|
The evidence for AGW (Anthropogenic Global Warming) is so overwhelming that the IPCC says, in its sixth assessment report (IPCC AR6 WG1, 2021): "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" There are many lines of evidence supporting their statement, well documented in the various IPCC reports and elsewhere. For example, NASA provides a good, easily accessible outline as a starting point for any reader. One such line of evidence is climate modelling. As well as providing part of the evidence for AGW that has occurred to date, and its causes (largely human activities in the time since the Industrial Revolution), climate modelling can also, to some extent and with uncertainty ranges, provide a means of projecting forward the amount of future warming. The latter is useful as reference material informing public policy decisions on responses to climate change at national and global levels. This is done largely through various modelled scenarios for future greenhouse gas emissions trajectories and their effects on global warming temperatures and impacts of weather extremes on human communities, systems (such as agriculture), health and wellbeing, patterns of migration etc. Especially because of this dual role of climate modelling (past evidence and future projections), it has for many decades been the target of anti-AGW dismissives, disinformers and propagandists, who seek to sow FUD (Fear, Uncertainty and Doubt) on the need to act robustly on AGW. For this reason, we need to be careful in talking about climate modelling in general discourse. We need to repeatedly point out how useful it is, and that it doesn't need to be 100% perfect to provide valid information that policy makers can base sound decisions on. At the same time, we need to recognise the limitations of climate modelling, and the uncertainties inherent in the use of such techniques, while not unwittingly providing material for the anti-AGW FUD propagandists. The right balance needs to be struck in communicating a very technical subject in language that imparts the usefulness of the technique, and the confidence that can be placed in its results, without unduly expressing uncertainties in ways that can be deliberately misinterpreted and misrepresented by anti-AGW propagandists. That is the context in which the main part of this article reviews the 2022 book "Escape from Model Land" by Erica Thompson, which was recently recommended to me in a discussion about AGW in an online professional channel. https://www.lse.ac.uk/granthaminstitute/profile/erica-thompson/ "Dr Erica Thompson is a UKRI Future Leaders Fellow based in the LSE Data Science Institute, where she works on a programme of research around the use and interpretation of mathematical models. Her research interests focus on how to identify meaningful and useful projections of future climate, how different types of model output can be used to inform these projections, and how to think about uncertainty. In the context of climate change, her recent work looks at the value judgements embedded within mathematical models of future climate and energy systems, the problems with statistical interpretation of climate model ensembles (with Dave Stainforth), and the need for diversity in climate modelling approaches." Dr Thompson's expertise spans maths, physics and climate, so she is in an excellent position to comment on the application of climate modelling to the problem of AGW. Researchers into the use of mathematical techniques to support public policy making, such as Dr Thompson, are making an important contribution to the interfaces between science and policy. In contrast, it's not easy for other people, who might not be either climate modellers or mathematicians, to understand sufficiently well all the statistical and mathematical aspects of climate modelling. It's a complex topic, because climate science itself is complex. At some point, most of us need to place reliance on experts such as Dr Thompson. Thompson, in her book, displays a sound knowledge of the application of mathematical modelling to climate change (and some other topics she mentions). The book provides some excellent insights into how models are constructed and operated, how they can provide useful information for policy makers, how they have limitations and how to overcome those limitations. One aspect she doesn't cover is the detection of the human-driver global warming "signal" from the overlayed natural climate variabilty "noise", which is one very important application of climate modelling. That's an important omission when talking about climate change, and I'll deal with it first, before dealing with other aspects of Thompson's book. Despite the complexity of climate science, it's a subject that has been studied extensively for several decades, one could even say hundreds of years. That has provided huge amounts of observational data. The "signal" of human-driven global warming was, until a few decades ago, only a theoretical possibility. With the advent of increasingly accurate climate models, and advances in statistical computing power and techniques, it has in more recent decades become possible to detect that signal with more and more confidence. Benjamin D Santer, Program for Climate Model Diagnosis and Intercomparison (PCMDI), Lawrence Livermore National Laboratory, Livermore, California, has written about that increasing confidence, expressed in terms of numbers of standard deviations. https://en.wikipedia.org/wiki/Benjamin_D._Santer In a paper co-authored by Santer in 2019, "Celebrating the anniversary of three key events in climate change science", he sets out how the five and six sigma (ie five and six standard deviations) confidence levels on the existence of AGW "signal" over the natural variations "noise" have been reached over recent decades, as per the attached graph.  The simple message from this work is that the chance of AGW being false is now vanishingly low. Any event that is extremely rare, beyond the sixth standard deviation in a normal distribution, is known as a six sigma event. The probability of such an event happening would be about two in a billion. In other words, the chance of observed climate variability (including global warming), since the Industrial Revolution, being because of natural climate variability alone, is about 2 in a billion. The IPCC includes in AR6 WG1 (2021) the following chart, which shows in a very visible way how the AGW signal emerges from the natural variability noise in about the 1980s. This was established by comparing climate model runs with and without the human drivers, and it demonstrated, with increasing levels of confidence, that current global average temperatures could not have occurred by natural climate variability alone, ie without the human drivers happening as well.  Thompson states her view about AGW and appears to have (somewhat guarded) confidence in climate modelling:
"Climate change is happening, it will have serious consequences and we must decide what to do about it in the absence of perfect information... " In the same breath, however, Thompson raises the matter of the Hawkmoth Effect, and it's not clear the extent to which she suggests this obscures or affects confidence we should place in climate models: "... [while] the Hawkmoth Effect serves to widen somewhat the ranges of plausible outcomes, it also supports a stance of humility about our ability to model, predict and adapt to changes, and therefore speaks in favour of greater efforts in climate change mitigation by reduction of [greenhouse gas] emissions... By the way, the existence of the Hawkmoth Effect is not a reason to doubt the characteristics of projections made by climate models are real possibilities. It is only reason to be cautious in interpreting the detailed projections of climate models, and especially cautious in assigning probabilities to outcomes..." Thompson explains the Hawkmoth Effect here: http://eprints.lse.ac.uk/57935/1/Thompson_Hawkmoth-Effect_LSEResearchFestival2014.pdf This theoretical effect is a form of structural instability of a particular model or modelled physical reality, in which a small perturbation or alteration in the starting state at one point in time can result in a very large change in the resulting outcome state or states at a later time in the modelled timeframe. In the context of climate modelling, this is particularly significant where the trajectories of human societies are modelled, in terms of societal characteristics such as demographics, energy use, urban designs, transport, industrial and agricultural systems etc. All these characteristics are subject to individual and collective decision-making, which is in itself influenced by the results of climate modelling, and provides inputs to some climate models. Thompson is right to advise caution in assessing the likelihoods associated with future scenarios projected using climate models, where such models include significant elements that can be subject to the Hawkmoth Effect (although it should be pointed out that many climate models are not subject to this effect, and those that do will be affected to a greater or lesser extent). At the same time, it's a genuine criticism of most, if not all, mainstream climate models, that they are not good at modelling climate tipping points. Tipping points are discontinuities in the trends in projected trajectories, resulting in dramatically different outcome states from those that might result from incremental, less discontinuous changes over time. An example in the AGW context is the potential for sudden (in geological timeframes) melting of both polar ice caps, which would result in hundreds of feet of sea level rise, perhaps in as little as a couple of hundred years. It is because such sudden discontinuities and tipping points are so rare and difficult to predict in advance, that it has proved extremely problematic for climate modellers to try to build such discontinuities into their models. There is a dilemma here. The more effort that is put into trying to model tipping points, the greater the chances that such models will produce outputs with greater ranges of uncertainty, which could reduce their perceived accuracy or precision. Nevertheless, the main message is that most climate models are, if anything, understating the possibilities of tipping points occurring and dramatically altered outcome states resulting. In that sense, models are "conservative" and this should give cause to consider applying the Precautionary Principle in designing public policy responses to AGW. Climate models understate the risks of tipping points, but public policy makers cannot afford to ignore tipping point risks. There's a good series of "explainers" about climate tipping points, with plentiful references to mainstream climate research supporting their significance and relevance to climate science and policy-making, in this 2020 series of articles: https://www.carbonbrief.org/explainer-nine-tipping-points-that-could-be-triggered-by-climate-change/ from which: "By its Fall Meeting of 2008, the AGU [American Geophysical Union] had an entire half-day session dedicated to climate tipping points. A Science [Magazine] briefing about the meeting declared that 'tipping points, once considered too alarmist for proper scientific circles, have entered the climate change mainstream'." I note that Thompson only mentions climate tipping points once in her book (on page 154), and only then in rather lukewarm terms: "At the moment, a very wide range of possible outcomes, especially for high warming scenarios, are thought to be plausible - these include the kinds of 'tipping points' often mentioned in the more pessimistic fringe of climate literature, such as changes in ocean circulation or rapid collapse of ice sheets... If we cannot rule these out, they become a dominant consideration for some decision-making frameworks, feeding 'doomer' scenarios of catastrophic climate change. If we could rule them out with confidence, we might increase the likelihood of agreement about climate policy." The closest she comes to recognising the possibility of tipping points as a valid climate modelling consideration is when she talks about the Hawkmoth Effect, as described above. Other than that, she appears to associate tipping points with "pessimistic" and "doomer" viewpoints. It's difficult to see clearly what she thinks about climate tipping points. For example, does she think they could actually happen, or is she dismissive of them, attributing them to the work of "pessimists" and "doomers" rather than the work of mainstream of climate researchers? In this respect, her book stops short on this topic, and this counts as another important omission. I'll draw towards closing by mentioning two more aspects Thompson addresses. The first is when she gives us cause to think about the use of climate modelling to make projection scenarios of possible and plausible future climate changes, where socio-economic trends, at scale, in human societies are one of the modelling inputs (and outputs), of particular importance in the realm of public policy making. This is where she brings in the concept of "performativity", where public policies themselves become akin to 'self-fulfilling prophesies' because of their impacts on people's behaviours. At a macro level of analysis, the collective behaviours of all of humanity is at the core of what has resulted in the problem of AGW, and will be our best hope of successfully tackling it. https://en.wikipedia.org/wiki/Performativity The other is the importance of addressing limitations in certain types of climate modelling by encouraging diversity of modelling, and of modellers. This overcomes the possibility of results from some types of climate models converging because of lack of differences in the underlying base assumptions used to build them. This might present a 'blind spot', where results are less accurate than they could be. Whereas this is less of a risk in climate models that only look at biological or physical processes and data, it is a more significant consideration in models that attempt to describe (and predict) socio-economic characteristics of human societies at scale globally or by region. Although I've given some attention to what I see as omissions, Thompson's book stands on its own very well and provides a very useful contribution to debates about the interfaces between science, mathematics and public policy making. Those interfaces will be increasingly important as the world progresses in its responses to AGW, moving from preliminary steps to full-scale transition, decarbonisation, global net zero and beyond. My comments here are intended to be pointers towards some additional material that helps to put Thompson's book into a wider context, in which it can become more useful in understanding the proper place and contribution of climate modelling to building a sustainable future for all. In this interactive interview at the Edison Electric Institute in 2023, Thompson talks about many of the points from her book, with a particular emphasis on climate change and the energy industry transition to global net zero (about an hour long): https://www.youtube.com/watch?v=wAQZQQJwHWk
0 Comments
Leave a Reply. |
