1 Introduction

Australia’s size and varied climates mean that it is affected by a range of weather-related natural hazards, including tropical and extra-tropical storms and associated extreme wind and hail, coastal and inland floods, heatwaves and bushfires. These hazards cause multiple human and environmental impacts, and collectively account for 93 % of Australian insured losses (Schuster 2013). In addition, drought—often treated distinctly from other hazards due to its more gradual onset—can cause substantial reductions in agricultural productivity, and places stress on municipal and industrial water resources and natural ecosystems.

Evidence is building that the frequency and cost of natural hazards are increasing both in Australia (Insurance Council of Australia 2013; Schuster 2013) and globally (Munich Re 2014). However, understanding the cause of these changes has proved to be difficult, with increases in reporting rates (Munich Re 2014), changes in societal exposure and vulnerability (Bouwer 2011; Neumayer and Barthel 2011) and anthropogenic climate change (IPCC 2013) all potentially playing a role in explaining the observed changes. Yet although the potential causes are many, correct attribution of the observed changes is necessary in order to identify appropriate policy responses, and to predict how the frequency and severity of natural hazards might change in the future.

This Special Issue focuses on the specific role of large-scale climatic changes on the observed and future incidence of Australian natural hazards. The Special Issue is divided into seven papers, each covering a major class of climate-influenced natural hazard: floods, drought, storms (including wind and hail), coastal extremes, bushfires, heatwaves and frost. The work was initiated by the Working Group on Trends and Extremes from the Australian Water and Energy Exchanges (OzEWEX) initiative, which is a regional hydroclimate project run under the auspices of the Global Energy and Water Exchanges (GEWEX) initiative.

2 Linking large-scale climate processes to Australian natural hazards

The structure of the Special Issue is illustrated in Fig. 1, in which anthropogenic climate change and associated large-scale climate patterns are related to each natural hazard through a complex cascade of processes and scales. Although the emphasis is on the natural hazards (bottom row of Fig. 1), the significant interrelationships between the hazards themselves, as well as between each hazard and larger-scale climate, has necessitated reviewing each hazard as part of a larger, interconnected system. This section briefly summarises the approach taken in this Special Issue to account for these interrelationships.

Fig. 1
figure 1

Illustration of the complex processes that link large-scale climate variability to a natural hazard. The arrows illustrate the processes for floods (see Johnson et al. 2016, for further information)

Starting at the global scale, Australia’s natural hazards are influenced by changes in the global energy balance, as well as by shifts in global circulation patterns. These changes are documented in the Intergovernmental Panel on Climate Change (IPCC) reports (e.g. IPCC 2013) and are not covered further in the Special Issue. Australia’s climate is also influenced by several hemispheric-scale patterns of climate variability, which can cause periods of lowered or heightened potential hazard activity. The most important patterns for Australia are the El Niño-Southern Oscillation phenomenon, the Indian Ocean Dipole and the Southern Annular Mode (e.g. Risbey et al. 2009b). An updated summary of historical and future changes to these patterns, as well as their connection to each of the seven natural hazards reviewed in this Special Issue, is provided in Table 1, and is referred to in a number of the specific hazard papers.

Table 1 Observed and projected changes in hemispheric-scale patterns of climate variability, and their link to Australian natural hazards

At smaller spatial and temporal scales, natural hazards are influenced by a variety of meteorological processes. A number of these processes have been summarised in Walsh et al. (2016), and include tropical cyclones, extratropical cyclones and their cold fronts, thunderstorms and east coast lows (coastal low pressure systems along parts of the east Australian coastline). These weather systems are often hazards themselves (e.g. leading to extreme wind and hail), and some are also causes of other hazards including floods (reviewed in Johnson et al. 2016) and coastal extremes (reviewed in McInnes et al. 2016).

In attempting to understand and attribute historical and future changes to natural hazards, significant emphasis is often placed on understanding changes to various atmospheric and oceanic variables, including temperature, rainfall, wind, humidity and atmospheric pressure (green circles in Fig. 1). There are multiple reasons for focusing on these variables as indicators of large-scale change, including the availability of long instrumental records, the ability of climate models to simulate the variables and the relatively limited influence of other human activities (e.g. land use change and the regulation of river systems) that can confound attempts to directly attribute changes to the natural hazards themselves. However, the connection between these variables and each natural hazard can be complex, with multiple variables usually acting jointly to influence the hazard (Leonard et al. 2014). For example, extreme rainfall is generally regarded as the proximate cause for most fluvial floods; however, annual average rainfall and the variables that drive evapotranspiration can collectively influence the catchment moisture content prior to the extreme rainfall event, and thus can also have a significant influence on flood magnitude (see Johnson et al. 2016, for a more detailed discussion).

In many cases, the mechanisms by which the atmospheric and oceanic variables and processes influence hazards are common to multiple hazards, albeit with subtle (but often important) distinctions. For example, heatwaves, frosts, bushfires and droughts are all influenced by atmospheric temperature, but in different ways. Heatwaves are one or several days of extremely high temperature (Perkins-Kirkpatrick et al. 2016), whereas frosts occur on timescales that are similar to heatwaves but at the other end of the temperature scale. Interestingly, Crimp et al. (2016) show somewhat surprisingly that the prevalence of frosts can increase despite an increase in mean atmospheric temperature. In the case of bushfires, high (but not necessary extreme) temperatures are a necessary but not sufficient condition for the occurrence of severe bushfires (Sharples et al. 2016). Finally, the relationship between temperature and drought arises through evapotranspiration processes that occur on timescales of months or years (Kiem et al. 2016).

Models are commonly used to describe our understanding of the relationship between the atmospheric and oceanic variables and the natural hazard, and these are depicted as red arrows in Fig. 1 (using the hazard ‘floods’ for illustration). The need for models arises because in many cases historical records of the natural hazards themselves are sparse, so that historical changes to the hazards need to be inferred from our understanding of changes to key causative processes. Therefore, models linking the climatic and meteorological variables (green circles, Fig. 1) to the hazards (blue circles, Fig. 1) often represent the primary line of evidence for how the hazards are affected by climate change. It is therefore critical to scrutinise the assumptions in the models, including decisions related to the processes that are included and the way they are represented, as this can have a significant influence on assessments of historical and future changes to the hazard.

There are also important interrelations between each of the natural hazards themselves (blue arrows, Fig. 1). For example, the prevalence of droughts can influence whether a catchment is wet or dry prior to a heavy rainfall event (linking drought and flood), whereas fires can influence the vegetation and soil properties of the catchment, and thus affect the conversion of rainfall to runoff (linking fire, drought and flood). In estuarine catchments, coastal processes including mean sea level and storm tides can combine with the fluvial flood to increase the overall flood hazard (linking sea level extremes with flood). These complex linkages between atmospheric/oceanic variables and the hazards highlight the need to take a consistent and unified approach to reviewing the evidence of change across all of the Australian natural hazards.

Given the complexity of the hazards and their causative mechanisms, this Special Issue therefore takes the following approach to reviewing historical and projected changes to Australian natural hazards:

  • Information on historical and projected changes in the hazards themselves (blue circles, Fig. 1) is covered in the relevant hazard paper. The models linking atmospheric and oceanic variables to that hazard are also covered in those papers (red lines, Fig. 1) as research into each hazard is typically informed by a hazard-specific set of models. Finally, the influence of other hazards on the topical hazard of each paper (blue lines, Fig. 1) is also covered; for example the presence of a drought can influence the catchment moisture stores, which is one of the factors that influence the flood hazard.

  • The atmospheric and oceanic variables (green circles, Fig. 1) are each covered in the most relevant natural hazard paper. A guide to where individual atmospheric and oceanic variables are covered is provided in Table 2.

    Table 2 Index of the atmospheric and oceanic variables that are described in each paper
  • The influence of large-scale patterns of climate variability that can influence Australian natural hazards are summarised in Table 1, with more detailed information provided in various Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report chapters, most notably the chapters by Christensen et al. (2013) and Hartmann et al. (2013).

3 Knowledge gaps and future research needs

This Special Issue documents our current understanding on historical and possible future climatic changes to the frequency and severity of Australian natural hazards. Although the science of detecting and attributing changes in the historical natural hazards—and developing projections of future changes—is progressing rapidly, a variety of knowledge gaps still exist. The authors of each paper have therefore identified research priorities for the hazard that would lead to a significant improvement in our collective understanding of the role of climate change in Australian natural hazards over a timeframe of about a decade.

Numerous suggestions for research priorities were common to many of the papers, including the need to revitalise our observational network to specifically monitor changes to the prevalence of Australia’s natural hazards. Similarly, increasing the resolution of our large-scale climate models (as well as including improved physics schemes to simulate certain processes such as tropical cyclones) continues to be a major priority, as it enables the inclusion of a greater number of scales within a single modelling framework. For many of the hazards, the role of paleo-climate data, which can assist in placing changes to natural hazards within a longer historical context, was also identified as a potential avenue for augmenting the often limited instrumental records.

There were also more specific suggestions that were unique to each hazard. Examples include the need for a unified framework to identify atmospheric heatwave events (Perkins-Kirkpatrick et al. 2016), and the need to better understand the role of alternative runoff-generating mechanisms and their relationship to future changes in flood risk (Johnson et al. 2016). In many cases the authors also called for better integration of research across different natural hazards; for example, the connections between drought and the land-atmosphere feedbacks that produce heatwaves (Kiem et al. 2016; Perkins-Kirkpatrick et al. 2016), and the link between coastal and inland processes in the context of flood hazard in estuarine regions (Johnson et al. 2016; McInnes et al. 2016).

Although most papers in this Special Issue highlighted the complexity and high levels of uncertainty of attributing historical changes in natural hazards and developing projections of future changes, there are grounds for optimism that the state of the science is improving. Land- and space-based remote sensing technology continues to yield data that enable investigations of change at increasing resolutions across the Australian continent; the increase in computing power and storage is leading to increasingly advanced models that can bridge a greater range of scales; and the increased information from the paleoclimate community is leading to improved understanding of how natural hazards have changed over long timescales. Furthermore, research into changes in natural hazards requires a focus on fostering interdisciplinary collaborations, and initiatives such as GEWEX and OzEWEX continue to serve the function of enhancing dialogue and collaborations between experts in diverse disciplines including meteorology, hydrology, oceanography, ecology, paleoclimatology, geography, engineering and statistics.

It is therefore hoped that, in addition to summarising the current state-of-the-science, this Special Issue will also provoke discussion and debate about future research priorities and directions. Only by taking a coordinated and strategic approach—one that accounts for the wide range of scales and processes that influence each hazard—will we be able to overcome the substantial scientific obstacles involved in understanding the nature and causes of historical and future changes to Australia’s natural hazards.