RealClimate logo

Technical Note: Sorry for the recent unanticipated down-time, we had to perform some necessary updates. Please let us know if you have any problems.

Ice Sheets and Sea Level Rise: Model Failure is the Key Issue

Filed under: — group @ 26 June 2006

Guest post by Michael Oppenheimer, Princeton University

A plethora of research articles has appeared over the past year reporting new observations of the Greenland and West Antarctic ice sheets along with associated modeling results. RealClimate has reviewed the issues raised by these articles and attempted to clarify the sometimes conflicting inferences about the current mass balance of the ice sheets, as well as their future contributions to global mean sea level rise (see here and here).

Nevertheless, the issue still seems to perplex many journalists and others because there are two entirely distinct aspects of the sea level rise problem that are emphasized, depending on which scientists are speaking. On the one hand, these ices sheets are large enough to ultimately raise sea level by 7m and about 5m, for Greenland and West Antarctica, respectively. On the other, the recent observations that caused such a stir report a current contribution to the rate of sea level rise not exceeding ~1mm/yr from both ice sheets taken together. If this rate were maintained, the ice sheets would make a measurable but minor contribution to the global sea level rise from other sources, which has been 1-2mm/yr averaged over the past century and 3mm/yr for 1993-2003, and is projected to average 1-9mm/yr for the coming century (see IPCC Third Assessment Report).

The key question is whether the ice sheet contribution could accelerate substantially (e.g., by an order of magnitude) either in this century or subsequently. Sea levels were indeed much higher in the distant, warmer past but the timing of earlier sea level rise is very uncertain. From the point of view of societal and ecosystem adaptation, the timescale over which ice sheets might disintegrate, which may be on the order of centuries or millennia according to the two extremes posited in the literature, is crucial.

The complexity of bridging the gap between past and future trends is familiar to the climate community, which has dealt with the same issues with regard to global mean temperature. Ice sheets aside, continuation of past warming trends based on the roughly 100-year temperature record (0.05-0.1ºC/decade) would pose a significant but manageable problem for most countries. Projected future warming (0.15-0.55ºC/decade) based on increasingly reliable general circulation models, poses much more serious, even unmanageable challenges. But the state of ice sheet modeling is far different from the state of atmosphere-ocean modeling, as underscored by the recent observations. At this juncture, numerical modeling simply does not provide a credible basis for quantitative projection of ice sheet behavior in a warmer world.

The limitations of ice sheet models were revealed starkly by the collapse of the northern sections of the Larson B ice shelf in 1998 and 2002. Glaciers bounded by the landward edge of the ice shelf accelerated toward the sea while glaciers bounded by the more southerly section of the ice shelf, which remained intact, didn’t. Apparently, backpressure on glaciers from the abutting ice shelf provides a significant portion of the restraining forces keeping land-based ice in place, at least in some instances. The recent behavior of glaciers farther south in West Antarctica, and in Greenland, points to a similar dynamical response to ice-shelf fragmentation.

Many glaciologists regarded these observations as a clear test of the ability of ice sheet models to forecast dynamical changes in a warming ice sheet, a test the models failed. The long-standing inability of ice sheet models to reproduce the ice streams of West Antarctica, unexplained dynamical contributions to the mass balance of the Greenland ice sheet during the late 1990s and its apparent basal response in one location to surface melt-water reinforced this skepticism. The problem is threefold: the physics in the models is incomplete, the numerical problems are very difficult particularly in the neighborhood of the grounding lines where the land-based ice begins to float (Vieli and Payne, 2005), and observations remain sparse. It may take more than a decade, perhaps much longer, to bridge the gap in the model world because human and financial resources dedicated to the ice sheet problem are woefully inadequate (see Kintisch, 2006: Science 312, 1296, for a discussion of problems with the planned National Polar-Orbiting Operational Environmental Satellite System (NPOESS), a proposed platform for crucial future ice-sheet observations).

We might also look to our experience with GCMs for some guidance as to how to evaluate the situation while we await an improved basis for numerical projection. GCMs are anchored in a broader range of observations than are ice sheet models at both the process and synoptic levels. But what would we do today if GCMs had failed several critical tests? With the climate already changing, we would likely not throw up our hands and say “let’s come back and reassess the situation once our numerical tools have improved”. More likely, we would increase our attention to paleoclimate analogs, a standard test bed for, and complement to, numerical modeling.

With the ice sheets already changing, the importance of analogs has come sharply into focus. Inferences from Eemian climate and sea level yield a wide range of estimates with regard to the climate changes that might result in widespread deglaciation of either or both ice sheets. Polar warming less than 5ºC (Overpeck et al, 2006) and global mean warming of 1ºC (Hansen 2004) or 2ºC (Oppenheimer and Alley, 2005) above recent temperature have been cited. Such global mean warming would almost certainly occur during this century, given forcing within the range of IPCC scenarios. In the southern hemisphere, the requisite circumpolar warming may occur more gradually, but a northern hemisphere trigger for WAIS deglaciation during the Eemian has also been proposed (Overpeck et al 2006). The information on potential rates of deglaciation is extremely sparse and its relevance to the future unclear. Potential rates of sea level rise equivalent to 1m/century (10mm/yr) have been suggested based on paleoclimate analogs (Overpeck et al, 2006) and by comparison to current ice discharge from West Antarctica (Oppenheimer 1998).

Filing the gap in knowledge between the risk (a significant probability of many meters of sea level rise) and the current reality (rapid local ice responses to local warming but small aggregate effect on sea level rise so far) will require a sharpened focus on all three fronts: observations, modeling, and paleoclimate assessment. Currently, the resources to do any one of these at the appropriate level are lacking. And because greenhouse gas concentrations and ice sheet loss are effectively irreversible, policy decisions need to be made based on the information in hand, which argues that deglaciation could be triggered by a modest warming.


E.Kintisch, Science 312, 1296-1297, 2006.
J.Hansen, Climatic Change, 68, 269-279, 2005.
M. Oppenheimer and R.B. Alley, Climatic Change, 68, 257-267, 2005.
M. Oppenheimer, Nature, 393, 325-332, 1998.
J.Overpeck et al, Science 311, 1747-1750, 2006.
A.Vieli and A.J.Payne, J. Geophys. Research, 110, F01003, doi:10.1029/2004JF000202, 2005.

90 Responses to “Ice Sheets and Sea Level Rise: Model Failure is the Key Issue”

  1. 51
    Leonard Evens says:


    As usual Michael says it all very well. Sequestering CO_2 is a much more serious possibility, at least as a short term strategy, if someone can figure out how to do it and our leaders were to pursue it. It is unfortunate that while sequestering CO_2 is often cited as a possibility tied to increasing use of coal, few if any actual proposed coal plants take it seriously. For the present, it seems more of a cover for increased use of coal without CO_2 emissions controls, which is actually the worst thing we could do.

  2. 52
    George A. Gonzalez says:


    I object to the notion that geoengineering would give the political and economic elite a license to continue fostering the emission of dangerous amounts of greenhouse gasses. They already have such a license, and are exercising it.

    My point in invoking geoengineering is to get a discussion on it going, and to draw on the expertise of the participants/readers of this list to gain a better understanding of the potential ramifications of geoengineering as a response to climate change emissions. In this regard, Mr. Tobis’s commentary is welcome and appreciated.

    I, nonetheless, believe that issues raised by geoengineering should be subjected to scientific treatment and rigor. It is my estimation that this is the next logical step of climate science with respect to climate change.

  3. 53
  4. 54
    Ron Taylor says:

    Concerning the possibility of meltwater percolating to the base of the Greenland icesheet, it is instructive to look at Greenland with Google Earth. Check the area, for example, on the west side at about 68 deg N, 49 deg W, from an eye altitude of 50-100 mi. You will see an incredible number of meltwater lakes. You can track them for miles inland from the western edge of the ice, and northward to some 75 deg N. Zooming in, they are accompanied by structures that appear to include sinkholes, crevasses and even streams between the lakes. With these structures in a context of shifting, fracturing and moving ice, it seems inevitable that water would be reaching the base.

  5. 55
    Gary says:

    Do any of the climate models take into account solar activity?

    The correlation between solar activity and the apparent heating and cooling of the Earth seems obvious. And, there seems to be a recent uptrend in solar activity that readily explains the apparent 20th warming.

    Unless this data can be explained and modeled, I don’t see how any climate model can be taken seriously. Seems like certain scientists are cherry picking their model parameters to fit thier politics.

  6. 56

    Emboldened by the reponse to my previous post, thank you Gavin, I thought I might grace you with my later thoughts.

    They are dominated by by reading Jim Hansen’s review, and his apology for not helping Al Gore earlier. I agree with nearly everything Jim wrote, but I do feel that it is too little too late :-( Jim seems to think that we still have ten years in which to act. But the Greenland glaciers have begun to melt and that melting is accelerating. They have not collapsed, but now it is inevitable that they will.

    Mauri argues in his first post that rapid collapse of the Greenland ice sheets is not posssible, but that is based on the premise that they are mountain glaciers which have melted slowly in the past. In his opinion, it is only low lying ice sheets, such as those that covered Canada that collapse suddenly. However, it seems to me to be more likely that sudden collapse is caused by undermining by the sea. That would have applied to the Canadian sheet, where much of the land was isostatically depressed and the rest, such as Hudson Bay, is still below sea level. Greenland may be three islands, and so its ice sheet culd also be undermined by the sea when melting begins. In fact the rise in level of 2 meters caused by melting must be due to the ice sheet floating.

    Mauri concludes his second post with the sentence “But the bottom of the glaciers are truly a land of unknowns.” My point is that whether we have ten years before the Greenland ice collapses is therefore also an unknown.

  7. 57
    Dan says:

    re: 55

    See for starters.

    Seems like certain posters are cherry picking their points which have been thoroughly explain previously to fit their politics.

  8. 58
    pat neuman says:

    … almost everywhere they looked, glaciers started to retreat about 17,500 years ago, …
    The only place that does not fit the warming pattern they observed is Greenland, which lagged about 2,500 years behind the rest of the planet in starting to thaw certain glaciers. …

    Above excerpts from:
    Ice-age study: World thawed out at same time
    by Lee Bowman: Tucson, Arizona – June 11, 2006

    Now I think we can say that over the last few decades, almost everywhere we look, glaciers are retreating rapidly (including Greenland).

  9. 59
    Daren Willcock says:

    If you are going to rely on paleo to sort this problem, you have an unlimited choice of models to choose from ranging from creationist to balance theory. WHICH DO YOU PICK?

    [Response: Simple. You pick the ones that are based on sound physics which has been tested in field observations and laboratory experiments. That knocks out the "creationist" models, whatever that may mean. Then you look at how well the models reproduce global mean temperatures, seasonal cycles, patterns of rainfall, etc., and how well they reproduce the 20th century climate trend. Maybe you also see how well the models reproduce the last glacial maximum, for which the climate forcing is pretty well known. Once you've done all that, you're still left with a spread of forecast temperature increase, all of which should be considered in the realm of possibility when assessing risk. The situation for ice sheet models is more problematic, since they're less easy to test. However, the same principles apply. --raypierre]

  10. 60
    Johnno says:

    I was surprised to see this was predicted on the BBC climate website way back in 1999
    The gist of the argument seems to be that melting clathrates shrink the volume of the seafloor.

  11. 61
    llewelly says:

    Jim seems to think that we still have ten years in which to act. But the Greenland glaciers have begun to melt and that melting is accelerating. They have not collapsed, but now it is inevitable that they will.

    After Greenland, there is the WAIS. After the WAIS, there is the EAIS. Preventing the melt of the WAIS though GHG emissions cuts is still feasible. The need to plan for the melting of Greenland’s Ice Sheet(*) does not mean we should throw up our hands and abandon GHG emissions cuts. I am not convinced that any of the geo-engineering schemes mentioned in this thread (assuming they work) could allow us to abandon GHG emissions cuts. So I’m opposed to the constant repetition of the notion that it is already too late to act, or that it will someday be too late act. There is no point at which more GHG emissions will not make matters worse.

    (*note: I believe the need to plan for the melting of Greenland is real, even though I think avoiding the melting of Greenland is still possible, if only barely.)

    [Response: Very well put, Llewelly. --raypierre]

  12. 62
    Claire Kenyon says:

    Someone told me that Kadhafi had at some point had a grand plan to build a canal and detour water from the sea into the low-lying areas of the Sahara desert. Would anyone like to comment on this as a possible response to significant sea-level rise?

  13. 63
    Hank Roberts says:

    Claire, I suspect your ‘someone’ got the story wrong. Try this:

  14. 64
    James Lane says:

    And yet, recorded “global” sea-level rise is measured in millimeters. Arctic sea-levels are reported as falling. Nevertheless it’s time to “head for the hills”?

  15. 65
    Eddy Ahnen says:


    I’m a newbie from Europe and have a question.

    Boreholedate from Greenland show, that the temperature has been higher there, for most of the holocene, as far as I understand these graphics.

    How was the sealevel behaving 1000, 2000 and 8000 Years ago.

    The glaciers in Europe (Alps) have retreated 2000 Years ago 300 Meters more than today, for a long period, so that trees have grown there and farmers have raised there cattle on the “alms”.

    What do we know about the sealevels of these times?

    Thanks for your hard work here!!!


    [Response: What Greenland graphic are you looking at? Do you mean oxygen isotopes or thermometry based on the downcore temperature profile? Neither one gives support for your statement that Greenland temperature has been higher for most of the Holocene, so the premise of using the Holocene to test ideas about warming and sea level isn't very promising. That's why Hansen uses previous interglacials -- some of which were warmer than the present time -- to give an idea of how much Greenland and Antarctica could melt in warm climates. We're probably not yet warmer than the warmest interglacials of the Pleistocene (roughly the past 2 million years, but unless sensitivity comes in at the very low end of the range, if we double CO2 we're almost certain to exceed the temperature of the warmest interglacials. That's what has Hansen worried about sea level. --raypierre]

  16. 66
    Eddy Ahnen says:

    Hello again,

    I don’t find the graphic from Scientific American (Spektrum der Wissenschaft), but the first graphic on this page is nearly the same:

    The author in Sciam made a hypothesis, that men had induced the global warming 8000 Years ago. I’ve seen that graphic already years before.

    As far as I know, the glaciers of Europe had nearly totally disappeared some 6000 years ago.

    What I mean, is, that if we consider Greenland alone, or the Northern Hemisphere, the ice must have melted in the same way than today.

    In Sciam, I also read about the most probabel models predicting a rise in temperatures of 3° C (from 2 to 4) for a doubling of Co2. I think you have been asked as well as Mr. Rahmstorf, closer to my country. Is that still your opinion today?

    But, does a rise of 3° C over a rather short? period have the same effects as a whole warm period.

    Is it true that the Eemian warm period was perhaps 5° C above todays temperatures?

    I hope you excuse my very bad english writing ;-)

    And thank you for the very quick answer!!! :-)



    [Response: Estimates of Eemian temperatures look to be about a degree or so warmer on average. Because of the orbital configuration, you might have expected NH summers to warmer still (possibly up to 5 deg locally, but you'd have to look at the primary literature for examples if they exist). The response to 2xCo2 is just a metric used to compare models (and 3 degrees is about right), estimates of what actually may occur range from 1.5 to 5 deg C by 2100. -gavin]

  17. 67
    Hank Roberts says:

    Lonnie G. Thompson, Ellen Mosley-Thompson, Henry Brecher, Mary Davis, Blanca León, Don Les, Ping-Nan Lin, Tracy Mashiotta, and Keith Mountain
    Inaugural Article by a Recently Elected Academy Member:
    Abrupt tropical climate change: Past and present
    PNAS published June 30, 2006, 10.1073/pnas.0603900103
    [Supporting Data Sets]

    Found thanks to mention here:
    “Ohio State University, June 27, 2006
    “First Compilation Of Tropical Ice Cores Shows Abrupt Global Climate Shifts

    “… a massive climate shift to a cooler regime that occurred just over 5,000 years ago, and a more recent reversal to a much warmer world within the last 50 years….
    “… in most of the world, glaciers and ice caps are rapidly retreating, even in areas where precipitation increases are documented. This implicates increasing temperatures and not decreasing precipitation as the most likely culprit.
    “… Ohio State University’s Byrd Polar Research Center and three other universities combined the chronological climate records retrieved from seven remote locations north and south of the equator.”

  18. 68
    Blair Dowden says:

    Re #66: I had never seen Greenland ice core temperatures presented for this time period, and I am rather surprised. I spot checked the graphs against the NOAA reference data, and they look correct.

    I think it is misleading to claim temperatures 3,000 years ago were three degress warmer on the basis of one or two hundred year peak. The key unknown is the time for the ice cap to reach equilibrium for a given temperature. Lets take that to be one thousand years. Hansen thinks it is less. Mentally running a thousand year smoothing filter over the Greenland data, I get a one degree difference. From what I have been able to find from paleoclimate data, he equilibrium temperature rise per degree C (global average) is about 6 meters. But we are now comparing Greenland local temperatures with the global average. Using another approximate relationship, which is 2.5 degrees of Arctic warming per degree global average, I get 6/2.5 = 2.4 meters that sea level should have been higher 3,000 years ago.

    This is all very approximate, and I would love to see this simplistic calculation corrected, but it looks to me there should have been a significant sea level difference that does not seem apparant in the paleo data.

  19. 69
    Eddy Ahnen says:

    Hello again,

    Thank you Gavin!!!!


    I’ve read a few things about the glaciers of Europe (I like skiing, and I would cry if the snow would not come in wintertime), that tell me that our glaciers behave like they did 1000 years ago and 2000 years ago. So this must not be so exceptional. But the temperature is rising quickly …!!!

    But on the map shown in your last link I see a glacier close to Punta Arenas. This link

    shows no exceptional warming since 1880. How could the glaciers melt without a rise of temperatures?

    The same thing happens to most of the Antarctic Station Temperatures.

    Even in Europe things don’t look so dramatic:

    The two last datasets are from high mountains in the Alps.

    We also know, that the glaciers retreated more 2000 years ago.

    I’m very confused about all these data and proxies. The temperature series I’ve picked out of many are very long datasets since 1880 sometimes until 2006.

    Might the melting perhaps be due to dust and particles in the air in the Alps ????

    Thanks again

    Eddy (Luxemburg)

  20. 70
    Hank Roberts says:

    Eddy, you may be doing what’s called ‘cherry picking’ — have you looked at the overall picture?

    Remember, while you have ice, the temperature measured stays at 32 degrees the whole time the ice is melting — the energy’s going into separating the molecules of water (melting) although the thermometer won’t change til the ice is gone. I don’t know where the GISS temps are measured but you say you picked sites around glaciers, so look at the averages. Look also at the Ohio State paper linked above that documents tropical ice cap melting.

  21. 71

    Re: #68

    FWIW I just did a Q&D numerical version of this smoothing and came up with -31.94 C for 0-1kybp and -30.30 for 3-4kybp, which is more like a difference of 1.6°C. The plot was average of the temperatures aggregated against floor(kypb). I can post the plot if anyone is interested.

  22. 72

    Re: #71

    P.S. The data was from here.

  23. 73
    Eddy Ahnen says:


    Hello Mr. Roberts,

    I have tried to pick out important temperature series from places, where global warming can be felt or seen, like the Alps or Skandinavia or for example Greenland. We also have a station in Luxemburg, that shows a warming according to the global temperatures.

    But also around here we see, when we compare longer series, that the warming is not SOO exceptional:

    That’s why I wanted to see the temperatures from important points, like the High Alps (Zugspitze and Jungfraujoch) (an area where glaciers are melting quickly)

    Aren’t they beautiful????

    I was born in 1961 and I love the snow. The warm winters I’ve lived when I was in highschool made me depressed and I was happy baout our cold winter 2005-2006, making me hope to live some more happy years … ;-)

    I have looked at all longtime temperatureseries in the Nasa-records and the only ones I found really extremely rising were the Japanseries. I got an E-mail-friend there, who told me about that.

    The temperatures around Antarctica seemed to be the ones with the least rise for the last 100 Years, like for example Punta Arenas about which I read a book (canoe).

    Environment protection is very important for me, but I hope that the predictions of the models are wrong ;-) because I like my Alps and Switzerland like they are ….

    I am not a scientist, but I “believe” in Science. Let’s hope that this time they are wrong ;-) (like R2D2 sometimes)

    If any of you have links to free papers like for example the glacier-report I would be very grateful.

    But thanks again for sharing your precious time with me!!!!! :-)

    This is a very nice site!!!!

    Have a good time


  24. 74
    Blair Dowden says:

    Re #71: Richard, I would like to see your graph. The thousand year figure for ice caps reaching equilibrium with temperature is rather arbitrary, so other values such as 500 years or 2500 years might also be interesting.

    This data leaves me with the impression that equilibrium takes longer than a thousand years, contrary to what James Hansen is saying.

  25. 75

    Re: #71

    Hi Blair. I’ll post a few versions later today after I get to work.

  26. 76
    Hank Roberts says:

    Eddie, see this thread, it may explain why you need to do the analysis rather than look at one station’s few data points to determine whether there’s a trend. Esp. response 11.

  27. 77
    Eddy Ahnen says:


    I’ve read the whole thread and again looked at many series in the Northern Hemisphere, where most of the data comes from. What striked me again is that, the longer the time period of records, the less the warming of today appears to be soo exceptional.

    We go back to 1880, that was a rather cool period here in Europe, as far as I know. I try not to be cherrypicking, but only considering longtime records. I think that’s the only reliable way to compare.

    There are many stations that go from 1960 to 1990 for example. How can they be part of it? The same happens with stations from 1990 to 2006. If there is a huge warming, how was it 100 years ago?

    I’ve read about Russia showing a real warming trend:

    But also here, the longer the record, the less spectacular the trend.

    It just strikes me.

    But if we see these in Japan :

    They show a clear trend, but they are not rural….

    That’s why I ask myself, how actual temperature series are put together with old data.

    This is the only site I can expect a serious answer to the question…

    Thank you all again and have a nice weeekend!!!!


    [Response: All of the series you show have a positive trend so I'm a little unclear about you are asking. If you want to know how these are put together to get the composites then I suggest you look at the papers describing the procedures themselves. They are quite accessible and will provide more information than we can get into here. In particular, read:Hansen et al (1999). - gavin]

  28. 78

    Re: #75.

    (Sorry to take so long, yesterday was rather busy.)

    Here is a set of plots of the data at resolutions of 500,1000 and 2500 years. It looks to me like the response is more on the order of 500 years as Hansen suggests.

  29. 79
    Ken Robinson says:


    This post is somewhat off-topic, but I can’t find an open thread that’s more appropriate. Please forgive the intrusion.

    I was reading LeGrande et al’s recent paper (LeGrande et al, Consistent simulations of multiple proxy responses to an abrupt climate change event (2006), PNAS, Vol. 103, No. 4, 837-842), to which you’ve referred a couple of times and for which you are a co-author. I have a few questions which I hope you can clarify.

    This paper states that:

    “In 12 coupled GCM simulations, we introduced fresh, isotopically depleted water (0°C, -30 permil d18Oseawater) into the Hudson Bay (adding to the total volume of water in the climate system). Additionally, we added a passive tracer to the melt water to track its movement through the simulation. The results discussed here use a range of volumes from 2.5 to 5 Sv yr, applied from 0.5 to 1 yr. Further experiments with volumes from 1.25 to 10 Sv and applied from 0.25 to 2 yr are qualitatively similar. The range of 2.5 Sv yr (0.788 x 10^14 m3 or ~23 cm rise of sea level) to 5 Sv yr (1.576 x 10^14 m3, 45 cm rise in sea level) is the most consistent with recent hydrologic model estimates for the volume and duration of the 8.2-kyr drainage of glacial Lakes Aggasiz and Ojibway (2).”

    In examining the referenced paper (GKC Clarke et al (2004) Q. Sci. Rev. 23, p 389-407), I was looking for the model which corresponded to volumes described above (which, if I’ve done my sums correctly, are 78,800 and 157,600 km3 respectively).

    Clarke et al contains 10 different models for possible flood volumes from Lake Agassiz-Ojibway (summarized in Table 4, each corresponding to different flood routes). The C(230) model contains the largest flood volume of 70,800 km3. The smallest is their case A1(230) which represents 27,900 km3. Their “preferred” model for an initial flood is B(230), containing 53,900 km3 of flood water, which corresponds to a 15 cm sea level rise. Clarke makes the distinction between an initial flood (Kinojevis event) and a subsequent one (Fidler event), the combined magnitude of which would fall within the ranges that LeGrande considers. However, Clarke also makes clear that the interval between such floods would range between 10 and 50 years, which doesn’t correspond to the method described in LeGrande where the fresh water pulse is introduced in no more than one year.

    In other words, the largest single flood volume model in Clarke is less than the minimum volume model whose results are discussed in LeGrande. The smallest pulse discussed in LeGrande is 46% larger than Clarke’s “preferred” model for a single flood event, while the largest is nearly three times the volume.

    These differences in this input parameter appear quite substantial, and lead to a couple of questions. First, can you clarify how LeGrande et al arrived at the flood volume range of 78,800 to 157,600 km3 which forms the basis of the discussed results? Second, assuming for the moment that Clarke et al’s flood volumes are correct, would using these figures significantly impact the results of the GCM runs in LeGrande and (potentially) lessen the match between the model runs and the paleoclimate proxies? If not, why not?

    Thanks and best regards;

    [Response: We made no attempt to exactly map the Clarke et al cases to the model experiments. Instead we took ballpark estimates of flood volume and duration and investigated the response. The context for these choices can be seen by examining other estimates of the flood volume (up to 15 Sv yr in von Grafenstein et al, 1999) and previous modelling work (which also used much larger estimates). Our simulations used a range from 1.25 Sv yr up to 10 Sv yr which encompasses Clarke's estimates, but also those of other workers. One of the key results was that the ocean response is not linearly tied to the volume - thus the model cannot be easily used to support any one value (or range) for the outflow. However, the range of results did allow us to calibrate the influence of appropriately-sized NADW slowdowns on the rest of climate, and in particular, on the proxies used to detect the 8.2 event itself. Thus in the results, we show the pooled response, not any particular run, and use that pooled response to come up with a likely change in the NADW (i.e. around 50% decrease). In doing this we are trying to avoid some of the obvious problems, such as the uncertainty in the initial floods and the uncertainty in the sensitivity of the model's ocean. As long as the inputs and the responses are in the ballpark, this calibration is possible. We are not claiming that any one of these model results was close to the 'truth', merely that the ensemble of model responses to reasonable perturbations are consistent with observations. This doesn't for instance rule out a smaller actual flood but a more sensitive ocean. -gavin]

  30. 80
    Ken Robinson says:

    Apologies for the messed up formatting above. It previewed fine. Hopefully the gist is evident.


  31. 81
    Ken Robinson says:

    Re: 79

    Thank you for the response Gavin. Reading through LeGrande, I was under the impression that the range of volumes had been selected only from Clarke as that was the only paper referenced in the particular paragraph, thus my confusion. Thanks also for cleaning up the formatting.


  32. 82
    Blair Dowden says:

    Re #78: Thanks, Richard. I would have preferred to cut the data off at 10 ky to vertically stretch the data of interest around 3 ky ago, but I get the idea. As to which window (500y, 1ky, 2.5ky) most reflects reality, that depends entirely on sea level data, which I don’t have. The searching I did yielded rather inconsistent results.

    Here are my assumptions:

    1) Sea level is a function of temperature.

    2) The intertia of the ice cap delays the response.

    3) The sea level response to a temperature increase follows a bell curve. For example, if the equilibrium period is 1000 years, half the rise in sea level will happen by 500 years, and the rate of increase will be the greatest at this time.

    4) Paleoclimate data suggests the sea level response to temperature rise is 6 meters per degree C, global average. We are measuring Greenland specific temperature, which will be much higher (about 3 times) than global average, suggesting 2 meters per Greenland degree.

    5) I am assuming that Antarctica has little response in this temperature range.

    All this suggests that the averaging you did around the center point is not quite right. You need to shift the curve to the right by one half of the averaging period.

    I don’t know if you want to persue the any further. I appreciate what you did.

  33. 83
    SteveF says:


    If you haven’t already seen it, you might be interested in:

    Ellison, C.R.W. et al (2006) Surface and Deep Ocean Interactions During the Cold Climate Event 8200 Years Ago. Science, 312, 1929-1932.

    Irritatingly, this paper comes from Chris’s PhD work. I get insanely jealous when people get their PhD published in Science or Nature!

  34. 84
    Ulf Erlingsson says:

    The problem of modeling ice sheets is real, and it is largely the fault of the water underneath, just as pointed out. In this article I addressed the issue in relation to a subglacial lake:

    Erlingsson, Ulf (2006) Lake Vostok Behaves Like A ‘Captured Lake’ and May Be Near To Creating An Antarctic Jökulhlaup. Geografiska Annaler: Series A, Physical Geography 88 (1), 1-7. doi: 10.1111/j.0435-3676.2006.00278.x

    This complication may also affect the Pleistocene ice sheets, which makes understanding past relations between ice sheets and temperature problematic. The final word is probably not said yet, just as pointed out above.

  35. 85

    [...] Why is ice flow rate important? An ice sheet affects sea level when it melts faster, or when some of its ice flows into the sea. There is evidence that portions of both the Greenland and Antarctic ice sheets are doing the latter at an increasing rate. The causes are unclear and the changes may not continue in the future, but parts of these ice sheets seem to be far less stable than previously thought. Changes such as these are very difficult to model. [...]

  36. 86

    [...] relevant. The rate at which this is likely to happen is however highly uncertain as we have discussed [...]

  37. 87
  38. 88

    [...] movement has often been a confusing one, but it is an important distinction. As Michael Oppenheimer explains, The issue still seems to perplex many journalists and others because there are two entirely [...]

  39. 89

    [...] that melt rate could occur with today&8217s smaller ice caps and two or three degress of warming., interannual, and long-term elevation change of the east …this to smaller annual [...]

  40. 90

    [...] movement has often been a confusing one, but it is an important distinction. As Michael Oppenheimer explains, The issue still seems to perplex many journalists and others because there are two entirely [...]

Switch to our mobile site