Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS All Models Are Wrong

How soon will the ‘ice apocalypse’ come?

I’m very happy to return to blogging here at PLoS after a long…pause. I talk about why after this post, which was originally published by the Guardian.


An emotive article on the ‘Ice Apocalypse‘ by Eric Holthaus describes a terrifying vision of catastrophic sea level rise this century caused by climate change and the collapse of the Antarctic ice sheet. But how likely is this – and how soon could such a future be here?

I’ve been gripped by the story of Antarctic ‘ice cliff instability’ ever since Rob DeConto and Dave Pollard published their controversial predictions last year. They suggested disintegration of ice shelves caused by global warming could leave behind coastal ice cliffs so tall they would be unstable, crumbling endlessly into the ocean and causing rapid, sustained sea level rise.

I’m glad Eric is writing about an impact of climate change that is both certain (seas will rise around the world, no matter what we do) and incredibly important (we must adapt). I’m sympathetic to his concerns about the future. But I think his article is too pessimistic: that it overstates the possibility of disaster. Too soon, too certain.

I’ll admit my fascination is personal. In late 2015, Catherine Ritz and I co-led a study about the future of Antarctica, and our predictions – though definitely bad news – were far less dire. So what evidence do we have that the Antarctic ice sheet could, or would, collapse this century?

First, it’s important to map out what is known. Yes, the Antarctic ice sheet is vulnerable to global warming. Not because it will melt in the warm air like an ice cube, but because the ice shelves around the coastlines act as bracing for the glaciers. When the surface of an ice shelf melts it can disintegrate, and the flow of ice into the oceans can speed up: we saw this for Larsen B in 2002. Our greenhouse gas emissions make this more likely.

Yes, scientists have proposed that disintegration of ice shelves could trigger two types of ‘instability’ that may have caused the ‘marine’ parts of the Antarctic ice sheet (lying on bedrock below sea level) to drastically reduce in the past, raising sea levels by several metres. Confusingly, these are named ‘marine ice sheet instability’ and ‘marine ice cliff instability’. Broadly speaking, both are based around the idea that the thicker the edge of the ice sheet, the faster ice can be lost into the oceans. Both may be ‘positive feedbacks’: in other words, self-sustaining. The marine parts of the ice sheet are thicker in the middle than at the edges, so once these proposed instabilities started, they would keep going until they ran out of ice. As Eric says, this would be over three metres worth of sea level rise.

Yes, one instability may already be happening. In 2014, new evidence suggested marine ice sheet instability may be underway in the Amundsen Sea area, forty years after it was first predicted. But Eric is wrong to say Antarctica’s ‘ice budget’ has tipped out of balance due to our burning of fossil fuels. Not only has it been out of balance before – such as the ancient West Antarctic collapse that causes concern – but the reason for the Amundsen Sea changes, where most ice is being lost, is that the ring of deep warm water around Antarctica has welled up onto the continental shelf and is melting the ice from underneath. We don’t know if human activities made this more likely.

As far as I know, scientists agree the basic theory of marine ice cliff instability by Jeremy Bassis and Catherine Walker is sound (it’s a shame Catherine’s joint contribution to that key 2011 paper wasn’t mentioned). Ice is not strong enough to form a sheer cliff taller than about 100 m above the water line. And indeed, we don’t see any ice cliffs taller than this, which gives us indirect evidence their calculations are correct. We also have the recent study led by Matthew Wise of marks left by the keels of ancient icebergs, scraping along the bedrock like heavy ships, which point towards this kind of cliff collapse in the distant past.

But there is little consensus in the scientific community about how this ice cliff instability could behave. That’s because there is a big leap from identifying a potential problem and predicting the real world consequences. Iceberg scrape marks, though important, can’t tell us how soon, how fast, or for how long. Rob DeConto and Dave Pollard have been the first to put their heads above the parapet, but others may come to different conclusions. Ted Scambos mentioned ‘piles of icebergs’ acting as new ice shelves. Could cold freshwater from melted icebergs also slow things down? Could cliffs partially, instead of catastrophically, crumble? Would a model with finer detail predict fewer tall cliffs? Eric describes the collapse rate as conservative compared with Jakobshavn in Greenland, but this compares apples with oranges: the rapid ice losses from that great glacier are mostly due to its unusually fast speed, rather than the rate of retreat of the ice edge.

Just as important is the initial trigger: how sensitive are the ice shelves to global warming? In Rob and Dave’s study, they disintegrated fast and early: predictions by others are less pessimistic. Eric compounds this by describing their highest scenario as ‘business as usual’, but it’s really business-worse-than-usual. Under current policies we’re headed for less warming than this scenario, and if national pledges under the Paris agreement are carried out this would decrease the warming more (though not meeting the two degree target).

Even Rob and Dave had a huge range of predictions for that very high scenario – anything from 30 cm sea level fall to nearly two metres rise – and they originally described their model as ‘speculative’. So it’s not justified to use such strong, certain phrases as ‘the destruction would be unstoppable’.

I was particularly concerned about some of the implied time scales and impacts. That ‘slowly burying every shoreline…creating hundreds of millions of climate refugees…could play out in a mere 20 to 50 years’ (it could begin then, but would take far longer). That ‘the full 11 feet’ could be unlocked by 2100 (Rob and Dave predicted the middle of next century). That cities will be ‘wiped off the map’ (we will adapt, because the costs of protecting coastlines are predicted to be far less than those of flooding). We absolutely should be concerned about climate risks, and reduce them. But black-and-white thinking and over-simplification don’t help with risk management, they hinder.

Is “the entire scientific community [in] emergency mode”? We are cautious, and trying to learn more. Climate prediction is a strange game. It takes decades to test our predictions, so society must make decisions with the best evidence but always under uncertainty. I understand why a US-based climate scientist would feel particularly pessimistic. But we have to take care not to talk about the apocalypse as if it were inevitable.


Since becoming a lecturer (i.e. a permanent job, with a fair amount of teaching) in late 2014, I’ve found it hard to find time to blog. Teaching and research have taken precedence, along with trying out a few new media outlets…

I wrote two pieces on reaching 400 parts per million of atmospheric carbon dioxide concentration:

One climate data point. For ‘Shaping the Future: 30 years of science and engineering’, Campaign for Science and Engineering (CaSE), November 2016.

The 12 key science moments of 2016. The Observer Tech Monthly, December 2016.

I talked about the DeConto and Pollard study to Adam Levy on the Nature podcast (Mar 2016), and then again in the context of making sense of uncertain climate predictions and the ‘pause’ in warming to Adam Rutherford on BBC Radio 4 Inside Science (May 2016, where we both nearly get the giggles). I also talked on Inside Science about whether good explanations of science make people stop listening (Jan 2017) and about Trump and the Paris agreement (Jun 2017). I had a fun podcast interview by Anthony Sharwood for the Huffington Post Australia (Jun 2017) – scroll down to ‘Breaking the Ice’, a series that aims to show the human side of climate scientists. I’ll let you decide whether he succeeds with me…


I’m writing this post while between jobs. Last Friday was my last day at the Open University, and from the 1st December I’ll be a lecturer in Physical Geography at King’s College London. I’m sad to leave wonderful colleagues, and an inspiring social mission to make education more accessible to students and the public. In three years I learned a huge amount about how to teach, and had the support to establish my research and public engagement too. Thanks so much to all who helped me there. I haven’t finished my relationship there: watch this space.

I’m now excited to join new welcoming colleagues and enthusiastic faces in the lecture theatre. Wish me luck…

  1. Very happy to see you back in the blogosphere.

    Thanks for injecting a note of sobriety into the catastrophe sweepstakes.

    Other than noting that the Grand Holocene Melt isn’t yet finished, it would seem that there is little in the way of hard data to inform our policy decisions. Glaciers don’t appear to be on rails easing their way to the sea–even if obstructions are cleared. The warm oceans eating away at the bases of projecting ice outflows are more subject to change than perhaps we admit.

    I guess the motto is keep an eye on this space–for another twenty years or so…

    Hope you can find time to post more often.

  2. I don’t think we need worry about over warming. All today’s non political science points towards upcoming cooling including ice formation at the poles. The Modern Warming Period has plateaued and it’s nothing to do with CO2

Leave a Reply

Your email address will not be published. Required fields are marked *

Add your ORCID here. (e.g. 0000-0002-7299-680X)

Back to top