How soon will the ice apocalypse come? (CCCR2018)

thank you so much for inviting me it's a real pleasure to see the breadth of topics here in the and the interesting climate change I'm a climate scientist with a particular focus at the moment on sea-level rise which I'm going to talk about as my sort of case study and I kind of want to emphasize I suppose that what I'm talking about with sea level rise doesn't apply to every type of climate prediction it faces a particular set of challenges where physical models have a shorter history the kind of global climate model so physical models for example of Greenland or of Antarctica the ice sheets themselves have a shorter history and also the observational data really we only started to get a handle on what the ice sheets were doing since the satellite era of the 1980s or so so it does have probably more challenges than a lot of areas of climate prediction and I want to talk about sort of the stories the narratives that have arisen around different types of evidence people have used to try and predict the possibility of extreme sea-level rise and in particular the collapse of the West Antarctic Ice Sheet I want to talk about how those narratives have played out in the literature among scientists with the IPCC the Intergovernmental Panel on Climate Change reports and also the media because if there's a sort of take-home message I guess that I'll give you now from this talk you know you don't need to worry too much about the numbers that I'll give but really but really the sense that the the pendulum keeps swinging of extreme sea level rise this century and and less extreme and so so it's almost been a sort of a struggle between those two polls so if you pardon the pun so what is it so I called it the ice apocalypse and I named this after an article by we try and adjust that slightly Eric Holt health last November in Grist talking about the ice apocalypse rapid collapse of Antarctic glaciers flooding coastal cities by the end of this century and that quote in the bottom there sort of one of these narrative points all this could play out in a mere 20 to 50 years much too quickly for Humanity to adapt so this talk and you'll see at the end as well it sort of almost a response to that article so what is collapse of the West Antarctic Ice Sheet or wastes as it's sometimes referred to so this dates back to a theory a paper back in 1968 about the possibility and this figures from there the possibility of a positive feedback of losing ice from the ice sheet so basically in areas where the bedrock that the ice sheet is lying on is underwater and in particularly in particular down sloping inland so it's sort of getting deeper into the center of the West Antarctic Ice Sheet that it's vulnerable to a positive feedback where once you start to lose a bit of ice at the edge whether that's through an ice shelf collapsing as we've seen in the peninsula in in in recent years or through some kind of direct melting from the ocean water from warm ocean water that if you start to lose ice basically it's almost like a bowl ball rolling down a hill it's a it's a sort of a runaway process it's called marine ice sheet instability and the kind of motivation for that theory really was some evidence in the 1960s that sea level had been quite a lot higher in the past and geological evidence that these kind of waterways that you can see these gaps on the right hand side had opened up in the past even if it was a bit uncertain exactly when and how much and so in this paper it was described as rapid perhaps even catastrophic a later paper on the same topic calling it a threat of disaster and so in terms of the impacts of that will you know in a way I don't need to say this you know three point three meters of sea level equivalent is locked up in that marine part of the West Antarctic Ice Sheet and of course he would have contributions under global warming from other sources so this is a map of sort of approximate elevation below five meters above sea level and so you'd get these kind of long-term changes in the world map of course over century sort of timescales set many century timescales but I'm talking perhaps more in this in this talk about my area which is the the maximum rates of sea level rise this century because of course they do there's much more literature in the area and of course they're very related what happens this century and how well we can predict that it's supposed to be the that's basically the easy part and then the kind of long-term predictions is harder so I'm going to talk about the struggle even to think about what this could mean for sea level rise this century so just as an idea I've sort of seen you know sea level rise is supposed to be one of the most difficult and expensive climate change impacts so for example under just over a meter of sea level rise this century we would start without if we didn't adapt to that we would start to get to kind of very very high numbers of global population flooded and annual damages you can see though so and in terms of kind of flood risk this is a review I did recently for the foresight project government office for science on kind of impacts for the UK so just as a sort of a ballpark these are estimates of annual damages at the end of the century you can see and the 60 centimeters of sea level rise this century seven hundred and sixty million per year and then if you go up to two and a half meters this century sort of roughly doubling and I put that partly to sort of give a sense of the scale of what I'm talking about this century this 2.5 meters is is the number used by the UK climate projections in 2009 for what they call the upper bound used in you know government adaptation planning Thames Estuary 2100 adaptation planning so Thames Barrier and so forth and so that's the sort of that's the sort of scale so I'm going to talk in the sort of first part of the talk about types of evidence we've used to assess extreme sea level rise and particularly from Antarctica this Antarctic collapse at a time when the physical models weren't really very well developed and so people were looking at different types of evidence and how to come up with ballpark numbers for this for the amount the maximum rates and so forth so up to the last IPCC report so I'm going to talk about sort of stories from the distant past and the Paleo climate record the recent past so 20th century and also expert elicitation and expert assessment so sort of one of the one of the very if you like sticky sort of stories of sea-level rise has been work by James Hansen there's an influential editorial essay in 2005 where he talked about sea-level rise during this period called meltwater pulse 1 a about 15,000 years ago this graph shows that very big blue bar is the is the rate of sea-level rise estimated during this period of 4 to 5 meters per century so very large explosively rapid as James wrote and so this this had influence at the time in terms of the kind of initial sort of sketching out idea of the UK climate projections and the Thames Estuary 2100 planning a sort of four metres or so is the absolute maximum we can think of for this century really the out there total total maximum right out on the tail so these are two these are rates you know reconstructed from things like fossil corals and stalagmites and stalactites in submerged caves and a whole range of different local sea level records put together to put to get a global sea level record so the meltwater pulse one a was a great concern but there's more kind of more understanding of that period now we're really we think that the the maximum those very high rates of sea level rise were because there was just a lot more ice in the world then we were coming out of the last ice age the heights of which was 22 thousand years ago and this paper shows for example the the North American ice sheet just before that time and as that sort of suddenly had a very big pulse of losing ice sort of down the middle of the ice sheet that contributed a lot to that thought up to that very high rate so the IPCC in despite this kind of explosively rapid sea level change said well we don't think we can use this as a kind of plausible maximum because it's primarily from large ice sheets that no longer exist so the left-hand blue bar is is kind of more looking at the rates over the whole kind of deglaciation period and apart from that meltwater pulse and so this was so this shows kind of lower rates of one to one-and-a-half meters a period that's a particular concern is a little bit further back so the last interglacial period before the last ice age there was an in-between ice age so an integrator a warm period and this has been a real concern because it was only probably half for one or the most kind of one and a half to two degrees warmer than pre-industrial in other words present-day too near future global warming equivalents but the sea level was six to nine meters higher than today so very much higher and there was an estimate back in 2008 of this kind of 1.6 plus or minus 0.8 meters per sec of the the rate of change which ended up effectively informing that 2.5 meters that I mentioned before for UK upper bound kind of planning but kind of recent estimates you can see a later one below there have kind of revised that down a bit to more like a meter there's a paper that looks at multiple ice age cycles over the last five hundred thousand years and what it shows similar to the previous example that I that I gave is that really the rate depends on how much ice you've got around at the time so this shows the maximum rate of sea level rise versus effectively the the amount of ice around and when we're down at present-day levels those rates are more like a meter per century or less so there's a kind of evolving understanding just even though those rates are uncertain they seem to be kind of recently revised down a little bit but similarly to that meltwater pulse one a even though it's perhaps more similar to today in the ice sheets the forcings were different so the the factors they're influencing climate were different it wasn't about you know co2 and greenhouse gases there were other factors that were affecting things so again the IPCC said the sort of limited use in a way from using these past periods so people have come up with kind of methods of using the recent past which are effectively extrapolation from the recent past getting a relationship between global sea-level rise and say global temperature for example this is different range of different studies reported in the last IPCC so that the vertical bars are showing the results they're up to kind of one one-and-a-half meters per century but in terms of the kind of narrative the the chapter authors from the IPCC I guess where were quite sort of there was a by they were largely ice sheet modelers and sort of physical modelers so forth rather than the people doing this kind of more back-of-the-envelope sort of extrapolation and so their kind of assessment of that evident line of evidence was this rather damning statement of them there is no consensus in the scientific community about their reliable reliability and that's really because we don't think that the future is going to be like the recent past the sources of sea-level rise are going to be different in the future compared to the past the balance between Greenland glaciers Antarctica thermal expansion and so on there was a structured expert elicitation just before the IPCC as well 26 ice sheet experts who were asked to think about the contributions from the three different from Greenland from West Antarctica in red they're an East Antarctica and the sort of take-home message of that is there was a very wide spread amongst experts you know principally for this reason that there are all these different lines of evidence limited amount of physical modeling and data available and so people just kind of came up with everything from almost nothing to a lot from Antarctica so again a lack of consensus was the with the take home and so in 2013 the IPCC had to sort of put all that together all the kind of lines of evidence I've shown so far and of course others and they really based their numbers for this century which are shown for the height sort of very high scenario they're going up to about a meter basically this century they based that around the literature of physical models so separate physical models for Greenland for Antarctica for glaciers for the thermal expansion of water as it were and so there was an awareness that physical models were incomplete there were very kind of course spatial resolution which is quite important for the the edges of the ice sheet or where all the action is and you really need high resolution to resolve the the physical processes there and importantly the the literature mostly was filled with point estimates so people would make one projection or perhaps a few different projections with one model there wasn't a systematic exploration of uncertainty with those models so they would take the model range and then kind of inflate it to something bigger and that was the red curve basically for that scenario and then for the for the possibility of West Antarctic collapse there were some estimates that sort of said it could happen and it could lead to half a meter or a meter of sea level rise just from Antarctica but because there was you know not great confidence in these models they just made this kind of semi quantitative statement of medium confidence that if a collapse were initiated it would not exceed several tenths of a meter this century so what does several mean so that was an arrow there and an arrow there and just after the IPCC report there was a kind of interesting kind of counter narrative put out in the literature so there was another survey so the first one was a cook method if you're familiar without calibrated experts I'm not so familiar with this one actually but they certainly the the thing that was distinguished with this survey from the previous one is that they defined their experts to be sea level experts rather than ice sheet experts so that incurred that of course encompasses all the people working on the Paleo and the extrapolation methods that I mentioned which were coming up with the higher numbers and so there their survey came up with higher numbers and and it was kind of amusing Lee sort of reported as experts say the IPCC underestimated future sea level rise well of course you know the experts made up the IPCC and so forth it's all about the sampling of the experts of what questions you ask and so forth I mean in this article talks about what they found isn't pretty so again there's kind of almost trying to kind of control the narrative I guess about lines of evidence so just a quick recap of those kind of paleo data and extrapolation and expert assessments of various kinds and you know people will agree that we could have up to a meter this century maybe meter and a half unlikely much more and then there was this kind of statement about several tenths of a meter if a collapse were initiated so then kind of a sort of the new phase the physics phase if you like so I'm going to talk about evidence from recent observations from satellite data and also from physical models I'm going to be focusing really on the Antarctic component of sea level rise now and so I sort of said if a collapse were underway we may have heard in 2014 that there were three papers came out which really suggested that this collapse could be underway in West Antarctica in the Amundsen sea impalement as I wrote a paragraph about this the satellite data showing you know lots of ice loss kind of consistent with a pattern we would see for this marine ice sheet instability suggested back in 1968 and to detailed physical modelling studies of the two big glaciers in that region also supporting that and so there's been this kind of new phase of using physical models to try to assess sea level contribution from Antarctica particularly kind of under that trying to assess the possible you know the potential sea level rise from that instability mechanism and we know that the models are still wrong you know and we know that they're they're wrong in important ways as well they're not just wrong they're sort of wrong in crucial ways but they are and particularly the low resolution is is is is really driving the key thing that they are trying to predict you know it's it's a very it's not a very slight it's not such a sliding scale of wrongness it's really an almost binary but there's an attempt now to address those limitations by using ensembles and Bayesian calibration and if you're not familiar with that basically just running the model lots of times using lots of different input values to sample all the possible you know future sea-level predictions that you could get from that model and then in most of these four studies shown here then comparing what those model different model versions say about the recent past or about the distant past and calibrating those with observations or reconstructions to try and give greater weight and greater credence to the most successful model versions to the future so trying to account for the limitations using sampling of uncertainty and calibration with observations and the numbers were coming up in these few papers since the IPCC basically so the the sort of story was that they are more effectively kind of consistent with the IPCC story you know they were sort of at the low end of the several tenths of a meter so that this paper Ritz Hotel I was co-author of and this was how it was reported things are not as bad as feared the IPCC greater than sixty six percent range is shown there so the idea was you know looking at this more systematically seems to show a low probability of those high sea level rise rates but then there's been a kind of wild card of big firecracker thrown into the mix which you may have heard of a paper by diccon turn Pollard in 2016 which came up with a new way of getting this instability of this collapse of the West Antarctic Ice Sheet so mersa had this marine ice sheet instability and they proposed a new hypothesis of marine ice cliff instability and it's basically a way that you can lose the edges of the ice sheet very fast and effectively if you have an ice shelf collapsing as we've seen as I say in the past if that leaves behind a very very tall ice cliff that that's structurally unstable and so it collapses into the ocean and then the next bit is also tall so that collapses into the ocean and it's always a sort of domino effect where the the ice sheet just loses and loses in a very in a much more rapid way than the original Mercer hypothesis so they're self-sustaining collapse of ice cliffs they they calibrated their model in a very simplistic at hoc sort of way with with paleo data and they came up with these very high numbers now remember this is just for Antarctica so this century a metre or more of sea level contribution and they went out to 2,500 and we're coming up with 16 meters from Antarctica alone so that's really coming up to those meltwater pulse one a numbers that I talked about before we need to take into account other contributions so this was understandably very highly you know it had a lot of influence of course it was sent by carbon brief to be the most cited climate paper in the media that year so so this one meter the reason it says doubling sea-level rise if you remember the IPCC's had 1 meter this century this basically takes us to two and it's and it was really the story behind that I suppose it's article that I mentioned at the beginning it was also used in the climate science special reports recently for the for the US so they revised their extreme scenario upwards from two to two and a half meters so something like the UK in 2009 but if you like I think there's a there's a feeling of amongst many of a greater credence in that number there's a there's a real there's been a real sort of concern about these predictions in a way that perhaps the previous estimates of two and a half meters were seen as a bit more vague and ballpark II so I've been with a few others reanalyzing their paper so this is a unpublished work where we we redid their calibration in a more formal statistical way that Bayesian not sorry not a Bayesian method better but a more well-defined statistically method an estimated PDFs because they just reported means and standard deviations and so the way that that the Contin pollard paper was used by others in the literature just in the last year or two is that they often would assume it was a Gaussian distribution because they were given a mean in a standard deviation and that makes people think of gaussians or in the absence information it's seen as an impartial destroy Stu make and of course it's not we think actually understanding of sea-level rise is probably quite skewed that it has a long tail that it's very not you know non-gaussian so this is an example of that in the top this paper by de Labarre that Hal where you can see the IPCC Antarctic contribution sort of basically around ten centimeters and then the the highest of the alternatives from the continent Pollard is actually 1.1 meters this century and our estimate was just using emulation if you're familiar with that effectively generating a much bigger ensemble than they had to estimate the shape of the PDF comes up with a lower mode a skewed PDF and we we kind of conclude that based on based on that analysis that these interpretations particularly kind of Gaussian interpretations would overestimate that probability of sea level rise so just to sort of make the sort of the kind of second thing that we found from from reanalyzing their results is that again one of the stories around that paper is that because they calibrated their model with two periods in the past which were very warm with very high sea level so one was the last interglacial period that I mentioned where we had higher sea level than today and not much more warming and the other was the Pliocene where three million years ago where it was much much higher and much much warmer and so there was a sense in which it was felt that this ice cliff instability this new hypothesis was required to explain the past and therefore that gave it credibility for the future and we really ought to be concerned about these projections but what we found when we looked into it in more detail is that really that mechanism isn't needed to explain those two periods you know the past data that we had to test the model with and with the emulation you can switch that mechanism off and test what the results would be without the predictions and they come up very they're very similar to the to the other results that I showed before so there's a sort of if you like an emerging consensus of sort of these are all quite different models they're simple models but they're quite different so without this wild card of cliff ice cliff instability there's a sort of convergence on bad but not too bad let's say if the sort of second thing part of this story is again in terms of narrative is the way that that mechanism has been described and the kind of credence that's given to it so when Dave Pollard first pub led a paper publishing this this hypothesis and using it in a model in the in the supplementary it describes it as somewhat speculative and that's because although it's a very sound theory that a nice cliff cannot be infinitely tall you know it has some limit on its what its height sort of structurally how that then translates into a positive feedback is very uncertain there are lots of ways in which that cliff collapse might be slowed down or stopped by kind of local factors and you know the physics of the ice sheet so it's called somewhat speculative in the nature paper a year later it was described as previously underappreciated so again this kind of how do we interpret evidence how much strength do we give to different kinds of effectively very simple descriptions of unknown physical theories and the kind of final point I want to make just about the long term so they made those predictions out several centuries so when we did our reanalysis of their data we found that so this is showing in pink this RCP 8.5 this high scenario a bit like I said before this asterisks are showing the mode so it's very skewed stays very very skewed in the future and crucially so very skewed these kind of purple and red lines sort of single lines below are from other studies and this Met Office one from other studies you can see that it's much higher than those other studies even under similar scenarios extreme scenarios but also I mentioned that we could switch off the cliff instability with the emulator and we find actually that the results the predictions are still very very high so it's not about just the cliff instability and because these results are so much higher than sort of any other models there's some suggestion that it could be oversensitive to warming maybe there other models are showing self limiting factors over the long term where the ice stops being lost after two three four centuries so quickly whereas this one keeps on going so that's kind of work in progress just very quickly I haven't really talked about this but you know that we've we're trying to work out if there's a threshold this century for collapse so we may think that the Amundsen sea is already collapsing but you know that could be limited it may not set off an entire West Antarctic Ice Sheet collapse it could be just that region and so there have been attempts to work out what what would be the threshold and they tend to come up with a usual kind of let's stick to one or two degrees of warming the strongest mitigation scenario two point six you know let's not you know we need to keep the ocean melting sort of at the similar levels but it's it's incredibly uncertain lots of the studies use the same model they don't really explore model uncertainties at all so just to kind of recap I wrote a response to Eric's piece in The Guardian which you can read if you're interested saying I thought his article was too pessimistic and overstating the certainties and the rate particularly this century and just a sort of sum up fee the sort of expert disagreements kind of up to the last IPCC report really come about because those those Paleo data areas the Paleo climate aren't good analogues for the future you know we know that even the recent past isn't a good analog for the future the physical model disagreements we know we have these limitations and missing processes and cost resolution and unfortunately we just don't really have enough data to constrain those limitations of the models so we need you know in our models we need greater diversity of different ways for example of representing this ice cliff instability the data you know there's lots we can do on Paleo data uncertainty quantification then computing will help and statistics can help bring it all together you know certainly there's still a role for for paleo data and expert judgment in all of that I think that's it thank you very much [Applause]

Channel: CSER Cambridge

Views: 611

Rating: 3.9333334 out of 5

Keywords: CSER, Centre for the Study of Existential Risk, University of Cambridge, Cambridge Conference on Catastrophic Risks 2018, CCCR2018, Tamsin Edwards

Id: OD0rT9FlEHY

Length: 29min 4sec (1744 seconds)

Published: Wed Apr 17 2019