I’m Jonathan Burbaum, and this is Healing Earth with Technology: a weekly, Science-based, subscriber-supported serial. In this serial, I offer a peek behind the headlines of science, focusing (at least in the beginning) on climate change/global warming/decarbonization. I welcome comments, contributions, and discussions, particularly those that follow Deming’s caveat, “In God we trust. All others, bring data.” The subliminal objective is to open the scientific process to a broader audience so that readers can discover their own truth, not based on innuendo or ad hominem attributions, but instead based on hard data and critical thought.
You can read Healing for free, and you can reach me directly by replying to this email. If someone forwarded you this email, they’re asking you to sign up. You can do that by clicking here.
Like Science itself, I refer to previously established facts, so, if this is your first encounter, I recommend reading past posts in order. Starting with this issue, I’m including an estimated reading time (retroactively) so that you can gauge your commitment to the content. To catch up to this point will take approximately an hour of your time, in 10 minute chunks. [Or, of course, more, if you decide to think.]
“Life is divided into three parts: past, present, and future. Of these, the present is brief, the future doubtful, the past certain. For this last is the category over which fortune no longer has control, and which cannot be brought back under anyone's power. Preoccupied people lose this part; for they have no leisure to look back at the past, and even if they had it, there's no pleasure in recalling something regrettable.” Seneca in De Brevitate Vitae (“On the Shortness of Life”) ca. 55 AD. Translated by Gareth D. Williams. Link here. [Emphasis mine]
This quotation is pretty self-evident, but here’s what it means to me: To understand where we may be headed, it’s foolhardy to ignore history. Sadly, as Seneca notes, it is entirely a human proclivity to dismiss the past, particularly for the preoccupied (which is, frankly, most of us). Consider this: Data always exists in the past. Divination1 of the future, on the other hand, is unconstrained by reality.
The story continues…
One of my thematic anxieties has become gauging the ability of computer models, however sophisticated, to predict the unprecedented. This anxiety can be traced to the previous installment “3. The Earth is getting warmer. So what? (Part 1).” In that installment, I found comfort by considering that hurricane predictions, despite the presence of chaos-produced uncertainties, have improved enormously over the past few decades thanks to increasingly accurate models run on supercomputers. Later, I learned that climate models, while encumbered by the innate sensitivity of the “butterfly effect,” require significantly different assumptions than weather models. So a conceptual extrapolation isn’t as comforting as I thought it might be.
To wit: While physical models of the atmosphere used in weather prediction and climate modeling are largely interchangeable, the sustained temperature change of global warming also affects the oceans too. In the context of tomorrow’s weather, variations in ocean currents and surface temperatures are minor. But they can become major in the context of climate over decades to centuries.
In the vernacular of the field, climate models are referred to as Atmosphere-Ocean General Circulation Models (AOGCMs), a description that is decipherable to mortals. Thus, such models employ fluid dynamics (the physics of gases and liquids as they circulate) to describe the Earth’s climate-controlling systems, generating flow patterns we recognize as the Jet Stream and the Gulf Stream, among others. These flows control our climates by steering weather systems like the “Polar Vortex” and warming Northern Europe, for example, so it makes intuitive sense to use them to gauge climate change. In the latest draft of the IPCC report, there are no fewer than 39 distinct AOGCMs of different complexities,2 integrated into the report’s Coupled Model Intercomparison Project [Version 5] (CMIP5). While the specifics differ, all are calibrated on recent data, mostly that gathered since 1960.
Because the dire outcomes associated with “climate change” have not happened yet, testing the predictions of these climate models directly, either individually or in concert, is impossible. Instead, climate models must be validated by “hindcasting”, the semantic reverse of forecasting. Unlike human experiments, computer models have the advantage that they can run the clock backward. Thus, rather than predicting the future, climate scientists validate models by testing to see how well they describe climates in the past.
There are, of course, drawbacks to this approach. First, the further into the past we go, the less data we have. On the other hand, the result is already known, so models can be tweaked (either purposely or subconsciously) to reflect a past climate that’s only coincidentally useful for predicting the future. Second, the data about the ancient past necessarily has to be inferred from some sort of fossil record, so all the assumptions inherent in deriving the measurements from the data become important. Third, the effects of tiny changes over millennia, ones that have no bearing on our weather day-to-day, now become significant. We currently view these as Earth’s constants (like the precise location of the North Pole and the tilt of the Earth’s rotational axis versus the sun) and geologic processes that are observable but gradual (like the movement of Earth’s tectonic plates). But, it’s the best we can do. So, the most comprehensive climate models are largely validated with an abundance of recent, high-quality data from a (slightly) cooler Earth.
The dilemma now is that we’ve crossed into uncharted waters, at least in carbon dioxide terms. This means that models that describe the recent past are being called upon to describe the near future when we know for sure that the present is significantly different from anything we’ve seen in recorded history. This sharpens my anxiety to a fine point. The models must be trusted to predict an unknown future in the context of a chaos-driven present. All we know for sure is that we don’t know for sure. It’s reminiscent of the late Donald Rumsfeld’s “known unknowns”.
How the heck can we be confident in a climate model validated by hindcasting? Well, that’s a big topic for debate, and it’s a pet peeve: IPCC spends most of its brainpower laying out exactly when and how much we’re screwed, rather than outlining our strategic choices to avoid a climate disaster. But, as we’ve learned, nobody can be fully certain of today’s predictions because of chaos. And, if we wait for certainty, it will certainly be too late. So what we would like to achieve is confidence in the model’s future predictions (not so much when as we approach a tipping point), and that ends up being confidence in hindcasting. For me, the test is whether we can roll the clock back far enough to be relevant.
As I said before, this is a steep hill, so today’s question is:
“How far back in time do we need a climate model to accurately ‘hindcast’ to be confident of its ‘forecast’?”
This is a personal question, so I’m going to describe my process—please do your own thinking!
Figuratively, I choose to return to Antarctica, specifically the ice core data. It is thought that Antarctica began to accumulate ice around 34 million years ago, give or take a few million years, so there’s a lot of fossilized air under the surface. The longest continuous ice core has been drilled two miles under the surface, representing about 800,000 years of snowfall. That’s a lot of self-consistent data! We know that we can measure carbon dioxide directly, but can we infer temperature?
It turns out that scientists can—at least to a point. Water, as H2O, contains small amounts of heavier forms (known as isotopes) of both hydrogen and oxygen, and this “heavy water” has a slightly higher (+1.5°C) boiling point. So, the warmer or cooler the ocean, the more or less heavy water is in the air. When it falls as snow, the amount of “heavy” vs. “light” snow indicates the Earth’s temperature! While there may be annual and regional variations in temperature, averages over decades and centuries can be derived. This is the awesome power of experimental Science at work!
So what is the data? The full 800,000-year data set shows many oscillations in both temperatures (derived from water isotopes) and carbon dioxide concentrations (measured directly). They roughly track one another—higher carbon dioxide correlates with higher temperatures. The last time Earth was this warm was a period called the Eemian, and that was over 100,000 years ago. That’s a long time! Many small changes could make the extrapolations and computer horsepower needed to run today’s models backward very tenuous indeed. The Eemian is also called the “last interglacial”: We are presently in an “interglacial” period, a term meaning between Ice Ages (the cores reveal 5!). Is there a closer point for testing?
Well, yes. Climate change can go either way, and we know that Earth has been both cooler and warmer in the past. However, as suggested above, there was a glacial period (an Ice Age) between the Eemian and the present day, so what if we look backward to the last time Earth warmed up after an Ice Age? That was about 20,000 years ago; the temperature & carbon dioxide measurements from the ice core are:
Graph of carbon dioxide levels (measured directly, downloaded from NOAA) and local atmospheric temperatures (estimated from isotope measurements reported in EPICA community members. Eight glacial cycles from an Antarctic ice core. Nature 429, 623–628 (2004). https://doi.org/10.1038/nature02599) for the EPICA Ice Core since the end of the last Ice Age. Note the alignment between carbon dioxide and temperature and the remarkable rise of carbon dioxide over the last few centuries! Despite the general agreement between carbon dioxide levels and water temperatures in the south Pacific Ocean, the temperature data here is not global. In fact, ice cores from Greenland show much larger temperature swings before 10,000 BC. [Based on the first installment, “1. Is Global Warming Real?” local measurements generally track global phenomena.]
Oh, man! This really puts modern carbon dioxide levels in context—current levels are literally “off the charts”. There’s been a larger deviation in the past 350 years than between the end of the last Ice Age (20,000 BC) and pre-industrial Earth (10,000 BC to 1750 AD), corresponding to a +10°C increase in the South Pacific ocean surrounding Antarctica.
Of particular interest, there is a period between about 12,000 BC and 10,000 BC when the temperature was relatively constant. It is arcanely defined in early sediment studies as the “Younger Dryas”, named from an arctic flower, Mountain Avens (Dryas octopetala), which appears in ancient lake sediments. The assumption is that these flowers thrived when the temperature surrounding the lake was much lower. Such sediments can be dated from annual layers, similar to ice cores and tree rings, and also from carbon dating, so the timing is well known.
This period is well documented, and we know that there were significant differences in climates before 10,000 BC. So, this was truly the last period of climate change. Therefore, I think I’d feel fairly confident in predictions of future climate change if an AOGCM model could “hindcast” to the “Younger Dryas” period.
I don’t know if that’s even a possibility, but that’s what I’ll be looking for in the coming weeks—How far back in time can these models be run in practice, and what do they say about Earth in the 12,000-10,000 BC time frame? For me, at least, this will determine whether I believe their future predictions. I’ll dig further and report back what I find.
Regardless of the precision of any predictive model, this data set alone emphasizes that we have to do something, something significant, to control carbon dioxide in our atmosphere. That’s the crux of “healing”, isn’t it? We’ve spent a few centuries injuring our planet with carbon dioxide. Now we have to figure out how to repair the damage. So I think that’s where I’ll spend the next few issues—what can we as humans actually do to heal the Earth?
In coming issues, I’ll explore the correlation between carbon and biological activity, a thread I began in “5. Leading with Data."
A little dessert for thought.
The opening quotation pointed out that history is important to understand where we are going as a species. But, in the context of climate change, it is irrational to educate everyone. Most humans are preoccupied with their own lives. As Kurt Vonnegut wrote:3
Flato, G., J. Marotzke, B. Abiodun, P. Braconnot, S.C. Chou, W. Collins, P. Cox, F. Driouech, S. Emori, V. Eyring, C. Forest, P. Gleckler, E. Guilyardi, C. Jakob, V. Kattsov, C. Reason and M. Rummukainen, 2013: Evaluation of Climate Models. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Vonnegut, K. (2009). Bluebeard, the Autobiography of Rabo Karabekian (1916–1988). United States: Random House Publishing Group. p. 100. See Wikipedia for characters in context.
