Extreme weather and climate events (e.g., heat waves, droughts, heavy rainfall, hurricanes) have always posed risks to human society. A matter of growing interest, however, is the degree to which humans are changing these risks through anthropogenic climate change. This concern has been driven by the growing impacts on ecosystems, communities, and infrastructure of recent extreme events across the United States and the world.
Efforts to attribute the causes of individual extreme events need to be understood in the broader context of what we already know about climate change. Humans have contributed to warming of the climate system globally (predominantly due to anthropogenic greenhouse gas [GHG] emissions). This finding is supported by multiple lines of evidence that originate from data from observing systems across the globe on land and sea and in the atmosphere and from structurally different analyses of multiple components of the climate system. A substantial body of evidence also shows that climate change has led to discernible and quantifiable changes in the intensity and/or frequency of some types of extremes (Donat et al., 2013; IPCC, 2014; Melillo et al., 2014; Seneviratne et al., 2012; Figure 1.1).
Extreme weather is one way that people experience climate change. Extreme events are abrupt, occur in the present, and are highly visible, as opposed to long-term climate change trends that seem abstract, distant, gradual, and complicated (Howe et al., 2014). The global news includes reports on extreme weather or climate events on a regular basis: for example, in 2015 there was a May-June India-Pakistan heat wave, both a “1,000-year rainfall event” in South Carolina (Figure 1.2) and Hurricane Patricia, the “strongest eastern Pacific or Atlantic hurricane in the historical record,” in October, as well as widespread flooding in northern England in December. Each of these cases has led to questions from the media and the public about whether the events were “caused” by climate change. Attribution draws the explicit connection between climate science as a whole and the specific event in the news, making the science concrete in a way that statements about broader trends and future projections do not.
Given that climate change affects the climate system globally, it is impossible to rule out some contribution from climate change to any extreme event. Each extreme event, however, has a host of possible natural and anthropogenic causes in addition to anthropogenic climate change. Examples of natural causes include large-scale circulation, internal modes of climate variability, and some aerosol effects (e.g., marine aerosol, stratospheric and volcanic aerosol). Furthermore, the resulting impacts of that event can be mitigated or exacerbated by other factors (e.g., the local topography, land use).
There are several motivations for investigating the causes of individual extreme events. From a scientific perspective, these studies increase our understanding of how and why the frequency and intensity of extremes has changed over time. These studies may also spur model improvements to ensure that the models used in event attribution studies adequately represent the event being studied. There is an element of scientific curiosity, but the primary motivation for event attribution goes beyond science.
Extreme events are directly traceable to loss of life, rising food and energy prices, increasing costs of disaster relief and insurance, fluctuations in property values, and concerns about national security. Extreme events can and have evoked policy changes: for example, Superstorm Sandy led to supplemental Congressional legislation to increase the National Weather Service’s numerical weather modeling capacity.1 Furthermore, global insurer Munich Re calculated that natural disasters caused more than $90 billion in overall losses and $27 billion in insured losses in 2015 alone (NatCatSERVICE, 2016).2 A study in the Bulletin of the American Meteorological Society (BAMS) (Lazo et al., 2011) noted that weather affects about 3.4% of the U.S. Gross Domestic Product, or more than $500 billion per year.
The strong societal impacts of extreme events explain public and policy maker interest in understanding their underlying causes. In addition, it is important to assess what is known about climate and non-climate causes of such events in order to evaluate whether they are likely to pose increasing risks to life and property in particular regions in the future. As is established in this report and many others, the climate contribution to risk associated with some kinds of extreme events is expected to increase over time as the concentration of GHGs in the atmosphere increases. Some of the anticipated impacts can be reduced, however, through such management strategies as land use planning if the connections between climate change and extreme events like intense precipitation are better understood. Such planning would ideally be based on a broad risk assessment, including projections of future trends in extreme events, and it need not rely specifically on attribution of individual events.
As they improve, event attribution studies can be a tool for informing choices about assessing and managing risk and guiding adaptation strategies. Such information may be critical to multiple decision makers, among them insurers, elected officials and policy makers, local and regional land and resource managers, zoning and infrastructure planners and engineers, litigators, and emergency managers who focus on disaster risk reduction.
In the past decade, the field of extreme event attribution has moved from generalized statements about expecting certain events to increase in frequency or magnitude,
1 See https://www.ametsoc.org/boardpges/cwce/docs/profiles/MurphyJohnD/2013-08-SCM.pdf (accessed May 31, 2016).
2 NatCatSERVICE is a natural catastrophe loss database that analyzes approximately 1,000 events every year.
to documented increases in frequency or intensity of extreme events, to probability-based attribution of individual events. Following an extreme climate or weather event, the standard response from scientists has typically been that global warming does not “cause” any single event in a deterministic sense, but it can make some of them more likely to occur or more intense when they do. Because of advances in the relatively young science of extreme event attribution, however, it is now possible in some cases to provide quantitative information about how climate change may have impacted the probability or intensity of an individual event and to cast this within a probabilistic causal framework.
In perhaps the first attempt at extreme event attribution, Stott and colleagues (2004) showed that climate change had at least doubled the chance of the record-breaking 2003 European summer heat wave that has been associated with the death of more than 70,000 people by some accounts (Robine et al., 2008). Since then, advances in the field have prompted numerous studies (e.g., the 2010 Russian heat wave [Dole et al., 2011; Otto et al., 2012]; the Texas drought and heat wave in 2011 [Hoerling et al., 2013; Rupp et al., 2012]; and the ongoing California drought [Cheng et al., 2016; Diffenbaugh et al., 2015; Williams et al., 2015]). BAMS now publishes annual special issues on event attribution (Herring et al., 2014, 2015b; Peterson et al., 2012, 2013b), which include a compilation of short studies on events that occurred during the previous year. An indication of the developing interest in event attribution is highlighted by the fact that in 4 years (2012-2015), the number of papers increased from 6 to 32.
Detection and Attribution of Long-Term Changes
Many elements of extreme event attribution research are derived from the more mature field of detection and attribution of long-term changes in the characteristics of the climate, such as changes in the frequency or intensity of extremes as well as changes in average climatic conditions.
The primary approach used in detection and attribution research is to compare observations (e.g., of spatial patterns of decadal mean temperatures) to expected changes, which are derived from climate model simulations. The methods used for detection of change continue to evolve, but repeatedly they have been demonstrated to be robust (see, e.g., Allen and Stott, 2003; the appendices in Hegerl et al., 2007, and Bindoff et al., 2013; the review of Hegerl and Zwiers, 2011; and recent papers that suggest further changes to the methods, such as Ribes et al., 2013, 2015; Ribes and Terray, 2013; and Hannart et al., 2015b).
Considerations Specific to Attribution of Extreme Events
Attribution is defined as the process of evaluating the relative contributions of multiple causal factors to a change or event with an assignment of statistical confidence3 (Hegerl et al., 2010). Many causal factors impact any given extreme event, so attribution to any of them could be studied in principle. Our statement of task covers attribution to both human-induced climate change and natural variability. In a number of respects the scientific issues are similar, whether human influence or natural variability is being assessed in an attribution study, so much of our discussion is general. Where attribution to human influence raises distinct scientific issues, our discussion prioritizes those, and our conclusions and recommendations do as well.
The occurrence of any individual extreme event, by itself, does not prove or disprove that the climate is changing. Nevertheless, event attribution studies seek to calculate how much human-induced climate change has affected an individual event’s magnitude or probability of occurrence (Stott et al., 2015).
Conclusions regarding attribution of extreme events are strongly affected by the way “extreme” is defined by scientists. Seneviratne and colleagues (2012) define climate extremes (extreme weather or climate events) as “the occurrence of a value of a weather or climate variable above (or below) a threshold value near the upper (or lower) ends of the range of observed values of the variable.” In fact, the threshold that is selected as “extreme” is generally based on 20th-century observations, but the baseline of what is “normal” is changing over time. In the future, events that are currently considered extreme may eventually be considered normal. Therefore, scientists generally establish metrics to characterize the extreme nature of the event being attributed in the context of a baseline period.
There are several important challenges related to event attribution (discussed in more detail in other chapters), including defining and interpreting an “event” and characterizing a “cause,” or a causal link. Further issues arise from the many different ways that scientists (who are often working with different sources of data and models) describe the degree of certainty about their findings and characterize the uncertainty.
This committee was asked to examine the science of attribution of extreme weather events to human-caused climate change and natural variability by reviewing current
3 In practice, not all attribution studies include statistical confidence.
understanding and capabilities; assessing the robustness of the methods; providing guidance for interpreting analyses; and identifying priority research needs (see Appendix A for the full statement of task).
This study is sponsored by the David and Lucile Packard Foundation, the Heising-Simons Foundation, the Litterman Family Foundation, the National Aeronautics and Space Administration (NASA), the National Oceanic and Atmospheric Administration (NOAA), and the U.S. Department of Energy, with additional support from the National Academy of Sciences’ Arthur L. Day Fund. Beyond the sponsors, the intended audiences for this report are the scientific community, decision makers, and the media. The committee’s goal for this report is to provide these audiences with guidance in interpreting new attribution studies and about the robustness of extreme event attribution science. The committee also hopes that this report guides future support and development of high-priority attribution and detection projects.
Although it is clear to the committee that communication is a critical issue in extreme event attribution, communication is not discussed in detail in this report, as it is not part of the committee’s charge. Indeed, a careful and comprehensive treatment of the many issues associated with science communication related to climate attribution could be a study in its own right. The committee recognizes the importance of communicating clearly and accurately framing any climate-related issue, however. Framing of event attribution questions—how they are posed, and the context within which they are posed—is a key issue, both in terms of communicating study results and in designing and conducting event attribution studies (e.g., Otto et al., 2013, 2015b; Trenberth et al., 2015). Different event framing can lead to large differences in the interpretation of evidence regarding whether human influence on the climate system played a role. The committee has included a detailed discussion on the framing of extreme event attribution questions in Chapter 2 and offers guidance on communicating event attribution study results in Chapter 3.
Although this report focuses almost exclusively on the physical aspects of the causes of extreme events, including the effect of anthropogenic climate change, it is important to acknowledge that significant human aspects (other than human-induced GHG emissions) do influence the severity of extreme events. This includes the perception of what is regarded as being “extreme” and the role that human activity plays in creating the vulnerability and exposure that determine the impacts of extreme events (Cardona et al., 2012). Event attribution is important because of its relationships to risk perception, disaster risk, climate change adaptation, disaster risk reduction, communication, and decision making. Human behavior can either exacerbate or mitigate the impacts of extreme events. For these reasons, understanding the social, ethical, and
human behavior issues that are connected to the experience of extreme events is an important research need.
This report focuses on several topics related to the committee’s statement of task. Chapter 2 discusses the framing of event attribution questions. Chapter 3 discusses the challenges and uncertainties associated with the implementation of the different approaches to extreme event attribution. In Chapter 4 the committee provides an evaluation of the robustness of the attribution work that has been completed for specific types of extreme events as well as identifies anticipated progress in research efforts. Chapter 5 provides guidance for future research.