Climate Science and Law For Judges

Part One: Scientific Foundations of Climate Change

Drawing the Causal Chain: The Detection and Attribution of Climate Change

Author
Lawrence Berkeley National Laboratory
Published
Collage showing traffic, heating atmosphere, and storm damage

This module describes the detection of human-induced climate and its attribution to causal factors. This rigorous body of scientific literature has provided the evidence that human activities, principally the burning of coal, oil, and natural gas for energy, have changed climate. This module will discuss two broad aspects of detection and attribution science. The first part describes the human influence on long-term trends in the climate system. The second part describes the human influence on specific extreme weather events and their impacts.

Abstract

The central issue in both climate science and the law is the attribution of effects to causes. In climate science, this is a two-step process. The first step is to detect that the climate has changed by demonstrating an observable change in a particular climate measure. The second step is to attribute that change to causal factors. Commonly known as D&A, the detection and attribution of climate change constitute an exercise in causality.

Quantifying the influence of the various human changes to the climate system is potentially important to assessing responsibility of the major polluters. Of particular relevance is the development of extreme weather event attribution. It is now possible to quantify the effect of global warming on a wide variety of actual specific individual weather events. The most recent research extends this quantification to the impacts of those weather events. Thus, it is possible to estimate the fractional cost of an extreme weather event due to human-induced climate change whether that be in dollars or lives lost.

I. Introduction

Complex phenomena such as climate change have many potential causal influences. Of principal concern today is the increase in atmospheric carbon dioxide (CO2) resulting primarily from the burning of fossil fuels for energy. While this powerful greenhouse gas makes up a small fraction of the atmosphere, its concentration has increased substantially from about 280pmm (parts per million) prior to the Industrial Revolution to over 400ppm. In fact, this is the highest level atmospheric CO2 in the last 800,000 years, well before the evolution of modern humans.1 This increased concentration has demonstrably caused an unprecedented increase in global temperatures and by other climatic changes. The current global average surface air temperature is the warmest since at least the last interglacial period, 125,000 years ago.2 D&A analyses attempt to determine whether changes in the composition of the atmosphere are linked to observed changes in the climate system.

CO2 is not the only atmospheric pollutant with the potential to alter the climate. Methane (CH4) from both natural and anthropogenic sources also acts to trap heat in the atmosphere, and its concentration in the atmosphere also has been increasing due to human activities. Various combinations of nitrogen and oxygen (known as nitrous oxides, or NOx), as well as the chlorofluorocarbons and bromocarbons now banned by the Montreal Protocol, are also greenhouse gases with the similar heat-trapping properties. Some D&A studies attempt to separately quantify the individual warming effect of these various pollutants, but most studies aggregate all greenhouse gases as a “CO2 equivalent,” or the amount of carbon dioxide that would be needed to produce the warming of all greenhouse gases combined.

Aerosols are another important atmospheric pollutant. Not to be confused with hair spray, aerosols are small atmospheric particles or liquid droplets, either natural or man-made. Some of these aerosols, such as sulfate caused by burning high-sulfur coal and oil or by large volcanic eruptions, reflect sunlight back to outer space and can have a cooling effect that counteracts the effect of increased greenhouse gases.3 Other aerosols, such as the soot or “black carbon” caused by forest fires or the burning of wood or dung for energy, can have a warming effect, thus exacerbating the effects of increased greenhouse gases.4 Dust blown off the deserts can be transported long distances and also can have complex interactions with aspects of the climate system.5

In addition to changing the composition of the atmosphere, humans have changed the surface of the earth for tens of thousands of years if not longer. Deforestation and subsequent reforestation change the amount of light reflected from the earth’s surface back into space, which in turn affects temperature. Forests tend to be darker than farmland and reflect less sunlight back to outer space, warming the earth’s surface, while snow-covered land is white and reflects more sunlight back to space than do areas covered with vegetation. Urbanization also affects the planet’s reflectivity, also known as albedo. For example, asphalt and dark roofs absorb more solar energy than do concrete or light-colored roofs. While the effects of urbanization are usually localized, D&A analyses have been used to quantify their consequences for climate change.

Variations in the intensity of sunlight received at the top of the earth’s atmosphere can also cause the climate to change. Long-term variations in the earth’s orbit are known to have caused massive swings in climate over long time periods, ranging from very cold ice ages to conditions warmer than today’s. However, these orbital changes and their associated climate effects occur on timescales of 1000s of years, thus very slowly compared with the global warming that has occurred in recent decades and are not generally part of D&A analyses.

Of more relevance on human timescales is the variability in the Sun’s luminosity. With a period of approximately 13 years, these solar variations and their impact on global temperatures have been well studied and will be discussed later in this module.

II. How Are D&A Analyses Done?

The causal factors described above are often referred to as external “forcing” factors. While these factors can be of both natural and anthropogenic origin, they are described as external because they are imposed upon the climate system rather than being an intrinsic part of it. Changes in climate due to these causal factors are the effects or “signals” being sought in D&A analyses.

However, the climate system also has a complicated internal variability. Some of these modes of internal variability are well known. For example, El Niño is part of a periodic redistribution of heat in the Pacific Ocean that occurs every few years. This natural variation in Pacific Ocean temperatures has far-reaching effects, such as modulating winter temperatures in North Dakota and influencing the number of North Atlantic hurricanes.

Other quasi-regular natural oscillations are not so well known to the public. For example, both the Atlantic and Pacific Oceans undergo regular changes over periods of years to decades that can influence temperature and rainfall patterns on land. While some aspects of these natural changes within the climate system are not thoroughly understood, enough is known about their mechanisms and effects to rule out their being responsible for the warming and associated climatic changes observed in recent decades.

Climatic measures such as average global temperature also vary from year to year due to weather “noise” or apparently random variations within the climate system. These variations are much more difficult to predict because they are the result of initially small influences that are magnified by the mechanisms of the climate system. The slower-moving components of the climate system aggregate short-term weather variations to longer-term fluctuations, so there is no intrinsic upper limit for the time duration of climate variability. The total internal variability of the climate system is therefore a mixture of known natural oscillations and this unpredictable chaotic noise.

The challenge in a D&A analysis is to extract the external signal of human-produced forcing factors from the natural variation of the climate system. This sort of problem arises in other areas of science and technology, such as in certain electrical engineering applications, and climate scientists have adapted techniques from that discipline.

However, unlike electrical engineers or other physical scientists and as was noted in the module on How Climate Science Works, climate scientists have only a single experimental planet to study. Lacking alternate planets to test a hypothesis, they must rely on climate models to determine how external forcing factors are changing the climate. Climate models are computer programs that simulate the physical processes that make up the climate. They vary from simple models of single components like the atmosphere or ocean to very large and complicated combinations of components including but not limited to the atmosphere, ocean, sea ice, glacial ice masses, land surfaces, biogeochemistry, and atmospheric chemistry, as shown in the What Is Causing Climate Change? module. The basic methodologies involved in using climate models are similar to those used in many other areas of science.

As an example, consider the most well-established aspect of the climate system, the global average surface temperature. The first step of a D&A analysis is to detect a change in the observed record, usually expressed as a trend. Fortunately, extensive observations of air temperatures over the land and in the ocean surface go back well into the 19th century, and indirect data can push this timeline further back. The black line in Figure 1 shows these measurements averaged over the entire globe each year from 1850 to 2020. These temperatures are shown as a difference from the average over the 1850-1900 period, which is centered around zero. The internal variability of climate is evident by the short-term ups and down in the black line. Around 1930, the observed global average surface temperature begins to increase above the previous average. By the 1980s, a detectable trend or change is obvious.

  • 1Sergey K. Gulev et al., Changing State of the Climate System, Chapter 2, in IPCC, Climate Change 2021: The Physical Science Basis (2021), https://www.ipcc.ch/report/ar6/wg1/.
  • 2Id.
  • 3Nicholas Bellouin et al., Bounding Global Aerosol Radiative Forcing of Climate Change, 58 Reviews of Geophysics 1 (2019), https://doi.org/10.1029/2019RG000660.
  • 4Tami C. Bond et al., Bounding the Role of Black Carbon in the Climate System: A Scientific Assessment, 118 JGR Atmospheres 5380 (2013), https://doi.org/10.1002/jgrd.50171.
  • 5Kevin A. Reed et al., Exploring the Impact of Dust on North Atlantic Hurricanes in a High-Resolution Climate Model, 46 Geophysical Rsch. Letters 1105 (2019).
Graph of the observed global mean surface air temperature (black line) with climate models containing human and natural influences (brown line) and not with models that include only natural influences (green line). Confidence intervals of the model simulations are shown by the shaded regions. Units: oC.

Figure 1. The observed global mean surface air temperature (black line) tracks with climate models containing human and natural influences (brown line) and not with models that include only natural influences (green line). Confidence intervals of the model simulations are shown by the shaded regions. Units: oC. Source: IPCC, Climate Change 2021: The Physical Science Basis, Summary for Policymakers (2021) (Figure SPM.1(b)).

If a trend has been detected, the next D&A step is the attribution of the observed change to a causal factor. To do that, D&A analyses typically compare the observations both with models that include a particular set of causal factors and with models that do not include them. The variations in simulations that do not include the causal factors reveal how much internal variability is present in the system, and this variability can be compared with the variability in the observations. The simulations that do include the causal factors then can be compared with an observation to determine whether an observed change can be attributed to that factor.

Correspondence between a simulation and an observed change does not necessarily mean that the change can be attributed to the causal factor included in the model. A model may not be “fit for purpose,” meaning that it does not accurately simulate the system being modeled, or cancelling errors or wrong combinations of external influences could spuriously agree with the observed record. Determining whether a model is fit for purpose is done through a process called model evaluation, which is a well-established science that has been discussed extensively in many reports and papers.1 Model evaluation involves such steps as comparing model outputs, contrasting simpler with more complex models, combining models, and quantifying uncertainties. This process increases the confidence with which attributions and projections based on models can be made.

Figure 1 demonstrates many aspects of this D&A process. The brown line represents the global mean surface temperature from climate model simulations with five external forcing agents: greenhouse gases, anthropogenic aerosols, anthropogenic ozone, volcanic aerosols, and solar variability. The curve is smoother than the observations (black line) because an ensemble of different climate models yields results that have been averaged together, which reduces internal variability. The range of different model results is shown by the brown-shaded region. Agreement of the averaged model simulations with the observed change leads to the conclusion that the detected observed change is externally forced and not an internal variation.

Figure 1 also compares the “all forcings” simulations with simulations that include only the volcanic and solar variability forcings, as shown in green. These simulations clearly do not contain the observed change. Furthermore, when comparing the range of natural simulations (shaded green) to the range of “all forcings” simulations (shaded brown), it is clear that the “all forcings” signal emerged from the noise of natural variability in the 1990s. Applying formal statistical tools to these data sets can quantify these statements in a rigorous manner.2

The conclusions of a D&A study are often made in attribution statements, which are constructed not to overstate the link between a cause and the observed effect. From Figure 1, such a conservative statement would be “It is very likely that at least half of the observed warming is due to human influences.” The italicized “very likely” is a reference to the Intergovernmental Panel on Climate Change (IPCC)-calibrated language denoting a 95% statistical confidence interval.3 The “at least” part of the statement refers to the lower bound of the brown shaded region, which is about half of the observed warming (in black). The IPCC statement of confidence is an expert judgment based on multiple lines of evidence, including observations, climate models, and statistical analyses.

This very conservative language belies the actual level of confidence in the attribution of global warming to greenhouse gases. An equally correct statement is “Our best estimate is that all of the observed warming is due to human influences.”

III. D&A Analyses Beyond Temperature

D&A analyses consider many aspects of the climate system other than global mean temperature, including precipitation, ocean temperature, sea ice extent, and sea level. Figure 2 shows aspects of climate that have been subjected to D&A analyses.

  • 1Zeke Hausfather et al., Evaluating the Performance of Past Climate Model Projections, 47 Geophysical Rsch. Letters 1 (2020), https://doi.org/10.1029/2019GL085378.
  • 2See, e.g., Gabriele C. Hegerl & Gerald R. North, Comparison of Statistically Optimal Approaches to Detecting Anthropogenic Climate Change, 10 J. Climate 1125 (1997), https://doi.org/10.1175/1520-0442(1997)010<1125:COSOAT>2.0.CO;2.
  • 3IPCC, Climate Change 2021: The Physical Science Basis, Summary for Policymakers (2021), https://www.ipcc.ch/report/ar6/wg1/chapter/summary-for-policymakers/.
Figure of some natural factors that have affected the climate in the past, and continue to do so today.

Figure 2. Studies in the peer-reviewed science literature have attributed changes in many aspects of climate to human emissions of heat-trapping gases and aerosols. Many natural factors have affected climate in the past and continue to do so today, but human activities are the dominant contributor to recently observed climate changes. This figure is only a partial depiction of a growing attribution literature.

  • 1-->John Walsh et al., Appendix 3: Climate Science Supplement, in Climate Change Impacts in the United States: The Third National Climate Assessment (2014), https://nca2014.globalchange.gov/report/appendices/climate-science-supplement.

The first challenge in a D&A study is acquiring long-term observational records, which can be inadequate even for temperature, much less other quantities of interest. Observational coverage is incomplete over the globe and varies with time. High-quality in situ observations of temperature and precipitation exist in the United States, western Europe, China, India, and coastal Australia. However, much of the Global South, especially Africa and South America lack quality observations. Many parts of the southern hemisphere are poorly observed, as are the portions of the oceans outside of shipping lanes. Satellites provide uniform global coverage, but the earliest satellites with relevant instrumentation were launched just in 1979. Furthermore, these early satellites’ primary mission was weather prediction, not climate monitoring.

Satellites nevertheless offer a good example of opportunities for D&A analyses. For example, one use of satellites is to compare temperatures close to the surface (in the troposphere) with those in the upper atmosphere. Microwave sounding unit satellites do not measure air temperature at different levels above the ground directly, but temperatures at different levels can be inferred from a retrieval and calibration algorithm.1 The human influence on these temperatures is very clear in D&A analyses. Figure 3 shows, on the left, the vertical profile of air temperature aloft over the 1979-1999 period. The troposphere, from the ground up to about 200 millibars (mb) of pressure, has clearly been warming, while the stratosphere, above about 100 mb of pressure, has been cooling both in the observations and in three models.2 More importantly, a series of model simulations with external forcing agents individually imposed (Figure 3, right) reveals that only greenhouse gases and ozone (panels A and C) can produce a cooling of the stratosphere. These studies also demonstrate that solar variations are not responsible for the observed climate change, because they would be expected to warm the stratosphere rather than cool it as observed (Figure 3, right panel E). There are many studies demonstrating the human cause of global warming. This example was chosen as it demonstrates a sophisticated understanding of not only the magnitude of the global warming, but of its distinct spatial structure that cannot be explained without a human influence.

  • 1Benjamin D. Santer et al., Identifying Human Influence on Atmospheric Temperature, 110 PNAS 26 (2013), https://www.pnas.org/doi/full/10.1073/pnas.1210514109; Benjamin D. Santer et al., Influence of Satellite Data Uncertainties on the Detection of Externally Forced Climate Change, 300 Science 1280 (2003), https://doi.org/10.1126/science.1082393.
  • 2Benjamin D. Santer et al., How Well Can the Observed Vertical Temperature Changes Be Reconciled With Our Understanding of the Causes of These Changes?, in Temperature Trends in the Lower Atmosphere (T.R. Karl et al., eds.) (2006).
Graphs of temperature change profiles (left) and simulations of external forcing factors (right)

Figure 3. Left: Over the 1979-99 period, air temperature averaged over both latitude and longitude has increased in the lower atmosphere and has declined in the upper atmosphere. Observations are the solid and dashed black lines. Results from climate models are shown as colored lines. Right: Simulations of external forcing factors produce different predictions of changes in vertical air temperature. Note that changes in solar luminosity alone (lower left panel) do not reproduce the observed vertical changes but that the “all forcings” simulation (lower right panel), which includes human greenhouse gas increases, do. Source: Benjamin D. Santer et al., How Well Can the Observed Vertical Temperature Changes Be Reconciled With Our Understanding of the Causes of These Changes?, in Temperature Trends in the Lower Atmosphere (T.R. Karl et al., eds.) (2006).

Because of the optical properties of water vapor, it can be remotely observed very accurately over the oceans. Although satellite observations of water vapor started only in 1989, the detected signal quickly rose above the noise and could be attributed readily to external forcing factors.1 These studies demonstrated the validity of D&A analyses, as the measured moisture changes were shown to be consistent with observed temperatures and could be predicted from well-established physical laws.

Changes in average precipitation have also been subjects of D&A analyses. As the atmosphere warms, when fully saturated, it can hold more water vapor, and precipitation might be expected to increase. However, the D&A problem for precipitation is complicated as changes in atmospheric circulation can cause precipitation to increase or decrease spatially depending on location, and season and natural variability is high.2 Hence, confidence in attribution of precipitation changes is lower than it is for temperature.3 In addition, frequent and accurate precipitation observations are mostly limited to North America and Europe, which imposes some conditions on the published attribution statements.4 As discussed later in this module, potential changes in extreme temperature and precipitation due to global warming are expected to be more robust, and D&A studies are more confident in these areas.5

Climate models require many long computations, and such simulations require substantial human and machine resources. Fortunately, climate science has matured to the point where a great deal of simulation data from the international climate modeling community is now publicly available.6 Collections of simulations are one way to evaluate whether models are fit for purpose. Other times, more specialized analyses may be required to make this determination.

IV. Assessing Confidence in Attribution Statements

Assessing confidence in attribution statements is critically important for decision- and policymakers. Most attribution statements are framed in the calibrated language developed by the IPCC of an objective “likelihood” and a subjective “confidence” (Table 1).7 As noted earlier, the IPCC’s phrasing tends to be conservative because of the focus on the lower bounds of statistical confidence intervals.

  • 1Carl A. Mears et al., Relationship Between Temperature and Precipitable Water Changes Over Tropical Oceans, 34 Geophysical Rsch. Letters 1 (2007), https://doi.org/10.1029/2007GL031936; Benjamin D. Santer et al., Identification of Human-Induced Changes in Atmospheric Moisture Content, 104 PNAS 15248 (2007), https://www.pnas.org/doi/full/10.1073/pnas.0702872104.
  • 2David R. Easterling et al., Precipitation Change in the United States, Chapter 7, in U.S. Global Change Res. Prog., Climate Science Special Report: Fourth National Climate Assessment, Vol. 1 (2017).
  • 3IPCC, Climate Change 2021: The Physical Science Basis, Summary for Policymakers (2021), https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_SPM.pdf.
  • 4Seung-Ki Min et al., Human Contribution to More-Intense Precipitation Extremes, 470 Nature 378 (2011), https://www.nature.com/articles/nature09763; Xuebin Zhang et al., Attributing Intensification of Precipitation Extremes to Human Influence, 40 Geophysical Rsch. Letters 5252 (2013), https://doi.org/10.1002/grl.51010.
  • 5Yeon-Hee Kim et al., Attribution of Extreme Temperature Changes During 1951-2010, 46 Climate Dynamics 1769 (2016), https://doi.org/10.1007/s00382-015-2674-2; Seung-Ki Min et al., Multimodal Detection and Attribution f Extreme Temperature Changes, 26 J. Climate 7430 (2013), https://doi.org/10.1175/JCLI-D-12-00551.1.
  • 6See, e.g., CMIP5 Monthly Data on Single Levels, Copernicus Climate Change Service, https://cds.climate.copernicus.eu/cdsapp#!/dataset/projections-cmip5-monthly-single-levels?tab=overview (last visited Jan. 26, 2023).
  • 7Sophie C. Lewis et al., Toward Calibrated Language for Effectively Communicating the Results of Extreme Event Attribution Studies, 7 Earth’s Future 1020 (2019), https://doi.org/10.1029/2019EF001273; Michael D. Mastrandrea et al., Intergovernmental Panel on Climate Change, Guidance Note for Lead Authors of the IPCC Fifth Assessment Report on Consistent Treatment of Uncertainties 3 (2010).
The calibrated uncertainty language developed by the IPCC treats both the nature of the evidence (left) and the likelihood of outcomes (right).

Table 1. The calibrated uncertainty language developed by the IPCC treats both the nature of the evidence (left) and the likelihood of outcomes (right). Left: Confidence increases toward the top-right corner as suggested by the increasing strength of shading. Generally, evidence is treated with greater confidence when multiple independent lines of high-quality evidence are consistent. Right: Objective likelihood statements range from virtually certain to exceptionally unlikely. Source: Michael D. Mastrandrea et al., Intergovernmental Panel on Climate Change, Guidance Note for Lead Authors of the IPCC Fifth Assessment Report on Consistent Treatment of Uncertainties 3 (2010).

The IPCC has also developed a method for assigning confidence to attribution statements regarding long-term changes in climate, as shown in Figure 4.1 This method can be used to assess contrasting studies when developing a weighted likelihood of particular events.

  • 1Friederike E.L. Otto et al., Toward an Inventory of the Impacts of Human-Induced Climate Change, 101 Bull. Am. Meteorological Soc’y E1972 (2020), https://doi.org/10.1175/BAMS-D-20-0027.1; Sonia I. Seneviratne et al., Weather and Climate Extreme Events in a Changing Climate, Chapter 11, in IPCC, Climate Change 2021: The Physical Science Basis (2021).
Series of questions to assess confidence in attribution statements

Figure 4. Confidence (low, medium, or high) in attribution statements can be assessed by answering a series of yes-no questions. Source: Friederike E.L. Otto et al., Toward an Inventory of the Impacts of Human-Induced Climate Change, 101 Bull. Am. Meteorological Soc’y E1972, E1975 (2020).

Confidence in an attribution statement is highest when multiple, independent teams arrive at similar conclusions using different observational data sets, different climate models, and different attribution techniques. This has occurred in only a few cases, in part because the attribution community remains small and much D&A work remains to be done. It is important to state that any confidence hinges on a deep understanding of the physical processes behind the changes in question.

Early attribution statements often relied on single climate models, and some rather specialized attribution efforts still do. But confidence increases with the number of climate models used, and the widespread and centralized availability of climate model output data increasingly makes the use of multiple models possible. Using multiple observational data sets also increases confidence, though many of these data sets are not independent, limiting the increase in confidence. Like many areas of climate science, attribution studies may draw upon high-quality weather model simulations informed by real-world observations. These simulations, known as “reanalyses,” fill in the gaps in space and time where no observations exist. However, confidence in attribution statements based on reanalyses depends on the uncertainties inherent to the reanalysis procedures.

V. Attributing Extreme Events to Climate Change

Traditional attribution statements have focused on long-term observed changes in climate. In contrast, extreme event attribution statements generally focus on the influence of human activities on a single event (or sometimes a single class of events).

In 2003, after his house in Oxford, England, was flooded by an exceptionally rainy storm, climate scientist Myles Allen proposed that the human influence on severe weather events could be quantified.1 Prior to that, climate scientists tended not to say much about the human influence on individual extreme weather events. Instead, they would say something like “While no individual event can be tied to climate change, what happened is consistent with expectations.”

Today, some kinds of individual events can be linked to climate change. Extreme event attribution techniques now make it possible under certain circumstances to formulate quantitative statements, with confidence intervals, about the human influence on many kinds of individual extreme weather and climate events.

For example, in 2003, shortly after Allen’s proposal, central Europe experienced a disastrous heatwave that caused over 70,000 excess deaths. Using the high-quality observational record of European temperatures and a single climate model, scientists estimated that climate change likely at least doubled the chances of the measured maximum daytime temperatures. Since then, the field of extreme event attribution has expanded to include many types of extreme weather in addition to heatwaves2 —including heavy precipitation, floods, droughts, and some extreme storms such as hurricanes.3 Figure 5 from a 2016 report of the U.S. National Academies of Sciences, Engineering, and Medicine illustrates the variety of events that are of interest to attribution scientists. Since that report, much progress has been made as discussed below. In particular, tropical cyclones and extreme precipitation would be moved significantly up and toward the right. Also, progress has been made in the attribution of certain types of drought and wildfire.

  • 1Myles Allen, Liability for Climate Change, 421 Nature 891 (2003), https://doi.org/10.1038/421891a.
  • 2Nat’l Acad. of Sci., Eng’g, and Med., Attribution of Extreme Weather Events in the Context of Climate Change (2016), https://doi.org/10.17226/21852.
  • 3Stephanie C. Herring et al., Explaining Extreme Events of 2020 From a Climate Perspective, 103 Bull. Am. Meteorological Soc’y S1 (2022); Stephanie C. Herring et al., Explaining Extreme Events of 2017 From a Climate Perspective, 100 Bull. Am. Meteorological Soc’y S1 (2019); Stephanie C. Herring et al., Explaining Extreme Events of 2016 From a Climate Perspective, 99 Bull. Am. Meteorological Soc’y S1 (2018); Stephanie C. Herring et al., Explaining Extreme Events of 2015 From a Climate Perspective, 97 Bull. Am. Meteorological Soc’y S1 (2016); Stephanie C. Herring et al., Explaining Extreme Events of 2014 From a Climate Perspective, 96 Bull. Am. Meteorological Soc’y S1 (2015); Stephanie C. Herring et al., Explaining Extreme Events of 2013 From a Climate Perspective, 95 Bull. Am. Meteorological Soc’y S1 (2014); Thomas C. Peterson et al., Explaining Extreme Events of 2012 From a Climate Perspective, 94 Bull. Am. Meteorological Soc’y S1 (2013); Thomas C. Peterson et al., Explaining Extreme Events of 2011 From a Climate Perspective, 93 Bull. Am. Meteorological Soc’y S1 (2012).
Graph of relative confidence in attribution of different extreme events.

Figure 5. A 2016 assessment of the relative confidence in attribution of different extreme weather events. Note that significant progress has been made since this assessment. Source: Nat’l Acad. of Sci., Eng’g, and Med., Attribution of Extreme Weather Events in the Context of Climate Change (2016), https://doi.org/10.17226/21852 (Fig. 4.7).

The ability to quantify the human influence on a particular event such as a flood or heat wave is not only important for the communication of climate change, but also of direct relevance for judges. Extreme event attribution statements are of two equivalent types. The first is “Did global warming change the magnitude of this event given its estimated rarity?” The second is “Did global warming change the chances of an event of this magnitude?” These two questions are not independent, as illustrated by Figure 6.

In Figure 6, the likelihood of a given temperature in Washington, D.C., as calculated by models is plotted as a function of its return time. Return time—the period in which we might expect an event to recur on average—is a key term for stating the likelihood of an event. The black line averages the model simulations under present-day conditions of global warming. The red line averages the simulations under pre-industrial climate conditions. The intersection of the vertical line with the black line indicates that, if the current climate were unchanging, temperatures would reach 41oC about once every 20 years on average over a long period of time. But the climate is changing, so a better way of describing current conditions is to say that there is a 1-in-20 or 5% chance of reaching 41oC this year. In a pre-industrial climate, the 20-year event would have been at about 39oC, as indicated by the intersection of the vertical line with the red curve. Therefore, climate change caused the 20-year event to be about 2oC warmer.

The second question regarding likelihood is more nuanced. The horizontal dashed line drawn at 39oC intersects the black curve at a return time of about 2.5 years under present-day climate conditions. It intersects the red line at a return time of 20 years under pre-industrial conditions. Hence, the chances of reaching 39oC have been increased by climate change by a factor of 20/2.5 or eight times.

Graph of a 20-year event of surface air temperature near Washington, D.C., is about 2oC (3.6oF) higher in a realistic climate (black line) than in a cooler counterfactual climate without anthropogenic climate change (red line) based on climate model experiments.

Figure 6. A 20-year event of surface air temperature near Washington, D.C., is about 2oC (3.6oF) higher in a realistic climate (black line) than in a cooler counterfactual climate without anthropogenic climate change (red line) based on climate model experiments. Source: Michael Wehner et al., Early 21st Century Anthropogenic Changes in Extremely Hot Days as Simulated by the C20C+ Detection and Attribution Multi-Model Ensemble, 20 Weather & Climate Extremes 1, 6 (2018) (Fig. 6).

Hence, the questions about human-caused changes in probability and magnitude of individual extreme weather events are two sides of the same coin. However, changes in magnitude are often more easily interpreted when considering changes in the impacts of extreme events as described below.

Confidence in individual extreme event attribution statements is increased if D&A studies have produced more general statements about the relevant variables or regions. However, this is not strictly necessary, and extreme event attribution statements can be made even if trends in similar events have not been detected.1

Extreme heat. Figure 6 also demonstrates some of the issues associated with attribution statements involving extreme heat. The high-temperature curves without climate change (red) and with climate change (black) approach values of 39oC and 41oC, respectively, and appear never to go higher. Was a temperature of 106oF impossible without climate change, as this curve would suggest? Most attribution statements would likely not make such a strong claim from this model calculation but would say that the likelihood of the temperature rising that high is very low but not absolutely zero because of statistical uncertainty. Quantifying the uncertainty in this upper bound is an ongoing topic in statistical research.

Figure 6 shows that the temperature change attributable to human activities in the once in 50- or 100-year heatwave is not very different than the once in 20-year event. This consequence of the distribution of extreme heat events in the atmosphere over time, in which high-temperature events of any kind are extremely rare, permits confidence in attribution statements about heatwaves in advance of their occurrence. Figure 7 applies this analysis to the continental United States, which shows one model’s estimate of the change in 20-year temperatures attributable to climate change. This change is nearly identical in pattern and magnitude to the model’s changes in 50-year temperatures.

  • 1Thomas Knutson et al., Detection and Attribution Methodologies Overview, Appendix C, in Climate Science Special Report: Fourth National Climate Assessment, Volume I (2017).
Map of attributable human temperature increase in rare heatwaves in the United States.

Figure 7. High temperatures during rare heatwaves in much of the United States are estimated to be 3.5oF to 4.5oF higher due to changes humans have made in the composition of the atmosphere. Adapted from Michael Wehner et al., Early 21st Century Anthropogenic Changes in Extremely Hot Days as Simulated by the C20C+ Detection and Attribution Multi-Model Ensemble, 20 Weather & Climate Extremes 1 (2018), https://doi.org/10.1016/j.wace.2018.03.001.

A confident attribution statement is therefore that almost any heatwave that occurs now in the United States is about 3.5oF to 4.5oF warmer than it would have been without climate change. This attribution statement can be made without estimating the probability of the heatwave temperature as long as it is thought to be rare. It would even extend to record temperatures, as long as the existing records are not broken by a large amount. However, in the case of far outliers, such as the 2021 Pacific Northwest heatwave, certain assumptions of this theory are violated and only less-definitive statements can be made.

Precipitation. Well-established physical laws indicate that the capacity of the atmosphere to hold water vapor increases by about 7% per degree Celsius of warming. However, recent attribution simulations suggest that this rate is a lower bound for precipitation increases in certain types of extreme storms and that the actual rate can exceed this lower bound by factors of two or three.1 Estimating the human influence on heavy precipitation events is more complicated than for heatwaves. Precipitation is a sporadic event, and extreme precipitation even more so. Thus, attribution statements for certain types of extreme precipitation events are weaker than for heatwaves.

Hurricanes / Storms. The computational demands of models with resolutions fine enough to capture the processes and conditions of severe storms, including hurricanes, restrict the duration of simulations using current supercomputers. However, shorter but more precise simulations using weather prediction models have proven to be useful in deriving more precise attribution statements. These so-called storyline attribution statements can answer the first question above about the human-induced change in magnitude of an event but cannot inform about the human-induced change in its probability. For instance, using a version of the Weather Research and Forecasting model, the author and a colleague,2 analyzed 15 different large tropical cyclones (that is, hurricanes) and were able to make robust predictions of precipitation increases. Other simulations of dozens of individual tropical cyclones suggest best estimates of anthropogenic increases in precipitation that are twice the typical rate of 7% for the most intense storms.3 In general, the human influence on hurricanes remains a topic of active research and public interest.

Other types of storms have received less attention from the attribution community. Recent research on atmospheric river storms, which carry intense plumes of moisture from the oceans onto land, impacting the San Francisco Bay Area has found that precipitation also can increase at about twice the 7% rate,4 though the physical mechanisms of change are very different than for tropical cyclones.

Little is known about the increases in extreme storms outside the tropics such as occur in the winter, or about the intense summer mid-scale systems that can occur in continental interiors. Limited studies have analyzed the human influence on the environmental conditions that support tornadoes, and a consensus on the influence of climate change on them has not been reached.5 Changes in large-scale atmospheric circulation patterns could also have implications for storm track locations. However, the character and magnitude of these changes are uncertain and their current influence is small compared to localized dynamic and thermodynamic processes.

Drought. The National Oceanic and Atmospheric Administration (NOAA) categorizes drought as a hierarchy of four related conditions. The first, meteorological drought, is characterized by a deficit of precipitation compared to normal conditions. The second, agricultural (or ecological) drought, is characterized by a deficit of soil moisture compared with normal conditions. The third, hydrological drought, is characterized by a deficit of water runoff compared to normal conditions. The fourth, socioeconomic drought, occurs when demand for water exceeds the supply.

Agricultural drought depends both on the precipitation that falls on the ground and on the loss of moisture from plants and soils into the atmosphere. Evaporation from bare ground depends strongly on air temperature. As climate change increases temperature, evaporation also increases, leading to drier soils. Transpiration from plants depends even more strongly on air temperature. As temperature increases, plants cool themselves by evaporating water from their leaves and stems. In very hot conditions, plants can draw moisture from their root system and release it into the atmosphere until there is very little soil moisture left. Because of these processes, many studies have attributed human-induced increases in agricultural drought conditions to this increased evapotranspiration, or release of moisture from land to the atmosphere.6  

Consensus on the effects of climate change on meteorological drought occurrences has not been reached in regions of the United States. Only in Mediterranean regions do studies demonstrate a consistent human influence on precipitation deficits, and even here confidence is low.7 As climate continues to warm, meteorological drought conditions in Mexico and the Southwest United States are projected to become more common,8 but a robust signal of this process has not yet been detected.

VI. The Impacts of Extreme Events

Attribution statements can also be made that link the extreme weather events influenced by human activities with the socioeconomic impacts of those events. As an example, consider Hurricane Harvey, which inundated much of the greater Houston area in 2017. What made Hurricane Harvey such an impactful event was that the storm stalled atop the Gulf Coast of Texas for about three days, dumping copious amounts of rain on land. Three independent analyses of Hurricane Harvey have quantified the increase in total rainfall that can be attributed to human-induced climate change.9 The average finding of these analyses is that global warming increased the region’s precipitation during Hurricane Harvey by about 19%, with a lower bound of 7 percent and an upper bound of 38 percent.

To evaluate the effect of a 19% increase in precipitation, researchers used a model that had demonstrated its ability to accurately simulate the flood caused by Hurricane Harvey given the available precipitation observations.10 To construct a counterfactual “flood that might have been” without climate change, they decreased the observed precipitation uniformly by the range of the published precipitation attribution statements.11 They found that climate change increased both the extent and depth of the flooding, with the magnitude of the increases depending on the amount of increased precipitation estimated to result from global warming.

Figure 8 shows the actual flood and two of the counterfactual floods in the South Houston and Pasadena neighborhoods, which represent a small subsection of the total region analyzed. The model has a resolution of 30 meters, which is about the size of a suburban house and its yard. The top panel shows the simulated flood using observed precipitation data during Hurricane Harvey and is a close approximation of the flooding that actually occurred. The middle panel shows the counterfactual flood simulation corresponding to the lower bound (a 7% increase) of published precipitation attribution statements. The area flooded is not substantially different between the two simulations, but the flood that actually occurred is about a foot deeper than it would have been if climate change had not produced a 7% increase in total rainfall. The lower panel shows the counterfactual flood corresponding to the upper bound (a 38% increase) of published precipitation attribution statements. In this case, many homes that were flooded would not have been if climate change had not increased total rainfall by 38%. In addition, the actual flood was more than 3 feet deeper than the counterfactual flood because of climate change.

  • 1Christina M. Patricola et al., Future Changes in Extreme Precipitation Over the San Francisco Bay Area: Dependence on Atmospheric River and Extratropical Cyclone Events, 36 Weather & Climate Extremes 1 (2022), https://doi.org/10.1016/j.wace.2022.100440; Kevin A. Reed et al., Attribution of 2020 Hurricane Season Extreme Rainfall to Human-Induced Climate Change, 13 Nature Commc’ns 1 (2022), https://doi.org/10.1038/s41467-022-29379-1.
  • 2Christina M. Patricola et al., Future Changes in Extreme Precipitation Over the San Francisco Bay Area: Dependence on Atmospheric River and Extratropical Cyclone Events, 36 Weather & Climate Extremes 1 (2022), https://doi.org/10.1016/j.wace.2022.100440.
  • 3Kevin A. Reed et al., Anthropogenic Influence on Hurricane Dorian’s Extreme Rainfall, 102 Bull. Am. Meteorological Soc’y S9 (2021), https://doi.org/10.1175/BAMS-D-20-0160.1; Kevin A. Reed et al., Attribution of 2020 Hurricane Season Extreme Rainfall to Human-Induced Climate Change, 13 Nature Commc’ns 1 (2022), https://doi.org/10.1038/s41467-022-29379-1; Kevin A. Reed et al., Forecasted Attribution of the Human Influence on Hurricane Florence, 6 Sci. Advances 1 (2020), https://doi.org/10.1126/sciadv.aaw9253.
  • 4Christina M. Patricola et al., Future Changes in Extreme Precipitation Over the San Francisco Bay Area: Dependence on Atmospheric River and Extratropical Cyclone Events, 36 Weather & Climate Extremes 1 (2022), https://doi.org/10.1016/j.wace.2022.100440.
  • 5Emily Bercos-Hickey et al., Anthropogenic Influences on Tornadic Storms, 34 J. Climate 8989 (2021), https://doi.org/10.1175/JCLI-D-20-0901.1; Noah S. Diffenbaugh, Robust Increases in Severe Thunderstorm Environments in Response to Greenhouse Forcing, 110 PNAS 16361 (2013).
  • 6Sonia I. Seneviratne et al., Weather and Climate Extreme Events in a Changing Climate, Chapter 11, in IPCC, Climate Change 2021: The Physical Science Basis (2021); Michael F. Wehner et al., Droughts, Floods, and Wildfire, Chapter 8, in U.S. Global Change Res. Prog., Climate Science Special Report: Fourth National Climate Assessment, Vol. 1 (2017).
  • 7Sonia I. Seneviratne et al., Weather and Climate Extreme Events in a Changing Climate, Chapter 11, in IPCC, Climate Change 2021: The Physical Science Basis (2021).
  • 8David R. Easterling et al., Precipitation Change in the United States, Chapter 7, in U.S. Global Change Res. Prog., Climate Science Special Report: Fourth National Climate Assessment, Vol. 1 (2017).
  • 9Mark D. Risser & Michael F. Wehner, Attributable Human-Induced Changes in the Likelihood and Magnitude of the Observed Extreme Precipitation During Hurricane Harvey, 44 Geophysical Rsch. Letters 12457 (2017), https://doi.org/10.1002/2017GL075888; Geert Jan van Oldenborgh et al., Attribution of Extreme Rainfall From Hurricane Harvey, August 2017, 12 Env’t Rsch. Letters 1 (2017), https://doi.org/10.1088/1748-9326/aa9ef2; S-Y Simon Wang et al., Quantitative Attribution of Climate Effects on Hurricane Harvey’s Extreme Rainfall in Texas, 13 Env’t Rsch. Letters 1 (2018), https://doi.org/10.1088/1748-9326/aabb85.
  • 10Michael Wehner & Christopher Sampson, Attributable Human-Induced Changes in the Magnitude of Flooding in the Houston, Texas Region During Hurricane Harvey, 166 Climatic Change 1 (2021), https://doi.org/10.1007/s10584-021-03114-z; Oliver E.J. Wing et al., A Flood Inundation Forecast of Hurricane Harvey Using a Continental-Scale 2D Hydrodynamic Model, 4 J. Hydrology X 1 (2019), https://doi.org/10.1016/j.hydroa.2019.100039.
  • 11Michael Wehner & Christopher Sampson, Attributable Human-Induced Changes in the Magnitude of Flooding in the Houston, Texas Region During Hurricane Harvey, 166 Climatic Change 1 (2021), https://doi.org/10.1007/s10584-021-03114-z.
Simulations of the actual flood that occurred in the South Houston and Pasadena neighborhoods can be compared with the floods that would have occurred without climate change.

Figure 8. Simulations of the actual flood that occurred in the South Houston and Pasadena neighborhoods can be compared with the floods that would have occurred without climate change. (a) The flood that was. (b) The flood that would have occurred in the absence of climate change if human activities increased Harvey’s storm total precipitation by 7%. (c) The flood that would have occurred in the absence of climate change if human activities increased Harvey’s precipitation by 38%. Source: Michael Wehner & Christopher Sampson, Attributable Human-Induced Changes in the Magnitude of Flooding in the Houston, Texas Region During Hurricane Harvey, 166 Climatic Change 1 (2021) (Figure 2), https://doi.org/10.1007/s10584-021-03114-z.

Over the greater Houston area, this analysis found that for the best estimate of a 19% human-induced increase in precipitation, the flood area was increased by 14%. The reinsurance companies estimate the insured losses of Hurricane Harvey to be about $90 billion. Assuming that damages were mostly from the flood, and that properties were equally valued and distributed uniformly throughout the region, yields a crude estimate of $13 billion for the insured loss due to climate change. The 19% precipitation attribution statement also corresponds to a fourfold human-induced increase in the probability of the actual flood. Thus, as a best estimate, the probability of an insured $90-billion hurricane loss in Texas was quadrupled due to climate change.

The very high resolution of the model and maps permits individuals to know if climate change flooded their own house. More generally, these maps permit much more detailed overall damage estimates. Projecting real estate value maps onto the flood maps reveals that, as a best estimate, 32% of flooded homes in Harris County would not have been flooded without climate change. Furthermore, regardless of climate change, 75% of the flooded homes were outside the federal 100-year floodplain and were thus uninsured, adding to the insured loss.1 Figure 9 shows the upper bound on the distribution of homes that were flooded in Harris County due to climate change.

  • 1Kevin T. Smiley et al., Social Inequalities in Climate Change-Attributed Impacts of Hurricane Harvey, 13 Nature Commc’ns 1 (2021), https://doi.org/10.1038/s41467-022-31056-2.
Map of Harris County, Texas during Hurricane Harvey, where each hexagonal bin symbolizes the number of residential buildings that would not have flooded without the added impact of climate change.

Figure 9. Each hexagonal bin symbolizes the number of residential buildings that would not have flooded without the added impact of climate change in Harris County, Texas, during Hurricane Harvey. These calculations were made using a 38% attributable precipitation increase from climate change. Source: Kevin T. Smiley et al., Social Inequalities in Climate Change-Attributed Impacts of Hurricane Harvey, 13 Nature Commc’ns 1 (2021) (Fig. 1).

Census data reveals that Hurricane Harvey’s flood damages were not distributed equally across socioeconomic groups. Figure 10 reveals that while Hispanic households comprise about 36% of the population Harris County, about one-half of the flooded homes were Hispanic households. The percentage was about the same whether or not climate change caused these homes to be flooded, as the percentage is relatively insensitive to which precipitation attribution statement is used. Additional analysis reveals that damages increased with wealth in white neighborhoods. In Hispanic neighborhoods, the situation was reversed, with damages increasing with poverty. With documentation of the relative contribution that wealthy households make to increases in greenhouse gases compared with poor households, such analyses can be used to quantify environmental and other social injustices.1

Other human impacts of extreme weather have been quantified. Of particular interest are the effects of climate change on deaths resulting from heatwaves, which are the deadliest of all extreme weather events.2 Epidemiology studies have developed relationships between mortality risk and temperature.3 These curves tend to steepen at very high temperatures, implying that small increases in temperature at the high end have large increases in mortality. By estimating the attributable human temperature increase during a heatwave and using the observedtemperature, the change in mortality risk can be estimated.

  • 1Kevin T. Smiley et al., Social Inequalities in Climate Change-Attributed Impacts of Hurricane Harvey, 13 Nature Commc’ns 1 (2021).
  • 2World Meteorological Org., Atlas of Mortality and Economic Losses From Weather, Climate and Water Extremes (1970-2019) (2021).
  • 3Michela Baccini et al., Heat Effects on Mortality in 15 European Cities, 19 Epidemiology 711 (2008), https://doi.org/10.1097/EDE.0b013e318176bfcd.
The average percentage of household properties flooded during Hurricane Harvey varied by racial and ethnic group.

Figure 10. The average percentage of household properties flooded during Hurricane Harvey varied by ethnic group. Green: Not flooded. Red: Flooded without climate change. Orange: Flooded because of climate change (with a 38% human-induced precipitation increase). Source: Kevin T. Smiley et al., Social Inequalities in Climate Change-Attributed Impacts of Hurricane Harvey, 13 Nature Commc’ns 1 (2021) (Figure 2).

Another method maps these mortality/temperature curves onto temperature changes to produce plots like Figure 6 of return periods for mortality risk. This makes it possible to estimate both the number of people who died because of the influence of climate change on a heatwave and the change in probability of mortality. This technique was used to estimate that over 500 people died because of climate change in Paris during the 2003 European heat wave.1 Methodologies to extend attribution statements about the weather to the human impacts of extreme events are an active area of research.2  

Finally, not all climate impacts come in the form of extreme weather events. Climate change also causes subtle shifts in weather, such as additional warm days per year or fewer cool days per year, that can have substantial human impacts. For instance, climate scientists have developed an index to characterize the influence of climate change on the temperature on any given day and region in the United States, including both extreme and more modest temperatures.3  

VII. Attribution of Climate Change to Sources

Who is responsible for climate change and its associated impacts? While this question extends beyond science and into the realm of ethics, philosophy, and law, scientific research in the field of source attribution can inform thinking on this complex issue.

One of the first things to consider when assessing responsibility for climate change is the source and the emissions derived from that source. The source may be an actor such as a country or a company, an economic sector, or a human activity. A given source’s contribution to climate change may be derived from observational data of greenhouse gas emissions, modeling, or corporate and governmental reports of emissions. Uncertainties in these estimates come from data gaps, the unknown climatic impacts of historical land use changes, and the nonlinear behavior of greenhouse gases in the climate system, among other factors.4  

With these uncertainties in mind, a source’s proportional contribution to climate change can be estimated by dividing the emissions associated with that source by the total of accumulated anthropogenic emissions. This is a reasonable approach since greenhouse gases are well-mixed in the atmosphere, and therefore a given molecule of carbon dioxide cannot be attributed to a specific source. From the perspective of the climate system, this also means that it does not matter from where a given molecule of carbon dioxide originates, because the emissions of a single source impact the climate at the global level rather than at just the location of that source.

To tie the emissions of a source to a specific climate impact, models must first be used to estimate the contribution of a source’s emissions to the concentration of greenhouse gases in the atmosphere. That incremental change in atmospheric concentration then must be linked to a given impact of climate change, such as sea-level rise or a heatwave.

The field of greenhouse gas accounting has important implications for climate law and governance. Notably, the methodological approach taken when conducting a greenhouse gas accounting survey can dramatically influence the results of that survey. Three such accounting methods have been devised for government-based accounting: (1) territorial accounting, which considers only emissions that are directly generated within a given country or territory, (2) consumption-based accounting, which considers additional emissions embodied in products that are imported into a country or territory, and (3) extraction-based accounting, which considers the emissions associated with the combustion of exported fossil fuels from the country or territory. While the United Nations Framework Convention on Climate Change currently uses the territorial accounting approach, there is an ongoing push for countries to quantify additional indirect emissions.5  

A private-sector analog to these government-based emissions accounting methods is the characterization of emissions into three “scopes.” Scope 1 includes direct emissions associated with company operations. Scope 2 includes indirect emissions associated with purchasing energy such as electricity, steam, heat, or cooling. Scope 3 encompasses all indirect emissions throughout the full value chain of a company not already covered by scope 2, especially those generated by the consumption of products created through the burning of fossil fuels. At the time of writing, the Securities and Exchange Commission (SEC) has proposed rules to make the reporting of scope 1 and 2 emissions, in addition to scope 3 emissions in some cases, mandatory.6  

In litigation related to climate impacts, the first step in assessing responsibility is attributing the emissions of a particular country or entity to its proportional contribution to climate change. The second step is assigning to an impact that source’s contribution to climate change. The first study to do this investigated the proportional contribution of the emissions of individual nation-states to global mean surface temperature and, subsequently to an Argentinian heatwave.7 Interestingly, the authors found that the framing of this question matters significantly to the outcome. Calculating a proportional contribution derived from quantifying the likelihood of the heatwave, if a given region had been the only region to emit, yields a different result than calculating a proportional contribution derived from the likelihood of the heatwave if that region had not emitted.

Judges are increasingly being asked to assign responsibility for climate change. Numerous states and several local governments have brought suit against the world’s largest oil companies, their associations, and others for climate-related damages, as described in the Applying Attribution module. One kind of lawsuit alleges that the companies worked to delay climate policies and are therefore responsible for some amount of the climate damages with which these governments are now burdened. If and when such cases come to trial, source attribution science will likely play a central role.

VIII. Conclusions

According to the Sixth Assessment Report of the IPCC: “It is unequivocal that human influence has warmed the atmosphere, ocean and land. Widespread and rapid changes in the atmosphere, ocean, cryosphere and biosphere have occurred”.8 Such a statement could not have been made without the many D&A analyses that underlie it.

Developments in attribution science over the past two decades have made possible many robust statements about the human influence on climate. These statements extend to both long-term trends and extreme events, including heatwaves, floods, droughts, and storms. The extension of attribution science to socioeconomic damages and inequality is now underway and is likely to become an important factor in assigning responsibility in legal proceedings.

  • 1Daniel Mitchell et al., Attributing Human Mortality During Extreme Heat Waves to Anthropogenic Climate Change, 11 Env’t Rsch. Letters 1 (2016), https://doi.org/10.1088/1748-9326/11/7/074006.
  • 2Sarah Perkins-Kirkpatrick et al., On the Attribution of the Impacts of Extreme Weather Events to Anthropogenic Climate Change, 17 Env’t Rsch. Letters 1 (2021), http://doi.org/10.1088/1748-9326/ac44c8.
  • 3The tool is available at https://www.climatecentral.org/tools/climate-shift-index.
  • 4Michael Burger et al., The Law and Science of Climate Change Attribution, 45 Colum. J. Env’t L. 57 (2020), https://doi.org/10.7916/cjel.v45i1.4730; Rupert F. Stuart-Smith et al., Filling the Evidentiary Gap in Climate Litigation, 11 Nature Climate Change 651 (2021), https://doi.org/10.1038/s41558-021-01086-7; Richard Heede, Tracing Anthropogenic Carbon Dioxide and Methane Emissions to Fossil Fuel and Cement Producers, 1854-2010, 122 Climatic Change 229 (2014), https://doi.org/10.1007/s10584-013-0986-y; B. Ekwurzel et al., The Rise in Global Atmospheric CO2, Surface Temperature, and Sea Level Rise From Emissions Traced to Major Carbon Producers, 144 Climatic Change 579 (2017), https://doi.org/10.1007/s10584-017-1978-0.
  • 5Burger et al, The Law and Science, supra note 42.
  • 6The Enhancement and Standardization of Climate-Related Disclosures for Investors, 87 Fed. Reg. 29059 (proposed May 12, 2022) (to be codified at 17 C.F.R. pt. 200).
  • 7Friederike E.L. Otto et al., Assigning Historic Responsibility for Extreme Weather Events, 7 Nature Climate Change 757 (2017), https://doi.org/10.1038/nclimate3419.
  • 8IPCC, Climate Change 2021: The Physical Science Basis, Summary for Policymakers (2021).
    Stay connected with the Climate Judiciary Project. This information allows us to better support education about climate science and the law and share new resources as they become available.

    Name Section

    I am a: (select one)
    This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.