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

From Ice to High Seas

My project ice2sea is currently in the news. For example, the BBC: ‘Best estimate’ for impact of melting ice on sea level rise.

Ice2sea is an enormous project, involving 24 institutions (such as universities and meteorological institutes), which were mostly in the EU but also further afield. Our aim was to improve observations and modelling of continental ice – the Greenland and Antarctic ice sheets, and the world’s glaciers and ice caps – to improve understanding about their behaviour in the recent past and projections of how they might change in the future. The main aim was to support the Intergovernmental Panel on Climate Change (IPCC) in assessing current understanding of sea level change. Ice2sea is an umbrella over many areas of research, which makes it difficult to summarise the results in a few quotes.

The reason we are in the news now is because we are having our final project meeting at the Royal Institution (Ri) in London this week. Many of the scientists involved in ice2sea, from across the EU and beyond, are gathering together to give talks on their work and discuss the results. Three of us are also giving a free public talk tomorrow at 7pm.

Publications and collaborations will continue long after the end of the project. We have a legacy of improved models, robust observations, and new scientists. I’m certainly planning to continue working with many of my ice2sea collaborators in the future.

Given that we are meeting at the Ri for three days, and the project is slowly drawing to a close, ice2sea held a press conference yesterday morning explaining what the project is, what our aims were, and what we feel we’ve achieved. Most of this focused on the benefits of this kind of project – the international collaboration, exploiting expertise from different countries, and the coordinated study of all aspects of continental ice.


We’ve produced a summary document, “From Ice To High Seas” that explains the science of continental ice and sea level to a general audience and highlights the individual publications that we’ve produced. It describes ice2sea papers that have been published in journals but also those that haven’t yet completed peer-review. So the document is a summary of our research as of April 2013, but the final results may change as these papers work their way through the system.

There are two sets of numbers from our summary document that have been highlighted by the media. I’ll explain where these come from:

“For that one scenario we have an ice sheet and glaciers contribution to sea level rise of between 3.5 and 36.8 cm by 2100,” – David Vaughan

“They concluded there was a one in 20 chance that the melting ice would drive up sea levels by more than 84 centimetres, essentially saying there’s a 95% chance it wouldn’t go above this figure.” – Matt McGrath

The first is a summary of the publications that make projections using models of the atmosphere, ocean, ice sheets and glaciers. The scenario David Vaughan (from the British Antarctic Survey, the ice2sea project leader) refers to is a “mid-range” emissions scenario, SRES A1B. Most, but not all, of these papers have completed peer-review. The list of papers relevant to that range is at the end of the post. We stated the minimum and maximum range of these projections in the summary document.

The second is from one publication (Bamber and Aspinall, 2013: listed below) that uses “expert elicitation”, a structured way of combining the judgement of a group of experts. This method deserves its own post, but here is an article about it by Willy Aspinall from the University of Bristol. This estimate is across all scenarios, not just one in particular.

So the first is the range of results from different plausible modelling options for the A1B scenario, while the second is an expert assessment of the probability of sea level rise given that the models do not incorporate every process and we may not follow the A1B emissions scenario.

Many of these papers will be included in the IPCC Fifth Assessment Report (AR5) being published this September; others will not, for example if they were not accepted by journals in time for the AR5 deadline. But this interim summary of our own work is not intended to “compete” with the IPCC, because that will assess all sea level research worldwide rather than only ice2sea results.

Not only that, but these papers are a small part of the results of ice2sea. We also have also led or contributed to major steps forwards in observing and modelling continental ice and sea level change, including:

We also have a special issue in The Cryosphere (open access).

Our focus has very much been on improving our understanding of the processes involved. This is a different approach to several recent studies that have used very simple assumptions or relationships to make projections for the future (such as extrapolation of past trends). Our work, such as our projections for Greenland glaciers (Nick et al.), shows that continental ice is complex, and can have both periods of rapid change and of relative stability. We can’t just extrapolate the past into the future. We need to understand the processes of ice, the detailed landscape of the bedrock underneath, and the potential range of changes to the atmosphere and oceans, to make physically-based projections of the future.

Update: some are asking about comparisons of our summary range with the previous IPCC projections (AR4).

First, the ice2sea headline results are the substantial improvements in models and observations, not our simple summary of the minimum and maximum projections.

Second, we should be cautious of comparing with AR4 land ice contributions for the same scenario (A1B), because those projections were from 1980-1999 to 2090-2099 while ours are for 2000-2100, and theirs are stated as a 5-95% range while ours don’t have a probability range attached.

Third, there’s another reason we can’t compare with the total sea level contributions: they combined the land ice and thermal expansion contributions using an assumption that they are independent, but we haven’t repeated this analysis.

There is a quote from David in the BBC article about our high end estimate being about 10cm higher than AR4, but this was a spur-of-the-moment, back-of-the-envelope comparison (in response to a press request) with the best-known AR4 range which includes all scenarios, just to provide some context. It’s not meant as a definitive, like-for-like comparison. Even if he had said about 17cm higher, the difference for the A1B land ice contributions, it would still have been a ballpark figure because of the different definitions of the contribution. That is why we haven’t made that comparison in the summary document.

In short, do wait for the AR5 in September for consensus sea level projections and how they compare with AR4. For ice2sea, please have a look at individual publications, and the summary document, but bear in mind that we are not trying to do the job of the IPCC: rather we have spent considerable effort trying to make their job easier, by providing science reported in individual publications.


Bamber and Aspinall (2013) “An expert judgement assessment of future sea level rise from the ice sheets”

Barrand et al. (2013), “Computing the volume response of the Antarctic Peninsula ice sheet to warming scenarios to 2200”

Fettweis et al. (2013), “Estimating the Greenland ice sheet surface mass balance contribution to future sea level rise using the regional atmospheric climate model MAR”

Fürst et al., in review.

Giesen and Oerlemans (2013) “Climate-model induced differences in the 21st century global and regional glacier contributions to sea-level rise”

Goelzer et al. (2013), “Sensitivity of Greenland ice sheet projections to model formulations”

Payne et al., in review.

  1. There was some confusion at the press briefing over the 1/20 risks but the report gives a more detailed projection on sea level rise risks than we have had before and also the regional variations which are bad news for some Pacific Islanders. We are getting a clearer picture of the actual risks we face.

  2. This does not foster good science communication and help to inform the public. The BBC reporting was particularly poor. For example, here is the headline that is all that most people will ever read:

    “Ice melt, sea level rise, to be less severe than feared – study”

    What journalists and editors needed was a clear statement of the new projections of total global SLR range, with the new estimates of the potential contribution from land ice mentioned in the same sentence — or at least the same paragraph. To its credit the Reuters report had something on the impacts of the “less severe” SLR of “16.5 to 69 cm”. While editors and journalists may have been confused by this statement:

    ‘They suggest a contribution from continental ice of 3.5-36.8 cm to global mean sea-level rise to the year 2100 for a “business as usual” mid-range emissions scenario.’

    — they may have missed the “contribution from continental ice” part and jumped directly to the “3.5-36.8 cm” part, ignoring the crucial following text: “To obtain a projection of total global sea-level rise, other contributions, not explicitly addressed by ice2sea, must be added” —

    it is up to scientists to make sure that the findings are unambiguous and that they are placed in context. All that most people will remember from this is “Ice melt, sea level rise, to be less severe than feared”.

  3. Hi melty,

    I agree it has been confusing, often because people are comparing with different things.

    The “less severe” refers to several recent estimates since AR4 which were much higher than AR4 (e.g. by Pfeffer, Rahmstorf). The ice2sea results are broadly consistent with, or a bit a higher than, AR4 – they may be a bit higher because we estimate the response of dynamical processes to future climate change while AR4 didn’t.


  4. The way in which this news story was announced leaves a lot to be desired.
    There has been much criticism of ‘science by press release’, but what happened here seems to be worse. It seems that ice2sea had a press briefing sometime yesterday, and press stories started coming out yesterday afternoon. These varied widely in tone, from the calm Reuters “Ice melt, sea level rise, to be less severe than feared – study” to the hysterical Independent “Floods could overwhelm London as sea levels rise – unless Thames Barrier is upgraded”.
    There was no way of knowing which if either of these was an accurate representation, because ice2sea did not produce any announcement until this morning.

    All four of the UK broadsheets have gone for scaremongering headlines about flooding in London:
    “Floods could ‘overwhelm Thames Barrier by end of century'” Guardian
    “Sea levels rise could mean floods in London” Telegraph
    “Risk to London from rising seas ‘worse than feared’” Times
    “Floods could overwhelm London as sea levels rise – unless Thames Barrier is upgraded” Independent.
    This is strange, since the one-page press release on the ice2sea web page doesnt say anything about the Thames barrier or London flooding (except to mention a talk later this week). This suggests that the press briefing was quite different from the written press release (something that has caused controversy before). Two of the headlines seem to be quoting something, but it’s not clear what, since neither ‘overwhelm’ nor ‘worse than feared’ appear in the website press release or the 50-page pdf document.

  5. Paul, I agree with much of what you’ve said. Let me take this apart bit by bit. It’s almost worth a main post update, but I think that’s already quite long.

    “The way in which this news story was announced leaves a lot to be desired….There was no way of knowing which if either of these was an accurate representation, because ice2sea did not produce any announcement until this morning.”

    I agree: we [as a project] should have put the press release and summary document online before/when the embargo lifted rather than waiting 14 hours or so. Our bad. I can’t say whether I could or should have predicted this problem – I only got involved in things quite late last night when I realised (partly from you and Gavin Schmidt) that this was becoming an issue. I had assumed the press release and document were going online at 6pm, but I hadn’t asked anyone about it.

    “All four of the UK broadsheets have gone for scaremongering headlines about flooding…This is strange, since the one-page press release on the ice2sea web page doesnt say anything about the Thames barrier or London flooding. This suggests that the press briefing was quite different from the written press release (something that has caused controversy before).”

    Logically, there is another possible alternative which is that the media reporting was different to the event. I wasn’t at the press conference so I don’t know exactly what was said.

    I think the origin of this “overwhelm” story is the 84cm expert elicitation (1 in 20) estimate, added to the upper end of thermal expansion from AR4 (32cm), combined with a graph David Vaughan often shows from Proudman Oceanographic Laboratory that shows a 1m sea level rise for London could alter the storm surge statistics such that a 1 in 1000 year event becomes a 1 in 12 year event.

    I agree that two headlines seem to be quoting someone. But the Guardian doesn’t attribute “overwhelm” to a name. Knowing the scientists on the press briefing panel, I do not imagine any of them using the word “overwhelm” (I don’t know if they did). Over-top, maybe, if they are talking about possibilities. I can’t see if the Times attributes that quote because I don’t subscribe, but the same applies – we all see this is as “less than feared” with respect to the post-AR4 studies, not worse.

    At every stage ice2sea scientists have tried to emphasise the scientific advances, rather than the summary of projections. We’ve also tried to emphasise that the projections are not too dissimilar to AR4 and within the envelope discussed in “Thames Estuary 2100” (TE2100) i.e. the Environment Agency are already aware of that level of estimated risk and feel they can manage it.

    It’s impossible for me to be 100% sure where these stories have arisen because I wasn’t at the press conference, it was 1.5 hours long, we don’t have our own recording of it, and there have been other interviews conducted with individual media outlets too. But the website press release and the summary document are the best reflection of the results of the project.

  6. Tamsin, I think the attack on the media headlines rather exposes a lack of understanding of how the media operates. It is hardly surprising that UK news journalists working for the mainstream media chose to interpret your technical results for their audiences in terms of the potential impacts on the UK. As you know, scientific papers tend to avoid any discussion of the implications for human society, which is the primary interest for journalists and their audiences. that leaves the journalists to try to fill in the gaps by asking the authors about implications at the press conference. Were there any questions at the press bruiefing about whta these results might mean for the UK?

  7. A question about glaciers after a recent trip to Glacier Bay that seems to be covered by the project. What caused the large observed retreat in glaciers from about 1760 to about 1880 and continuing retrest after that? What do the models show as the main causes?

  8. I think the “less than feared” headlines are unfortunate. It’s a case of comparing apples to oranges which none of the mainstream coverage has made clear. Personally, I put the post-AR4 semi-empirical studies in a different category to the process based ice2sea approach. To my mind they are relatively immature and don’t represent the community’s best estimate. As such they shouldn’t be compared in the way they have.

    My take home is that the ice2sea results suggest a sea level rise envelope a little higher than AR4, increasing the risk. However, we do now have more confidence that dramatic dynamic events are unlikely, at least before 2100.

    Another key sentence from the press release is: “For the period after 2100, sea levels will continue to rise, initially at an accelerating rate, for many centuries.”
    For many of our major coastal cities, which measure their histories in many centuries this is serious. London has been there for a couple of millennia, it seems likely we can adapt to 2100 sea level. But what about 2200 or 2300? In historical terms, are we in the final 10% of London’s existence as a thriving city?

  9. Hi Bob,

    I think “attack” is unfair. I pointed out Paul’s logical error, the lack of attribution of a quote, and said those scientists don’t usually say those things in those ways; but also said I wasn’t there so I don’t know exactly what was said.

    And of course I understand how things work in the media. It’s just interesting that the same story, the same press briefing, was “sold” differently by different papers. It depends whether they are comparing to AR4 (“worse than feared”) or post-AR4 (“less severe than feared”) projections. Carbon Brief wrote a post on exactly this point of confusion.

  10. Another reply Paul, where I don’t agree with you – the press release does mention regional sea level and storm-surge events, and the summary document does mention the Thames Barries and London, so these aren’t too different to the press briefing/coverage.

  11. You are right that there is a confusion. But I am not convinced that journalists are to blame for it. It seems to me that the research community has been very bad at communicating this information, even though it is complicated.

  12. Sorry Bob. Doing the best we can. I think my post has cleared up quite a few questions, but let me know if you have more.

  13. Rob, this may be helpful:

    Mann, D.H. and Streveler, G.P. (2008) Post-glacial relative sea level, isostasy, and glacial history in Icy Strait, Southeast Alaska, USA. Quaternary Reesearch, 69, 201-216.

    From the discussion:

    “Correlations with other glaciers in the region suggest that climate changes triggered these terminus fluctuations, which in turn altered isostatic depression and hence RSL. Nonetheless, the progressive down-fjord expansion of glacial ice in Glacier Bay over many millennia is consistent with the main driver of this glacier system being the dynamics of its terminus. The terminus dynamics of calving glaciers involve feedback interactions between the rate of terminal moraine construction, water depth at the grounding line, iceberg calving, glacier flow rate, and RSL (Meier and Post, 1987, Post and Motyka, 1995 and Van der Veen, 1996). As in other calving glacier systems (Mann, 1986 and Powell, 1991), the effects of climate change on the overall glacier system become slaved to the slower, larger-scale dynamics of the glacier’s terminus, which may then modulate the system’s overall responses to climate change. If the terminus is securely anchored on a terminal moraine shoal, climate change can cause only a brief and modest retreat or thinning of the glacier; however, if the terminus lies in an insecure position, as at a fjord mouth, even minor changes in climate can potentially trigger a fjord-emptying, catastrophic calving retreat. In the case of Glacier Bay, it was only after the glacier and its terminal moraine shoal entered Icy Strait ca. AD 1750 that the glacier became susceptible to catastrophic calving retreat, which in this case proved irreversible before the LIA ended ca. AD 1880.”

  14. “which in this case proved irreversible before the LIA ended ca. AD 1880”

    Thanks, I guess my main question related to this is why did the LIA end ~1880. I assume some combination of lower precipitation and higher ocean/air temperatures during the 18th/19th centuries must have been a large driver of such a significant glacial retreat.

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