groenemeijer, & R. Sausen (2016). Validation of Convective Parameters in
MPI-ESM Decadal Hindcasts (1971-2012) against ERA-Interim Reanalyses.
Meteorologische Zeitschrift, 25, 753–766, doi:10.1127/metz/2016/0649.
Power, S., Casey, T., Folland, C., Colman, A., & Mehta, V. (1999). Inter-decadal modulation of
the impact of ENSO on Australia. Climate Dynamics, 15(5), 319–324.
Power, S. B., & Callaghan, J. (2016). Variability in severe coastal flooding, associated storms,
and death tolls in Southeastern Australia since the mid-nineteenth century. Journal of
Applied Meteorology and Climatology, 55(5), 1139–1149, https://doi.org/10.1175/JAMC-
D-15-0146.1
54
Proctor, F. H. (1989). Numerical simulations of an isolated microburst. Part II: sensitivity

experiments. Journal of the Atmospheric Science, 46, 2143-2165
Púčik, T., et al. (2017). Future changes in European severe convection environments in a
regional climate model ensemble. Journal of Climate, 30, 6771–6794, doi:10.1175/JCLI-D-
16-0777.1
Purich, A., Cowan, T., Cai, W., van Rensch, P., Uotila, P., Pezza, A., et al. (2014). Atmospheric
and Oceanic Conditions Associated with Southern Australian Heat Waves: A CMIP5
Analysis. Journal of Climate, 27(20), 7807–7829, https://doi.org/10.1175/JCLI-D-14-
00098.1
Quinting, J. F., & Reeder, M. J. (2017). Southeastern Australian Heat Waves from a Trajectory
Viewpoint. Monthly Weather Review, 145(10), 4109–4125, https://doi.org/10.1175/MWR-
D-17-0165.1
Quinting, J. F., Parker, T. J., & Reeder, M. J. (2018). Two Synoptic Routes to Subtropical Heat
Waves as Illustrated in the Brisbane Region of Australia. Geophysical Research Letters,
700–708, https://doi.org/10.1029/2018GL079261
Rädler, A. T., Groenemeijer, P., Faust, E., & Sausen, R. (2018). Detecting severe weather trends
using an additive regressive convective hazard model (AR-CHaMo). Journal of Applied
Meteorology and Climatology, 57(3), 569-587.
Rasmussen, K. L., A. F. Prein, R. M. Rasmussen, K. Ikeda, & C. Liu (2017). Changes in the
convective population and thermodynamic environments in convection-permitting regional
climate simulations over the United States. Climate Dynamics, 1–26,
http://dx.doi.org/10.1007/s00382-017-4000-7
Raupach, T. H., Martius, O., Allen, J. T., Kunz, M., Lasher-Trapp, S., Mohr, S., Rasmussen, K.
L., Trapp, R. J. and Zhang, Q. (2021). The effects of climate change on hailstorms. Nature
Reviews Earth & Environment, 1-14.
Reeder, M. J., Spengler, T., & Musgrave, R. (2015). Rossby waves, extreme fronts, and
wildfires in southeastern Australia. Geophysical Research Letters, 42(6), 2015-2023.
Rogers, C. D. W., Gallant, A. J. E., & Tapper, N. J. (2019). Is the urban heat island exacerbated
during heatwaves in southern Australian cities? Theoretical and Applied Climatology,
137(1), 441–457, https://doi.org/10.1007/s00704-018-2599-x
Rudeva, I., & Simmonds, I. (2015). Variability and trends of global atmospheric frontal activity
and links with large-scale modes of variability. Journal of Climate, 28(8), 3311–3330,
https://doi.org/10.1175/JCLI-D-14-00458.1
Schwalm, C. R., Glendon, S., & Duffy, P. B. (2020). RCP8. 5 tracks cumulative CO2
emissions. Proceedings of the National Academy of Sciences, 117(33), 19656-19657.
Scott, A. A., Waugh D. W., & Zaitchik, B. F. (2018). Reduced Urban Heat Island intensity
under warmer conditions. Environmental Research Letters, 13, 064003,
https://doi.org/10.1088/1748-9326/aabd6c
Seeley, J. T., & Romps, D. M. (2015). The effect of global warming on severe thunderstorms in
the United States. Journal of Climate, 28, 2443–2458, doi:10.1175/JCLI-D-14-00382.1
Smith, B. T., Castellanos, T. E., Winters, A. C., Mead, C. M., Dean, A. R., & Thompson, R. L.
(2012). Measured Severe Convective Wind Climatology and Associated Convective Modes
of Thunderstorms in the Contiguous United States, 2003–09. Weather Forecasting, 28(1),
229–236.
55
Soderholm, J. S., H. McGowan, H. Richter, K. Walsh, T. M. Weckwerth, & M. Coleman

(2017). An 18-year climatology of hailstorm trends and related drivers across southeast
Queensland, Australia. Quarterly Journal of the Royal Meteorological Society., 143, 1123–
1135, doi:10.1002/qj.2995
Spassiani, A. C. (2020) Climatology of severe convective wind gusts in Australia (PhD. thesis).
The University of Queensland School of Civil Engineering, doi: 10.14264/uql.2020.901
Taszarek, M., Brooks, H. E., & Czernecki, B. (2017). Sounding-Derived Parameters Associated
with Convective Hazards in Europe. Monthly Weather Review, 145, 1511–1528,
doi:10.1175/MWR-D-16-0384.1
Taszarek, M., Allen, J., Brooks, H., Pilguj, N., & Czernecki, B. (2020). Differing trends in
United States and European severe thunderstorm environments in a warming climate.
Bulletin of the Americal Meteorological Society, 1–51, doi:10.1175/BAMS-D-20-0004.1
Taylor, K. E., Stouffer, R. J. & Meehl, G. A, (2012). An Overview of CMIP5 and the
experiment design. Bull. Amer. Met. Soc., 93, 485-498, https://doi.org/10.1175/BAMS-D-
11-00094.1
Thatcher, M., & McGregor, J. L. (2011). A technique for dynamically downscaling daily-
averaged GCM datasets using the conformal cubic atmospheric model. Monthly Weather
Review, 139(1), 79-95.
Thatcher, M., et al. (2021). Technical Report on downscaling for the ESCI project. CSIRO and
BoM, Victoria, Australia. Report available from ESCI project website:
https://www.climatechangeinaustralia.gov.au/en/projects/esci/
Timbal, B., & Drosdowsky, W. (2013). The relationship between the decline of Southeastern
Australian rainfall and the strengthening of the subtropical ridge. International Journal of
Climatology, 33(4), 1021–1034, https://doi.org/10.1002/joc.3492
Trapp, R. J., N. S. Diffenbaugh, H. E. Brooks, M. E. Baldwin, E. D. Robinson, & J. S. Pal
(2007). Changes in severe thunderstorm environment frequency during the 21st century
caused by anthropogenically enhanced global radiative forcing. Proceedings of the National
Academy of Science, 104, 19719–19723, doi:10.1073/pnas.0705494104
Trapp, R. J., K. A. Hoogewind, & S. Lasher-Trapp (2019). Future changes in hail occurrence in
the United States determined through convection-permitting dynamical downscaling.
Journal of Climate, 32, 5493–5509, doi:10.1175/JCLI-D-18-0740.1
Ukkola, A. M., Pitman, A. J., Donat, M. G., De Kauwe, M. G., & Angélil, O. (2018).
Evaluating the Contribution of Land-Atmosphere Coupling to Heat Extremes in CMIP5
Models. Geophysical Research Letters, 45(17), 9003–9012,
https://doi.org/10.1029/2018GL079102
Ukkola, A. M., Roderick, M. L., Barker, A., & Pitman, A. J. (2019). Exploring the stationarity
of Australian temperature, precipitation and pan evaporation records over the last
century. Environmental Research Letters, 14(12), p.124035.
Ukkola, A. M., De Kauwe, M. G., Roderick, M. L., Abramowitz, G., & Pitman, A. J. (2020).
Robust future changes in meteorological drought in CMIP6 projections despite uncertainty
in precipitation. Geophysical Research Letters, 47(11), p.e2020GL087820.
van Oldenborgh, G. J., Krikken, F., Lewis, S., Leach, N. J., Lehner, F., Saunders, K. R., van
Weele, M., Haustein, K., Li, S., Wallom, D. and Sparrow, S. (2021). Attribution of the
Australian bushfire risk to anthropogenic climate change. Natural Hazards and Earth
System Sciences, 21(3), 941-960.
56
Van Wagner, C. E. (1974). Structure of the Canadian forest fire weather index, Vol. 1333,
Ontario: Environment Canada, Forestry Service.
Voigt, A., Albern, N., Ceppi, P., Grise, K., Li, Y., & Medeiros, B. (2020). Clouds, radiation,
and atmospheric circulation in the present‐day climate and under climate change. Wiley
Interdisciplinary Reviews: Climate Change, p.e694.
Walsh, K., et al. (2016). Natural hazards in Australia: storms, wind and hail. Climate Change,
139, 55–67, doi:10.1007/s10584-016-1737-7
Wang, C., Wang, X. & Khoo, Y. B. (2013). Extreme wind gust hazard in Australia and its
sensitivity to climate change. Nat. Hazards, 67, 549–567, DOI:10.1007/s11069-013-0582-5
Wang, G., & Cai, W. (2013). Climate-change impact on the 20th-century relationship between
the Southern Annular Mode and global mean temperature. Scientific Reports 3, 2039,
https://doi.org/10.1038/srep02039
Wang, B., Chen, G. & Liu, F. (2019). Diversity of the Madden-Julian Oscillation. Science
Advances, 5, eaax0220. https://doi.org/10.1126/sciadv.aax0220
Waugh, D. W., Garfinkel, C. I., & Polvani, L. M. (2015). Drivers of the recent tropical
expansion in the southern hemisphere: Changing SSTs or ozone depletion? Journal of
Climate, 28(16), 6581–6586, https://doi.org/10.1175/JCLI-D-15-0138.1
Weckwerth, T. M. (2000). The effect of small-scale moisture variability on thunderstorm
initiation. Monthly Weather Review, 128(12), 4017-4030.
Wehrli, K., Guillod, B. P., Hauser, M., Leclair, M., & Seneviratne, S. I. (2019). Identifying key
driving processes of recent heatwaves and droughts, 21, 13643,
https://doi.org/10.1029/2019JD030635
Williams, S., Nitschke, M., Sullivan, T., Tucker, G. R., Weinstein, P., Pisaniello, L., Parton, K.
A., & Bi, P. (2012). Heat and health in Adelaide, South Australia: Assessment of heat
thresholds and temperature relationships. Science of the Total Environment, 414, 126–133,
https://doi.org/10.1016/j.scitotenv.2011.11.038
Woollings, T., Barriopedro, D., Methven, J., Son, S. W., Martius, O., Harvey, B., et al. (2018).
Blocking and its Response to Climate Change. Current Climate Change Reports,
https://doi.org/10.1007/s40641-018-0108-z
Wouters, H., et al. (2017). Heat stress increase under climate change twice as large in cities as
in rural areas: A study for a densely populated midlatitude maritime region, Geophysical
Research Letters., 44, 8997– 9007, doi:10.1002/2017GL074889
Zhao, L., Oppenheimer, M., Zhu, Q., Baldwin, J. W., Ebi, K. L., Bou-Zeid, E., Guan, K., & Liu,
X. (2018). Interactions between urban heat islands and heat waves. Environmental Research
Letters, 13, 034003, https://doi.org/10.1088/1748-9326/aa9f73
57

ESCI Technical Report - Standardised Method - 1

Uploaded by

Copyright:

Available Formats

ESCI Technical Report - Standardised Method - 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ESCI Technical Report - Standardised Method - 1

Uploaded by

Copyright:

Available Formats

Extreme temperature, wind and bushfire weather

projections using a standardised method

Bureau Research Report - 055

Extreme temperature, wind and bushfire weather

CSIRO Oceans and Atmosphere, Aspendale, Australia

Bureau Research Report No. 055

National Library of Australia Cataloguing-in-Publication entry

Series: Bureau Research Report – BRR055

Enquiries should be addressed to:

Lead Author: Andrew Dowdy

Copyright and Disclaimer

1. INTRODUCTION AND OVERVIEW

2.1 Standardised method for projections information

Step 1 – Produce the Lines of Evidence Table

Step 2 – Produce the Projections Likelihood Information

Additional details on the method

High agreement Medium Medium-high High confidence

2.2 Examples of method outputs being used

3. EXTREME TEMPERATURE PROJECTIONS

3.2 Summaries for physical processes

Cloud cover and solar radiation

Blocking / high pressure systems

Modes of variability - ENSO

Modes of variability – IOD

Modes of variability - SAM

Modes of variability – MJO

Urban effects including urban heat island

3.3 Summaries for historical climate

Model assessment for historical climate

3.4 Summaries for future climate

3.5 Lines of evidence table

Subtropical ridge Moderate influence, primarily in southern Australia. Potential increase

Fronts Moderate influence. Future change uncertain.

Blocking / High Moderate influence. Future change uncertain.

ENSO Small to moderate influence. Uncertain future change; potentially more

IOD Small to moderate influence. Uncertain future change; potentially more

SAM Small to moderate influence. Positive trend in SAM relevant for

MJO Small influence. Uncertain future change.

GCMs: CMIP5 Strong increase (high confidence).

RCM: CCAM Strong increase (high confidence).

RCM: NARCliM Strong increase (high confidence).

3.6 Projections and confidence information

4. EXTREME WIND PROJECTIONS

4.2 Summaries for physical processes

Combining these factors through the use of environmental thunderstorm diagnostics

Severe convective wind environments

Modes of variability – ENSO, IOD and SAM

Other phenomena that can cause severe wind gusts

4.3 Summaries for historical climate

Trends in severe convective winds based on diagnostic methods

4.4 Summaries for future climate

Leslie et al. (2008) used a convection-permitting model to dynamically downscale

4.5 Lines of evidence table

Thunderstorm Moderate-strong influence. More favourable environments in

SCW Strong influence. Many uncertainties and few studies to date.

Thunderstorm Strong influence. Uncertain changes (relating to extratropical

ENSO Small influence. Uncertain future change.

IOD Small influence (moderate in northeast). Uncertain future change.

SAM Moderate influence in east. Projected shift towards positive