A few days ago a team of climate scientists (Catherine Ritz, Tamsin Edwards, Gaël Durand, Antony Payne, Vincent Peyaud, and Richard Hindmarsh) published a study of “Potential sea-level rise from Antarctic ice-sheet instability constrained by observations.”
The study asked how much Antarctic ice sheets might contribute to global sea level by 2100 and 2200 AD. The results contradicted some earlier estimates which are on the high end, but conformed very closely to the current IPCC estimate, raising that number by a negligible amount.
The authors note that rising seas due to global warming is a significant problem. In other words, this research could be good news on one way, in that the highest estimates were not supported. But it is bad news in another way, in that the Antarctic ice sheet will contribute enough that when added to other sources of sea level rise, coastal regions will be seriously affected.
One of the study’s authors, Tamsin Edwards, wrote a summary of the paper in The Guardian. That essay provides a useful summary of the history of Antarctic ice-sheet research, and places the new research in perspective. In particular, Edwards notes,
We’re not the first to predict the consequences of Antarctic instability. So what’s new? We are the first to use all three elements I think are essential for climate predictions: physics, observations, and statistics.
I’m not sure if this is the first study to use data, physics, and statistics, but if it is, wow. However, there may be one very important thing missing from Ritz and Edwards Et Al: A full consideration of the factors involved in ice sheets turning into ocean because of global warming.
The study involved developing a computer model simulating the behavior of the ice sheet. This model was refined by comparing results of different runs, each using slightly different values for the relevant variables, with observations, in order to weight the model variants to get a more plausible set of results. Several thousand runs of the model were evaluated in this way.
My impression of the study, which I partially wrote up here, was that there were two possible problems. One derives from those earlier higher-end estimates that the new study contradicts. Some of those estimates are based on paleo data, which attempt to link either CO2 levels or global temperatures with known sea levels contemporary with those values. Looking at sea level from a paleo perspective, one could argue that current levels of atmospheric CO2 should be associated with much higher sea levels than we have today. Since added CO2 takes decades to be realized as surface warming, and surface warming takes, we assume, considerable time to be manifest as polar ice sheet melting or deterioration, the timing of sea level rise is very much an open question. In other words, a paleo-based estimate of many feet of sea level rise does not necessarily conflict with the results of this paper, which predict “that the Antarctic ice sheet will contribute up to 30 cm sea-level equivalent by 2100 and 72 cm by 2200.” Both could be right, because it may simply take several hundred years for sea levels to reach an equilibrium consistent with between 400 and 500 or so parts per million of CO2 in the atmosphere.
The second problem concerned me a bit more. This is the idea embodied in the “Ice swan analogy” I outlined in my post. The transformation of a continental ice sheet (and its nearby sea-situated parts) into ocean water could be somewhat over simply characterized as having two parts. One is simply the melt of ice being greater than the replacement of ice from precipitation and cold conditions. The other is the physical collapse of parts of the glaciers, causing large amounts of ice to slough off into the sea where they will quickly melt and contribute to sea level rise. It is likely that the latter would affect the former, so melting would increase because of changes to the structure and position of ice after physical collapse of large parts of it. Removing the distal part of a glacier’s tongue may unplug upstream sources of meltwater, and cause further rapid deterioration by destabilizing the ice sheet’s structure.
If the catastrophic deterioration of parts of the ice sheet (catastrophic in the sense that nothing happens, then more of nothing, then still more, then suddenly a threshold is reached huge chunks fall of for a time, then back to nothing again) is not accounted for, or insufficiently accounted for, in a model, then the model may be underestimating total ice sheet contribution to sea level rise, and the rate at which that may happen.
The possibility that large scale or at least rapid deterioration of parts of the ice sheet could happen has potentially important consequences. First, if such a thing does occur in large scale, the rate of sea level rise could be very rapid for a period of years. A sea level that goes up a few millimeters a year is potentially different, as a problem to which we must adapt, than one that rises in fits and starts. Second, the total contribution of Antarctic ice sheets to sea level rise may be both larger, and less predictable.
Richard Alley is a climate scientist at Penn State who studies ice, glaciers, sea level change, and abrupt climate change. I asked him for his opinion on the Ritz, Edwards, et al. paper. I am happy-sad to say that many of his remarks mirrored my own thoughts. Happy because it is always nice to have one’s ideas about complex science confirmed by an expert to not be completely wrong. Sad, because the Ritz, Edwards et al paper does look like it may be underestimating the total amount and rate of Antarctic ice sheet contribution to sea level rise.
Alley is concerned about the lack of attention in the Ritz, Edwards et al study to important relevant mechanisms.
Alley told me that among the many factors that contribute to sea level rise (melting of mountain glaciers transferring water from the land to the ocean, expansion of ocean water as it warms, possibly from mining of groundwater exceeding water trapping from dams and other human activities) that “the largest uncertainties are attached to the ice sheets. For the 20 years leading up to the IPCC Fifth Assessment Report, the Shepherd et al. IMBIE assessment (Science, 2012) found an accelerating contribution to sea-level rise from the ice sheets, but with an average of only ~0.6 mm/yr out of the ~3 mm/yr total. At that rate, loss of all the ice sheets would require just over 100,000 years; the rate of loss of 0.001%/yr is equivalent to me as a professor losing 1/3 of one potato chip per year on a diet. Both I and the ice sheets could lose weight more rapidly; we generally would consider my weight loss to be good and that of the ice sheets to be bad.”
Alley notes that the projections made by the IPCC are a good starting point for understanding sea level rise, but that work done since the IPCC projections were solidified for the most recent report tend to indicate slightly higher rates. As with other features of climate change such as climate sensitivity, the distribution of possible sea level rise rates has a long tail at the high end. This means that rates below the average estimate are highly unlikely, but higher rates are not as unlikely, and there is a small possibility of much higher rates. The tail at the high end of the distribution is lengthened primarily by uncertainty with what will happen in Antarctica. This problem is central to current research on the contribution of Antarctica to sea level rise.
Alley notes, “Because the ongoing changes are relatively slow in their contribution to global sea-level rise, and based on other research showing how some of the processes involved in ice-sheet shrinkage cannot accelerate hugely, there has been some optimism that the long tail won’t be realized. However, a small but growing body of scientific literature has looked at the possibility that fracturing could greatly speed shrinkage; meltwater can wedge open crevasses on ice shelves or non-floating ice near the coast, thinning beyond some threshold tends to lead to complete ice-shelf loss, giant icebergs calving off the resulting ice cliffs can move the grounding line back rapidly especially if aided by meltwater wedging, and theoretically estimated limits on cliff heights suggest that much faster iceberg loss and cliff retreat are possible.”
Alley was co-author of a review here that addresses this topic. Here’s the abstract from that paper:
Ocean-ice interactions have exerted primary control on the Antarctic Ice Sheet and parts of the Greenland Ice Sheet, and will continue to do so in the near future, especially through melting of ice shelves and calving cliffs. Retreat in response to increasing marine melting typically exhibits threshold behavior, with little change for forcing below the threshold but a rapid, possibly delayed shift to a reduced state once the threshold is exceeded. For Thwaites Glacier, West Antarctica, the threshold may already have been exceeded, although rapid change may be delayed by centuries, and the reduced state will likely involve loss of most of the West Antarctic Ice Sheet, causing >3 m of sea-level rise. Because of shortcomings in physical understanding and available data, uncertainty persists about this threshold and the subsequent rate of change. Although sea-level histories and physical understanding allow the possibility that ice-sheet response could be quite fast, no strong constraints are yet available on the worst-case scenario. Recent work also suggests that the Greenland and East Antarctic Ice Sheets share some of the same vulnerabilities to shrinkage from marine influence.
Alley lauds the Ritz, Edwards, et al paper as representing “a great amount of careful work, and provid[ing] a particularly broad exploration of some of the poorly known parameters that control the ice sheet.” However, he finds that the study did not address some important mechanisms.
…the model does not allow loss of any ice shelves, does not allow grounding-line retreat from calving of icebergs following ice-shelf loss, and does not allow faster retreat from breakage of cliffs higher than those observed today, especially if aided by meltwater wedging in crevasses. The model restricts grounding-line retreat to the rate given by thinning of ice during viscous flow of an unbuttressed but still-present ice shelf, with a specified upper limit enforced on the rate of that retreat. The model also does not allow retreat up a sloping bed under forcing, something that is widely observed. The Supplementary Information includes discussion of checks that the authors did to assess the importance of these assumptions, which the authors argue justify omitting the mechanisms. However, it remains that with the model not allowing very rapid retreat, not allowing ice-cliff crumbling after ice-shelf loss, and not allowing retreat up sloping beds, the model cannot exhibit some possible behaviors that could cause rapid ice-sheet shrinkage.
So, I view this as an important step forward for the scientific community, but the qualification in the last sentence of the paper leads to additional information showing that we cannot yet confidently place quantitatively reliable limits on the possible sea-level rise from the Antarctic ice sheet. I personally hope that the new paper is right, but I will continue research on this topic in the hope of providing improved estimates. Until such work is successful, I do not believe we can exclude the possibility of faster sea-level rise than suggested in the new paper.
I did ask Edwards questions about these missing elements, but have not heard back yet. If I do, I’ll either post her response as a separate item or add them here, as seems appropriate.
Have you read the breakthrough novel of the year? When you are done with that, try:
or Paperback
*Please note:
Links to books and other items on this page and elsewhere on Greg Ladens' blog may send you to Amazon, where I am a registered affiliate. As an Amazon Associate I earn from qualifying purchases, which helps to fund this site.
There are two important things to consider. One is the equilibrium volume of the ice sheet under a given climate regime. This is probably reasonably constrained by the paleo data. The second thing is how long does it take to approach equilibrium. Until a decade or two ago, the time scale was assumed to be many thousands of years. Now we have discovered that the rates of glaciation and deglaciation are assymetrical, with the later being faster -but how much faster is the million dollar (actually trillions) question. For growth of an ice sheet, its reasonably simple, excess of snow deposition versus net losses, with the former being limited by the rate of precipitation. But for melting, there are no hard limits. Especially for marine terminating glaciers, the loss at the ice water interface, due to both melting and calving is significant -and for Antarctic glaciers this is the dominant loss mechanism. Key uncertainties would involve the calving front, and the transport of warm water under the ice sheet. The later would be determined by the climate, acting through ocean currents.
I don’t think we need worry about the ice loss possibly being episodic, rather than quasi-steady rate. Even for the Antarctic, there are dozens of ice basins, whose dynamics should be relatively independent of each other (excepting being affected by similar climate changes). So meltrate spikes (if they occur), would likely not be simultaneous.
Greg. You say: “…causing large amounts of ice to slough off into the sea where they will quickly melt and contribute to sea level rise.”
I’m sure you didn’t mean that the ice contributes to sea level rise as it melts. Can I suggest: “i>”…causing large amounts of ice to slough off into the sea where it contributes immediately to sea level rise.” Once it’s floating it, of course, it can no longer affect sea level.
I immediately found myself sceptical of this study’s findings, so thanks for your work on this. The fact that ice loss is so slow to take effect only makes it more obvious that adding “…by 2100” to every projection tends to blind us to what a drastic impact global warming will have on future generations. I guess by 2100 the truth will have finally sunk in and people will be kicking themselves that they were so slow to accept what (by then) 500ppm can do to a planet.
Yes, I certainly did not mean it had to melt in the sea to raise sea level. No Comma, but I could certainly made that more clear.
Omega, how variability works out will be interesting. Yes, there are many sources so over some longer period it will even out. But most high volume sources would not be operative at any one moment of time.
The summary is, “The mechanics involved are not known well enough to model with any confidence.”
The “by 2100” meme has always irked me, and most especially in formal scientific publications that don’t provide any longer term context. I think that it’s past time that wherever possible both science and the media report any projections to the end of the 21st century in conjunction with additional projections to either a series of longer-term dates in the future, or to the new equilibrium of parameters discussed, preferably with time periods provided for context.
We don’t expect research into a virulent disease to focus just on the first stage of infection/affliction, so why do we so readily accept it in studies of climate?
Reviewers, publishers and media should ask for these long-term sequelae wherever possible, and if longer-term projection is difficult this should be noted, as should the confidence intervals in both directions .
My $0.02.
Prof. Alley writes:
In other words, Ritz 15 excludes the very things that are now widely regarded as making ice sheet collapse a nonlinear response to forcing change. Yet the study’s conclusions are drawn from matching modelled behaviour with current observations of Antarctic ice sheets. This approach is not going to yield reliable information about the rate of mass loss in the future if that process becomes increasingly nonlinear.
BBD: “This approach is not going to yield reliable information about the rate of mass loss in the future if that process becomes increasingly nonlinear.”
The issue appears to be one of loads and forces acting on those loads (mechanical physics) instead of thermodynamics. Ice-penetrating sonar, if I recall correctly, only mapped parts of the “drain point” of the West Antarctic; as far as I know, no one knows the Antarctic’s geology well enough to even take a wild guess at constructing a ice loss model.
Greenland melting is the much greater risk compared to Antarctic. By the time Greenland ice mass mostly gone with “business as usual” (200 or 300 years), humanity will be back to banging rocks together as our highest knife-making technology. In the long run, this is a good thing.
So let me see if I got this right. Everybody who’s done a study on this has given a range from not too bad to kind of bad. But they all agree that the graph of sea level rise over time is a straight line? When has nature ever graphed a straight line (in complex events)?
Richard Chapman: “But they all agree that the graph of sea level rise over time is a straight line?”
Well, no. The rudimentary models are “straight lines” because that’s the best one can do with the tiny amount of data currently available. It’s the “Antarctic’s contribution to sea level rise” that is a “straight line,” since no one at the moment knows any better. The modelers don’t think it will be linear.
As I noted, I did ask Edwards for her perspective, but she was not able to get to that over the weekend, and will be sending me comments. They should be interesting.
One thing I do want to point out is that this team is a bunch of good scientists. The work is regarded as good work. It just may not be the whole story. To be fair, it is hard to get the whole story into one peer reviewed paper.
Stay tuned!
In a much simpler context, a vaguely-similar problem arises in trac king money flows into think tanks, i.e., :
1) They reported $X[i] income year-by-year.
2) One can find $Y[i] of donations from known private foundations, year-bu year.
3) That leaves Z[i] = X[i] – [Y[i] of “dark” money you can’t find, and unlike Y[i] where sometimes can see steady donations or trends, you have no idea what’s happening with Z[i].
The question is:”Is there good chance that climate warming can cause a so-called Heinrich Event”. If the mechanism that trigger such an event already exist, I believe it is the case, then there is hardly a chance to escape from a global catastrophy. What we do not know is, when and how fast will it happen.
Does anyone have input on this “Take the $100,000 Global Warming Believer Challenge!” meme infecting the denialosphere in the past nine hours?
http://www.informath.org/Contest1000.htm
It seems to me that this person is demanding that a statistician do that which no statistician has ever claimed is possible, therefore human-caused climate change is a lie.
There are two parts to the “challenge,” neither of which any statistician has said is possible to pass.
The first is “to anyone who can demonstrate, via statistical analysis, that the increase in global temperatures is probably not due to random natural variation” — an oxymoron. Anyone who tells him that demand is nonsensical must first pay US$10 before he will read it. Sheeeish.
The second part is to unscramble what the writer called a series of 1,000 “trendless statistical model fit for global temperatures,” which also makes no sense: if there is a fit to global temperatures, each one of the 1,000 series must have a trend. He then added another trend to a number of pseudo-randomly selected series.
He then insisted he will pay US$100,000 to the person who can identify at least 900 of the series by which have the additional trend, and which are “trendless.” BUT NONE OF THE SERIES ARE “TRENDLESS” if the “fit for global temperatures” is applied.
Every temperature series of 135 years, yearly average, will have a trend— going back hundreds of thousands of years. That is because yearly global temperatures have trends, with and without humans even existing at the time.
Desertphile;
Yes, his challenge is a load of crap, but he seems to be amazing at that.
In his “critique” of the 2013 IPCC report, he makes this comment
dean, quoting Lunatic: “To illustrate the easy way, consider our example of tossing two coins. Instead of calculating the probability of getting two heads, we could estimate the probability as follows: take two coins, toss them a million times, and count the number of times that both coins come up
heads. I tried doing that and counted 249943 times that both coins came up heads…”
Hee! Well golly, everyone can now thank him for checking to see if numbers still add and divide properly. But then, it wasn’t 250,000 therefore maybe human-caused climate change isn’t happening.
Argh. I read the “challenge” yet again. It still makes no sense. A “trendless” series must always sum to zero— none of his series do that. Am I just too stupid to understand (which is very often the case), or is he the idiot?
Curse no preview: he does go on to say that he didn’t really flip coins a million times but simulated it: his point is about basic statistical models, and he’s talking about fair coins. The implication seems to be that using a computer indicates the model is valid: all his example provides is the basic idea that probability, as he presents is, is a long-term percentage occurrence.
dean: “Curse no preview: he does go on to say that he didn’t really flip coins a million times but simulated it: his point is about basic statistical models, and he’s talking about fair coins. The implication seems to be that using a computer indicates the model is valid: all his example provides is the basic idea that probability, as he presents is, is a long-term percentage occurrence.”
I am very glad he didn’t toss a coin one million times: I would then have to pity him. 🙂 His meta-thesis is that all of the (seven?) long-term temperature reconstructions going back 100 years and longer are Drunkard’s Walks (random walks), with the probability of going up, down, or stay the same for 135 “x” values, 1,000 series of 135.
The problem is, Drunkard’s Walks *HAVE* *TRENDS* and so do global yearly average temperatures, regardless of human activities. His “challenge” is not winnable because he is demanding that someone do that which no climatologist has said is possible. And I very much suspect he knows that fact.
Check A (stupid) $100000 bet at andThenTheresPhysics, the challenge is pretty much rigged so Koonan can’t lose.
https://andthentheresphysics.wordpress.com/2015/11/20/a-stupid-100000-bet/
Johnl: “His problem is that he takes no account of all the physical evidence there is for human-caused climate change. To think it can be reduced to statistics is just loony.”
Thank you. It appears the silly goose believes, or wants other people to believe he believes, that all of the independent temperature reconstructions are random walks, and the world’s climatologists are basing their conclusions only on these reconstructions— ignoring the laws of physics.
Another thought occurred to me. His 1,000 series of 135 data points each should sum very close to between +5 and -5. If it does not, his “model” is wrong.
Doug Keenan’s $100,000 challenge is discussed at length here: https://andthentheresphysics.wordpress.com/2015/11/20/a-stupid-100000-bet/
His problem is that he takes no account of all the physical evidence there is for human-caused climate change. To think it can be reduced to statistics is just loony.
Snow is falling, and I don’t want to go feed the hens, so I ran the numbers—
His data series has a cooling bias of 56%, whereas his “model” should be within +5% / -5% (there are far more decreasing trends than increasing). Also, all 1,000 trends start at -0.23 as the first “year” even though almost all of the series’ remaining 134 “years” are small fractions of that. WTF? Did he even look at his series to check for errors?
His data don’t even do what he said they do. And again, I bet he knows that.
The first new thing you are supposed to learn when (if) you are getting experience in consulting is that you need to learn as much as possible about the subject the person you’re helping is an expert in. Real world data does not equal text book data.
Golly. I spent another 15 minutes looking at Keenan’s “challenge.” It sucks having nothing more productive to do when snow is falling.
The data show his randomization method(s) failed: the data are not pseudo-random, therefore the “walks” are not “random walks.” There is a very strong bias towards decreasing numbers, by 7d:3u ratio. A random walk will be 1d:1u even with a underlying linear trend added.
Post filled with Keenanspam, ruined now for its original purpose.
Heard anything from Tamsin, Greg?
Is climate change a matter of statistics? Clearly no. It is a matter of its natural effects.
Hartmut Heinrich: “Is climate change a matter of statistics? Clearly no.”
No one here claimed otherwise. Human-caused climate change is a matter of physics.
“It is a matter of its natural effects.”
Yes: the natural effect of humans producing greenhouse gases. We know. Your point was lost to me, some how.