Hothouse Earth and Heatwaves

This episode, in response to the Steffen et al. paper of 2018, was first published in September 2018, and can be accessed here.

This is a mini episode in a format I’m hoping to try out a little more often, to keep you updated on the news and latest developments. Every so often, I get fascinated by a couple of science and technology stories that make the rounds, and I want to tell you about them and provide a perspective on what’s going on — and hopefully spark a discussion about the stories with you! It’s going to be more informal and shorter than a standard show. And, as with all podcasting experiments, highly irregular. But you should be used to that by now.

So the story I wanted to discuss at the moment is a new climate science paper that’s got an awful lot of media attention over the last few days — partly because it fits into the cultural zeitgeist in Europe and North America, where we’ve just had an extremely hot summer.

Incidentally, those of you wild enough to follow me on Twitter @physicspod will know that I got kind of annoyed about the coverage of this wave of heatwaves. For the first few days, there were endless comparisons here in the UK to the UK heatwave of 1976 — but if you look at a temperature map, you’ll see the two events are really nothing alike. The UK heatwave of 1976 was more or less isolated over the UK and the surrounding area. This heatwave, the 2018 heatwave, covered practically the entire surface of the globe.

The problem is that the link between extreme weather and climate change can be very difficult to explain — because we like to think of things in terms of cause and effect, whereas — particularly with extreme weather events — we should think about them in terms of probability distributions.

Let’s start super basic with weather and climate. Weather is essentially the day-to-day or week-to-week conditions of temperature, precipitation, and so on. Climate is the average of weather over many many seasons. The climate of the UK is generally very different to the climate of Brazil — even though you could probably pick a handful of days where the weather was actually surprisingly similar; and the same is true even between places like Britain and Italy, for example.

But the climate of a particular region obviously determines the kinds of weather that you’re likely to see. Here in the UK, a summer’s day that reaches 33 degrees Celsius happens once in a blue moon. In Florida, which has a different climate, this kind of weather is far more likely to occur.

One way to think about it — especially in terms of extreme weather events — that I’ve found helpful is a bell curve. If you plotted the temperatures for, let’s say, all days in July across the UK — you would find that it formed a bell curve, with the average at (say) 20C. Maybe the bell curve stretches out for 9C either side — such that getting an 11C day in July or a 29C day are both uncommon, but possible.

Climate change shifts the average of that bell curve. In fact, you can see from NASA’s temperature records that over land in the Northern hemisphere, since the 1960s, that bell-curve has shifted to the right by around 1 degree Celsius. In other words, maybe now the average July day is 21C.

But the extremes shift further. In fact, again, according to the NASA GISS temperature records, they’ve shifted by two degrees Celsius. So climate change over the last 50 years has meant that “uncommon, but possible” warm summer days have shifted from 29C to 31C in our example.

The problem is that, for most weather events, you can’t definitively say that climate change “caused” that event to happen. Even in a world without climate change, it might still be possible to get a freak event where we see the same global heatwaves. Scientists — who I work with — who study extreme weather attribution would never look at the heatwave and say “Is this *caused by* climate change?” because, for most weather events, you can’t give a simple yes/no answer.

Instead, what they tend to do is run many hundreds of simulations on climate models that are then rescaled by the actual, historical temperature data. In other words, they’ll run ten thousand simulations of the world without human influences, and ten thousand with human influences. If you see 500 extreme weather events in the world with climate change, and only 100 in the world without climate change, then you’ll be able to make a quantitative statement like “Climate change made this extreme weather event five times more likely” or “Events like this should now occur once every ten years rather than once a century.”

That’s not the same as causation. But, of course, eventually, it becomes as near to “caused” as makes no odds. Let’s say I drink forty triple vodkas and then go out for a drive, and I get in a car accident. If I stood up and argued in court that the accident wasn’t caused by my decision to get hammered, because there’s a non-zero probability that I would’ve got into an accident anyway, then I would probably be laughed out of the room. Beyond a certain threshold of loading the dice, we’re happy to call the game fixed.

If you can define that threshold, you can eventually start to say “this weather event was so ridiculously unlikely without human influence, that it may as well have been caused by climate change.”

There is a serious tightrope balance for everyone who’s concerned with climate change to walk here. Because the most devastating effects of climate change will be the increase in probability and extremity of severe weather events. When a severe weather event like the heatwave we’ve just had happens, we naturally want to draw attention to the fact that we need to do more to adapt, because the science tells us that these events are getting more and more frequent — and, if we do nothing and don’t address climate change, worse things will lie in store. But we have to be extremely careful not to say “this was caused by climate change”, because people tend to hear that as “now thanks to climate change this will happen every year” — and enough people are either baffled by the difference between weather and climate, or else pretend to be, that reinforcing this even by accident is a bad idea.

Nevertheless, I was mad with the early news coverage of the heatwaves and the wild-fires in Europe, North America, and Japan. At one point, the BBC News frontpage — for listeners abroad, the BBC is constantly being accused of having a liberal political bias — had four or five extreme weather related stories about heatwaves or wildfires, and not one even mentioned climate change.

I think all such articles should feature *somewhere* in the main body, even if it’s not the headline, that “Many scientific studies have shown that extreme temperature events are made more likely by anthropogenic climate change.” That’s a perfectly scientific thing to say. In fact, it’s obvious that if you increase the global average temperature, then the extreme temperature events will become more likely — in the same way that if you add one to every number on your dice, you’ll be more likely to roll greater than a 3 and escape from the dragon in your D+D game. It is in no way alarmist, or overstating the case; it’s just providing vital context to the story.

Eventually, the BBC and other outlets did step up their game and start to mention the impact of climate change on extreme weather — although, frankly, some of them then went too far again. Whenever you see a scientific story in the news, particularly in the tabloid press, just remember that their key goal is to garner clicks and sell papers: informing you accurately comes second, and for many outlets, sadly, it’s a very distant second. I’m reminded of a paper from years ago which explained that the most likely temperature increase from doubling carbon dioxide was around 3 degrees Celsius, but that you couldn’t rule out values as low as 1.5 degrees Celsius or as high as 11 degrees Celsius based on the models that they were using. (We’ve now narrowed it, somewhat, to between 1.5 and 4.5, according to the IPCC.) If you read the paper, it was clear that 11C was extremely unlikely, just barely within the realms of physical possibility: yet, of course, this was the number that got splashed all over the front page of the newspapers as if that had been the main prediction…

So it’s in this context, of trying to frame climate-related news stories in an accurate way in a world that’s more focused on clickbait than accuracy, while still conveying the gravity of the situation and the need to take action, that I want to talk about the other big climate-related story this week: Hothouse Earth.

You might have seen lots of news articles on this subject: here’s the BBC’s take:

“Researchers believe we could soon cross a threshold leading to boiling hot temperatures and towering seas in the centuries to come.

Even if countries succeed in meeting their CO2 targets, we could still lurch on to this “irreversible pathway”.

Their study shows it could happen if global temperatures rise by 2C.

According to the research paper, crossing into a Hothouse Earth period would see a higher global temperature than at any time in the past 1.2 million years.

The climate might stabilise with 4–5 degrees C of warming above the pre-industrial age. Thanks to the melting of ice sheets, the seas could be 10–60 metres higher than now.

Essentially, this would mean that some parts of the Earth would become uninhabitable.

The impacts would be “massive, sometimes abrupt and undoubtedly disruptive,” say the authors.

The only upside, if you can call it that, is that the worst impacts may not be felt for a century or two. The downside is that we wouldn’t really be able to do anything about it, once it starts.




This arises from a paper from Steffen et al. which is open access — anyone can read it — called Trajectories of the Earth System in the Anthropocene.

So there are a couple of things to point out here right off the bat. First of all, this is a “research perspective” — essentially a literature review. It doesn’t contain any particularly new science — it’s not some new study with a new and improved climate model, or some additional data that demonstrates that the Earth is tipping into this Hothouse regime. It is a perspective — one endorsed by lots of scientists who are way smarter than me, who have synthesised and read lots of different information, but nevertheless a perspective which draws together lots of other scientific papers rather than a new study. Think of it as an overview of the science with a salt-shaker of opinion added in.

That’s why saying things like “this study shows” is a little misleading. The study *suggests* that, given the evidence we have at the moment, this *may* be a possibility. But it’s more speculative than this might suggest.

So let’s talk about that evidence. The idea here is all based around climate feedbacks. We’ve discussed this before, in our episodes on climate change. The basic idea is that the climate is driven by all kinds of things — solar activity, volcanoes, human activities like emitting aerosols and greenhouse gases — that kind of thing. Yet it is a complex system with many moving parts that can influence itself. Feedbacks occur when changes to the climate cause other changes that reinforce, or act in the opposite direction, to the initial forcing. So, for example, emitting carbon dioxide traps infrared radiation, which directly acts to heat up the Earth, including the Arctic Circle. That’s the direct forcing. But heating the Arctic also causes sea ice to melt. Sea ice naturally reflects more of the solar radiation back to space than open ocean — it’s brighter and more reflective. So when this ice melts, it means more of the solar radiation gets absorbed by the darker ocean. This is one of the reasons why the Arctic circle heats faster than anywhere else on the globe. The warming due to this secondary effect — melting ice, which reflects less sunlight, which heats the arctic more — is a feedback called the ice-albedo feedback.

We’ve already discussed in those old episodes, too, an example of a pretty stable climate equilibrium — it’s called the Snowball Earth — which arises due to precisely this feedback. This has happened several times in Earth’s distant history. Something — maybe a supervolcano — caused a sudden, rapid cooling. Ice sheets expanded from the poles, and they in turn reflected more sunlight, which caused the Earth to cool further. Eventually, the Earth can be entirely covered with ice under this scenario — until you get another big kick, for example from bacteria that increase carbon dioxide levels in the atmosphere, to reverse the trajectory again.

So we know from Earth’s history that it’s perfectly feasible that you can kick the climate system into a position where these feedbacks take over, and in fact become the dominant effect. At this point, although your direct forcing is still important, the climate feedbacks that result can push you the rest of the way. So it might be that cooling the planet by 3–4C would be enough to trigger a snowball Earth, which would run away with itself and result in much more cooling overall.

What the scientists of the Hothouse Earth are proposing is that there might be another stable point that feedbacks can lead us to — another state of the climate that’s relatively robust, like the snowball Earth — but, instead, it’s 4–5C warmer than the world we live in today. And their argument is that, perhaps, a warming of 2C — in other words, where we’ll be at mid-century without any major action — might be enough to trigger one feedback, which then triggers another, in a disastrous domino-style effect that kicks the climate on this new trajectory towards this new, warmer, Hothouse Earth.

Let’s talk about some of the feedbacks they mentioned. One, we’ve already discussed: ice-albedo feedbacks. The Arctic summer sea ice is smaller than it’s ever been on record. This is leading to heating of the Arctic circle. Perhaps, as the summer sea ice shrinks, it might destabilise the ice sheet over Greenland, which would in turn heat the planet more, resulting in a tipping point for the Antarctic ice sheet — which is currently pretty stable for a variety of different reasons, give or take a few major glaciers — which would then in turn start to melt.

Another class of feedbacks that were considered are carbon cycle feedbacks. Currently, more than half of the CO2 we emit is absorbed by the land (plants) or the oceans; it makes some of the plants grow quicker, and makes the oceans more acidic, but doesn’t go into the atmosphere and impact climate.

As the world warms, there’s a risk that forests — such as the Amazon rainforest or the Boreal forests in high Northern latitudes — would start to die, and may even start to become a net source rather than a net sink of carbon dioxide. In other words, not only might the land and ocean become less effective sponges for CO2, but they might actively start emitting extra CO2 to the atmosphere on top of what humans already do — a feedback that would clearly make things worse. Similar concerns arise due to the respiration rates of bacteria in the ocean; in a more acidic and warmer ocean, these bacteria might respire more and emit more carbon dioxide to the atmosphere. The circulation of the oceans is driven by temperature and by salinity, or salt concentrations. In a warmer world, where freshwater from Greenland or glaciers dilutes the ocean, that circulation can change. Water has a much greater heat capacity than air; this will be clear to anyone who’s ever left a bowl of hot water lying around. This means that the oceans, thus far, have taken up around 95% of the additional thermal energy that’s kicking around in the climate system due to human influences. But it may be possible that, in the future, if the ocean circulation changes, the oceans will become less effective at taking up heat — perhaps they will circulate this heat energy down to the deep ocean more slowly. If that happens, then more heat may end up in the atmosphere.

And, on the more exotic side of things, there is permafrost — permanently frozen ground — up in the Arctic circle which contains an awful lot of methane, a potent greenhouse gas. If that permafrost thaws, the methane could be released, which would lead to rapid and sudden warming. A few years ago — and, today, amongst a group of deeply concerned people — it was the green apocalypse. Recent studies suggest that it might be more of a slow burn that takes decades or centuries to fully arise, rather than a sudden cataclysm.

You’ll notice a lot of the following in these previous descriptions: “may”, “might”, “could”. The reality is that to fully understand all of these climate feedbacks requires a huge amount of science, and a great deal of expertise in fields as diverse as the study of peat soils, the biogeochemistry of the oceans, and how bacteria respire.


What’s more, we’re talking about conditions that the Earth has never seen before. When we cross the 2C Paris threshold, it will take us out of the range of temperatures during which human civilization has ever existed — into a world we haven’t seen for hundreds of thousands of years. The whole range of temperatures for the last geological epoch has been around +/- 1C. We’re already moving out of what the world has known of late.

So what do we really have here? We have plenty of very real feedback mechanisms that might kick the planet onto a new trajectory. But it’s very difficult to quantify how bad they might be, or when they’d “kick in”. And we also know that in the past, paleoclimate records have shown rapid warming events that resulted in the Earth being 4–5C warmer than it is today. Much of that may have been due to feedbacks. But the climate is in a totally different state today. The feedbacks themselves could operate differently. We do not have very many reliable data points for how the Earth will respond to something that’s never happened before. And, unlike in other branches of science, we don’t have miniature Earths to do experiments on.

This is where we get to the framing question, and why I really wanted to talk about this particular paper. For all sorts of reasons — many of them political — I can understand why authors and journalists want to say “If we fail to hold to the Paris limit on temperatures, then all of these feedbacks will kick in, and the world will warm by 4–5C, and there’ll be no stopping it.” They may well be right. The 2C Paris limit is important because it represents uncharted waters for a climate setup like the one we have today. Here be dragons, and here, maybe, be accelerating feedbacks. It would be madness not to warn people — especially when we’ve spent more than thirty years warning them, during which carbon emissions have risen pretty much every year save for the Great Recession. In some ways, though, this paper — rather than necessarily giving new information — reminds us of all of the risks we still can’t be sure about.

What concerns me is drawing thresholds in climate science. I appreciate that a domino effect could be possible. We know that some feedbacks, like the forest diebacks and the Arctic sea ice, are likely to be triggered before others, like the permafrost, Greenland, and the Antarctic icecap. So it makes perfect sense to say that, perhaps, getting to 2C triggers some feedbacks at 2C that then push us to 3C where more feedbacks will be triggered in turn.

But we cannot allow this to be phrased as “if we miss the Paris Agreement, it’s all over.” Because we simply do not know that this is the case. This paper in no way conclusively proves that; it just points out that feedbacks exist, and they aren’t represented very well in our climate models, and they could cascade over timescales of decades or centuries if we don’t take action. It doesn’t show that, after 2C, it’s all over and you might as well burn all the oil you want.

And, furthermore, it seems much more likely that *all* of these feedbacks basically just get worse and worse the further from our nice, stable, Holocene climate that we drift. In other words, it’s not like you trigger five feedbacks at 2C suddenly that lock you into a certain amount of warming. Rather, the more carbon dioxide we emit, the more feedbacks we’ll trigger, and the worst the outcome is likely to be. A 4C world has more risks than a 3C world which has more risks than a 2C world, and so on. Targets and goals give us something to aim for, and something concrete to agree upon and build plans around. But our ultimate goal should always be to reduce our dependence on fossil fuels — and, more broadly, a way of life that does inevitable and irreversible damage to the environment — as quickly as we possibly can. We’re driving, a little drunk, at top speed, in the fog. No one knows how near the cliffs might be. That’s hardly a good reason not to slow down.

So this was the rather lengthy minisode. I urge you to read up on the Steffen et al paper, and see how your favourite news outlets have covered it. As we learn more about feedbacks, we’ll hopefully get a greater understanding of how likely a Hothouse Earth is to happen in reality. But the risk is always worth bearing in mind. When you move into unprecedented territory, there’s a very real danger that things will spiral out of our control altogether. The faster we act, the less likely that is to happen.