The media is again awash with claims of “human footprints” in extreme weather events, with headlines like these:

“Global warming is making hot days hotter, rainfall and flooding heavier, hurricanes stronger and droughts more severe.”

“Global climate change is making weather worse over time”

“Climate change link to extreme weather easier to gauge”– U.S. Report

“Heat Waves, Droughts and Heavy Rain Have Clear Links to Climate Change, Says National Academies”

That last one refers to a paper just released by the National Academy of Sciences Press: Attribution of Extreme Weather Events in the Context of Climate Change (2016)

And as usual, the headline claims are unsupported by the actual text. From the NAS report (here):

Attribution studies of individual events should not be used to draw general conclusions about the impact of climate change on extreme events as a whole. Events that have been selected for attribution studies to date are not a representative sample (e.g., events affecting areas with high population and extensive infrastructure will attract the greatest demand for information from stakeholders) P 107

Systematic criteria for selecting events to be analyzed would minimize selection bias and permit systematic evaluation of event attribution performance, which is important for enhancing confidence in attribution results. Studies of a representative sample of extreme events would allow stakeholders to use such studies as a tool for understanding how individual events fit into the broader picture of climate change. P 110

Correctly done, attribution of extreme weather events can provide an additional line of evidence that demonstrates the changing climate, and its impacts and consequences. An accurate scientific understanding of extreme weather event attribution can be an additional piece of evidence needed to inform decisions on climate changerelated actions. P. 112

The Indicative Without the Imperative

The antidote to such feverish reporting is provided by Mike Hulme in a publication: Attributing Weather Extremes to ‘Climate Change’: a Review (here).

He has an insider’s perspective on this issue, and is certainly among the committed on global warming (color him concerned). Yet here he writes objectively to inform us on X-weather, without advocacy: real science journalism and a public service, really.

Overview

In this third and final review I survey the nascent science of extreme weather event attribution. The article proceeds by examining the field in four stages: motivations for extreme weather attribution, methods of attribution, some example case studies and the politics of weather event Attribution.

The X-Weather Issue

As many climate scientists can attest, following the latest meteorological extreme one of the most frequent questions asked by media journalists and other interested parties is: ‘Was this weather event caused by climate change?’

In recent decades the meaning of climate change in popular western discourse has changed from being a descriptive index of a change in climate (as in ‘evidence that a climatic change has occurred’) to becoming an independent causative agent (as in ‘climate change caused this event to happen’). Rather than being a descriptive outcome of a chain of causal events affecting how weather is generated, climate change has been granted power to change worlds: political and social worlds as much as physical and ecological ones.

To be more precise then, what people mean when they ask the ‘extreme weather blame’ question is: ‘Was this particular weather event caused by greenhouse gases emitted from human activities and/or by other human perturbations to the environment?’ In other words, can this meteorological event be attributed to human agency as opposed to some other form of agency?

The Motivations

Hulme shows what drives scientists to pursue the “extreme weather blame” question, noting four motivational factors.

Why have climate scientists over the last ten years embarked upon research to provide an answer beyond the stock phrase ‘no individual weather event can directly be attributed to greenhouse gas emissions’?  There seem to be four possible motives.

1.Curiosity
The first is because the question piques the scientific mind; it acts as a spur to develop new rational understanding of physical processes and new analytic methods for studying them.

2.Adaptation
A second argument, put forward by some, is that it is important to know whether or not specific instances of extreme weather are human-caused in order to improve the justification, planning and execution of climate adaptation.

3.Liability
A third argument for pursuing an answer to the ‘extreme weather blame’ question is inspired by the possibility of pursuing legal liability for damages caused. . . If specific loss and damage from extreme weather can be attributed to greenhouse gas emissions – even if expressed in terms of increased risk rather than deterministically – then lawyers might get interested.

The liability motivation for research into weather event attribution also bisects the new international political agenda of ‘loss and damage’ which has emerged in the last two years. . . The basic idea is to give recognition that loss and damage caused by climate change is legitimate ground for less developed countries to gain access to new international climate adaptation funds.

4. Persuasion
A final reason for scientists to be investing in this area of climate science – a reason stated explicitly less often than the ones above and yet one which underlies much of the public interest in the ‘extreme weather blame’ question – is frustration with and argument about the invisibility of climate change. . . If this is believed to be true – that only scientists can make climate change visible and real –then there is extra onus on scientists to answer the ‘extreme weather blame’ question as part of an effort to convince citizens of the reality of human-caused climate change.

Attribution Methods

Attributing extreme weather events to human influences requires different approaches, of which four broad categories can be identified.

1. Physical Reasoning
The first and most general approach to attributing extreme weather phenomena to rising greenhouse gas concentrations is to use simple physical reasoning.

General physical reasoning can only lead to broad qualitative statements such as ‘this extreme weather is consistent with’ what is known about the human-enhanced greenhouse effect. Such statements offer neither deterministic nor stochastic answers and clearly underdetermine the ‘weather blame question.’ It has given rise to a number of analogies to try to communicate the non-deterministic nature of extreme event attribution. The three most widely used ones concern a loaded die (the chance of rolling a ‘6’ has increased, but no single ‘6’ can be attributed to the biased die), the baseball player on steroids (the number of home runs hit increases, but no single home run can be attributed to the steroids) and the speeding car-driver (the chance of an accident increases in dangerous conditions, but no specific accident can be attributed to the fast-driving).

2. Classical Statistical Analysis
A second approach is to use classical statistical analysis of meteorological time series data to determine whether a particular weather (or climatic) extreme falls outside the range of what a ‘normal’ unperturbed climate might have delivered.

All such extreme event analyses of meteorological time series are at best able to detect outliers, but can never be decisive about possible cause(s). A different time series approach therefore combines observational data with model simulations and seeks to determine whether trends in extreme weather predicted by climate models have been observed in meteorological statistics (e.g. Zwiers et al., 2011, for temperature extremes and Min et al., 2011, for precipitation extremes). This approach is able to attribute statistically a trend in extreme weather to human influence, but not a specific weather event. Again, the ‘weather blame question’ remains underdetermined.

3. Fractional Attributable Risk (FAR)
Taking inspiration from the field of epidemiology, this method seeks to establish the Fractional Attributable Risk (FAR) of an extreme weather (or short-term climate) event. It asks the counterfactual question, ‘How might the risk of a weather event be different in the presence of a specific causal agent in the climate system?’

The single observational record available to us, and which is analysed in the statistical methods described above, is inadequate for this task. The solution is to use multiple model simulations of the climate system, first of all without the forcing agent(s) accused of ‘causing’ the weather event and then again with that external forcing introduced into the model.

The credibility of this method of weather attribution can be no greater than the overall credibility of the climate model(s) used – and may be less, depending on the ability of the model in question to simulate accurately the precise weather event under consideration at a given scale (e.g. a heatwave in continental Europe, a rain event in northern Thailand) (see Christidis et al., 2013a).

4. Eco-systems Philosophy
A fourth, more philosophical, approach to weather event attribution should also be mentioned. This is the argument that since human influences on the climate system as a whole are now clearly established – through changing atmospheric composition, altered land surface characteristics, and so on – there can no longer be such a thing as a purely natural weather event. All weather — whether it be a raging tempest or a still summer afternoon — is now attributable to human influence, at least to some extent. Weather is the local and momentary expression of a complex system whose functioning as a system is now different to what it would otherwise have been had humans not been active.

Results from Weather Attribution Studies

Hulme provides a table of numerous such studies using various methods, along with his view of the findings.

It is likely that attribution of temperature-related extremes using FAR methods will always be more attainable than for other meteorological extremes such as rainfall and wind, which climate models generally find harder to simulate faithfully at the spatial scales involved. As discussed below, this limitation on which weather events and in which regions attribution studies can be conducted will place important constraints on any operational extreme weather attribution system.

Political Dimensions of Weather Attribution

Hulme concludes by discussing the political hunger for scientific proof in support of policy actions.

But Hulme et al. (2011) show why such ambitious claims are unlikely to be realised. Investment in climate adaptation, they claim, is most needed “… where vulnerability to meteorological hazard is high, not where meteorological hazards are most attributable to human influence” (p.765). Extreme weather attribution says nothing about how damages are attributable to meteorological hazard as opposed to exposure to risk; it says nothing about the complex political, social and economic structures which mediate physical hazards.

And separating weather into two categories — ‘human-caused’ weather and ‘tough-luck’ weather – raises practical and ethical concerns about any subsequent investment allocation guidelines which excluded the victims of ‘tough-luck weather’ from benefiting from adaptation funds.

Contrary to the claims of some weather attribution scientists, the loss and damage agenda of the UNFCCC, as it is currently emerging, makes no distinction between ‘human-caused’ and ‘tough-luck’ weather. “Loss and damage impacts fall along a continuum, ranging from ‘events’ associated with variability around current climatic norms (e.g., weather-related natural hazards) to [slow-onset] ‘processes’ associated with future anticipated changes in climatic norms” (Warner et al., 2012:21). Although definitions and protocols have not yet been formally ratified, it seems unlikely that there will be a role for the sort of forensic science being offered by extreme weather attribution science.

Conclusion

Thank you Mike Hulme for a sane, balanced and expert analysis. It strikes me as being another element in a “Quiet Storm of Lucidity”.

Is that light the end of the tunnel or an oncoming train?

 

Recently I addressed this claim by referring to this chart produced by scientists from AARI.

 

In their commentary, it is clear why the data does not support claiming fossil fuels cause global warming.  From Frolov et al. 2009:

The WFC curve shows an exponential increase, which doubles approximately every 30 years, increasing 25-fold since the middle of the nineteenth century. The global air temperature anomaly curve shows a positive trend of +0.06°C/10 years (Sonechkin et al., 1997). At the same time, there are cyclic changes with periods of about 60 years. The correlation between these curves changes its sign every 30 years, varying from —0.88 (1940 1970) to +0.94 (1970 2000). Hence, there is no direct linear connection between WFC (which indirectly represents CO2 concentration in the atmosphere) and global air temperature. The authors of this study therefore conclude that the WFC increase is not an obvious cause of the increase in global air temperature.

In this post, I am bringing the analysis up to date by showing World Fossil Fuel Consumption (WFFC) compared to Global Temperature Anomalies.

The WFFC numbers come from US EIA and can be accessed here. I have included only the statistics for coal, oil and gas, which comprise 91% of total energy consumed. The remainder are hydro, nuclear and other renewables. 2015 numbers are not yet available. UAH version 6 provides global temperature anomalies for the lower troposphere.

The correlation overall is moderately positive at 0.60, but the two patterns are markedly different because of the 1998 event. 1980 to 2000 (the period overlapping with the AARI graph) shows a weakly positive 0.48 correlation. From 2000 to 2014 the correlation is almost non-existent at 0.07.

Summary

In the long-term and in the recent short-term, use of fossil fuels is not the obvious cause of temperature changes. The context and background for reaching this conclusion is provided below (From the previous post.)

Legal Test of Global Warming

In a previous post (here), I discussed the Bradford Hill protocol that has become precedent for trials concerning scientific evidence for legal liability.

Bradford Hill was the jurist who brought clarity and  methodology for the courts to consider and rule on accusations such as:

Thalidimide is causing birth defects;
Asbestos dust is causing lung disease;
as well as frequent claims of causal relationships between illness, injury and conditions of work.

The Global Warming Claim

When it comes to Global Warming, the proposition is straightforward:
Rising fossil fuel emissions are causing rising global temperatures.

The procedure to test that claim is described by Nathan Schachtman here.

Proper epidemiological methodology begins with published study results which demonstrate an association between a drug and an unfortunate effect. Once an association has been found, a judgment as whether a real causal relationship between exposure to a drug and a particular birth defect really exists must be made. 

Step 1: Establish an association between two variables.
Proper epidemiological method requires surveying the pertinent published studies that investigate whether there is an association between the medication use and the claimed harm. The expert witnesses must, however, do more than write a bibliography; they must assess any putative associations for “chance, confounding or bias”:

Step 2: Rule out chance as an explanation
The appropriate and generally accepted methodology for accomplishing this step of evaluating a putative association is to consider whether the association is statistically significant at the conventional level.
“Generally accepted methodology considers statistically significant replication of study results in different populations because apparent associations may reflect flaws in methodology.”

Step 3: Rule out bias or confounding factors.
The studies must be structured to analyze and reject other factors or influences, such as non-random sampling, additional intervening variables such as demographic or socio-economic differences.

Step 4: Infer Causation by Applying Accepted Causative Factors
Most often legal proceedings follow the Bradford Hill factors, which are delineated here.

By way of context Bradford Hill says this:

None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis and none can be required as a sine qua non. What they can do, with greater or less strength, is to help us to make up our minds on the fundamental question – is there any other way of explaining the set of facts before us, is there any other answer equally, or more, likely than cause and effect?

Such is the legal terminology for the “null” hypothesis: As long as there is another equally or more likely explanation for the set of facts, the claimed causation is unproven.

The Causative Factors

What aspects of that association should we especially consider before deciding that the most likely interpretation of it is causation?

(1) Strength. First upon my list I would put the strength of the association.

(2) Consistency: Next on my list of features to be specially considered I would place the consistency of the observed association. Has it been repeatedly observed by different persons, in different places, circumstances and times?

To test the Global Warming claim, let’s consider the association between world fuel consumption (WFC) and surface air temperatures (SAT):

In Figure 5.1, the dynamics of global air temperature anomalies obtained from instrumental measurements over the last 140 years is compared with changes in world fuel consumption (WFC) (Makarov, 1998). The WFC curve shows an exponential increase, which doubles approximately every 30 years, increasing 25-fold since the middle of the nineteenth century. The global air temperature anomaly curve shows a positive trend of +0.06°C/10 years (Sonechkin et al., 1997). At the same time, there are cyclic changes with periods of about 60 years. The correlation between these curves changes its sign every 30 years, varying from —0.88 (1940 1970) to +0.94 (1970 2000). Hence, there is no direct linear connection between WFC (which indirectly represents CO2 concentration in the atmosphere) and global air temperature. The authors of this study therefore conclude that the WFC increase is not an obvious cause of the increase in global air temperature.

The other causative factors could be applied, but can not add weight against the argument above.

Case Closed

The legal methodology above is used to decide the causal relationship between two variables. Clearly, in Climate Science the starting question is: Do rising fossil fuel emissions cause temperatures to rise? Those who have been following the issue know that there are many arguments underneath: Why do not temperatures always rise along with CO2? Has chance been eliminated? Are not natural factors confounding the association? And so on.

For myself, I will join in the conclusion reached by Frolov et al., who go on to further explain their position:

In general, although climate models are based on physics, they inevitably include a number of adjustable parameters that are fitted to past temperature changes. We are not aware of a single climate model based on fundamental physics without adjustable parameters that has been subjected to a rigorous test against actual climate data. Climate modelers appear to assume that the Earth’s climate would continue without change, were it not for greenhouse gas emissions. They do not take into account the possibility that natural climate cycles are also acting independently of effects induced by buildup of greenhouse gas concentrations. As we have shown in Chapter 4, there is evidence for cyclic variability of Arctic climates. Furthermore, there is considerable evidence for past variability of global climate as expressed in the so-called Medieval Warm Period (900-1100) and the Little Ice Age (1600-1850). These fluctuations appear to be as great as the temperature rise of the 20th century, yet, there was no contribution of greenhouse gases to these climate changes.

A major challenge in climate modeling is to understand the range of natural fluctuations, and separate these from climate changes induced by human activity (greenhouse gas emissions, land clearing, irrigation, …). The models neglect natural fluctuations because they have no means of incorporating them, and put the entire blame for climate changes since the 19th century on human activity. As a result, they appear to project an extreme view of the future that seems unlikely to be reliable.

Again my thanks to Dr. Bernaerts for the copy of this book:

 

 

 

 

 

In a previous post (here), I discussed the Bradford Hill protocol that has become precedent for trials concerning scientific evidence for legal liability.

Bradford Hill was the jurist who brought clarity and  methodology for the courts to consider and rule on accusations such as:

Thalidimide is causing birth defects;
Asbestos dust is causing lung disease;
as well as frequent claims of causal relationships between illness, injury and conditions of work.

The Global Warming Claim

When it comes to Global Warming, the proposition is straightforward:
Rising fossil fuel emissions are causing rising global temperatures.

The procedure to test that claim is described by Nathan Schachtman here.

Proper epidemiological methodology begins with published study results which demonstrate an association between a drug and an unfortunate effect. Once an association has been found, a judgment as whether a real causal relationship between exposure to a drug and a particular birth defect really exists must be made. 

Step 1: Establish an association between two variables.
Proper epidemiological method requires surveying the pertinent published studies that investigate whether there is an association between the medication use and the claimed harm. The expert witnesses must, however, do more than write a bibliography; they must assess any putative associations for “chance, confounding or bias”:

Step 2: Rule out chance as an explanation
The appropriate and generally accepted methodology for accomplishing this step of evaluating a putative association is to consider whether the association is statistically significant at the conventional level.
“Generally accepted methodology considers statistically significant replication of study results in different populations because apparent associations may reflect flaws in methodology.”

Step 3: Rule out bias or confounding factors.
The studies must be structured to analyze and reject other factors or influences, such as non-random sampling, additional intervening variables such as demographic or socio-economic differences.

Step 4: Infer Causation by Applying Accepted Causative Factors
Most often legal proceedings follow the Bradford Hill factors, which are delineated here.

 

By way of context Bradford Hill says this:

None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis and none can be required as a sine qua non. What they can do, with greater or less strength, is to help us to make up our minds on the fundamental question – is there any other way of explaining the set of facts before us, is there any other answer equally, or more, likely than cause and effect?

Such is the legal terminology for the “null” hypothesis: As long as there is another equally or more likely explanation for the set of facts, the claimed causation is unproven.

The Causative Factors

What aspects of that association should we especially consider before deciding that the most likely interpretation of it is causation?

(1) Strength. First upon my list I would put the strength of the association.

(2) Consistency: Next on my list of features to be specially considered I would place the consistency of the observed association. Has it been repeatedly observed by different persons, in different places, circumstances and times?

To test the Global Warming claim, let’s consider the association between world fuel consumption (WFC) and surface air temperatures (SAT):

In Figure 5.1, the dynamics of global air temperature anomalies obtained from instrumental measurements over the last 140 years is compared with changes in world fuel consumption (WFC) (Makarov, 1998). The WFC curve shows an exponential increase, which doubles approximately every 30 years, increasing 25-fold since the middle of the nineteenth century. The global air temperature anomaly curve shows a positive trend of +0.06°C/10 years (Sonechkin et al., 1997). At the same time, there are cyclic changes with periods of about 60 years. The correlation between these curves changes its sign every 30 years, varying from —0.88 (1940 1970) to +0.94 (1970 2000). Hence, there is no direct linear connection between WFC (which indirectly represents CO2 concentration in the atmosphere) and global air temperature. The authors of this study therefore conclude that the WFC increase is not an obvious cause of the increase in global air temperature.

The other causative factors could be applied, but can not add weight against the argument above.

Case Closed

The legal methodology above is used to decide the causal relationship between two variables. Clearly, in Climate Science the starting question is: Do rising fossil fuel emissions cause temperatures to rise? Those who have been following the issue know that there are many arguments underneath: Why do not temperatures always rise along with CO2? Has chance been eliminated? Are not natural factors confounding the association? And so on.

For myself, I will join in the conclusion reached by Frolov et al., who go on to further explain their position:

In general, although climate models are based on physics, they inevitably include a number of adjustable parameters that are fitted to past temperature changes. We are not aware of a single climate model based on fundamental physics without adjustable parameters that has been subjected to a rigorous test against actual climate data. Climate modelers appear to assume that the Earth’s climate would continue without change, were it not for greenhouse gas emissions. They do not take into account the possibility that natural climate cycles are also acting independently of effects induced by buildup of greenhouse gas concentrations. As we have shown in Chapter 4, there is evidence for cyclic variability of Arctic climates. Furthermore, there is considerable evidence for past variability of global climate as expressed in the so-called Medieval Warm Period (900-1100) and the Little Ice Age (1600-1850). These fluctuations appear to be as great as the temperature rise of the 20th century, yet, there was no contribution of greenhouse gases to these climate changes.

A major challenge in climate modeling is to understand the range of natural fluctuations, and separate these from climate changes induced by human activity (greenhouse gas emissions, land clearing, irrigation, …). The models neglect natural fluctuations because they have no means of incorporating them, and put the entire blame for climate changes since the 19th century on human activity. As a result, they appear to project an extreme view of the future that seems unlikely to be reliable.

Again my thanks to Dr. Bernaerts for the copy of this book: