2000 Years of Sea Level (+updates)

Eine Übersetzung in deutsch finden Sie hier.

67 comments on this post.
  1. Kevin McKinney:

    Very interesting indeed. On the graph, the increase in SLR appears to set in somewhere around 1800, which seems a bit early for big anthropogenic effects–though it’s a bit hard to be definite due to the scale and image size. Can you say more about that? (It’s a question the ‘usual suspects’ would raise, to be sure.)

    [Response: Actually, the sea level reconstruction (blue curve) doesn’t show an increase until later (roughly 1900). Its the semi-empirical model (red curve), that shows an increase since around 1800. Since the semi-empirical model has been driven by the global temperature reconstruction, it is by definition consistent with it. The global temperature reconstruction does in fact show a minimum in temperature at about 1800 (clearer if you look at the relevant figure of the paper itself, i.e. Figure 2a), and hence a warming since then. – mike]

    Also, a slight typo–the second sentence of the “How Did Sea Level Evolve?” section should read “There *are* four phases.”

    [Response: Fixed. Thanks. – mike]

  2. Thomas:

    It may sound a bit nitpicking, but water sloshing in a bathtub is a gravity wave, water height in the ocean varying on decadal or longer scales must be totally different physics.

    [Response: Stefan’s analogy here, as I interpret it, is reasonable. There is a strong regional pattern of sea level variation on multidecadal and longer timescales associated with e.g. the differential gravitational pull of growing or decaying continental ice sheets. See e.g. Mitrovica et al (2009). – mike]

    [Response: You’re right about the gravity waves – it was intended as a simple analogy that can’t be carried too far. The ocean is a stratified fluid on a rotating sphere, and the water motions are rather more complex than those in a bath tub. -Stefan]

    I presume North Caroline is far enough south to not be affected by postglacial rebound, but Massachusetts IIRC is.

  3. sidd:

    Thanks for the nice article. Would you care to explain further (or direct me to a reference) the bit about determining the height relative to mean sea level

    ” exploit the fact that each level within the tidal range is characterized by a particular set of organisms that live there.”

    I have become rather pessimistic about the fate of the large ice sheets of late. If West Antarctica were to deliver something similar to MWP1B this century, how much warning would we have before sea levels rose at a couple inches a year for a hundred years ? (Although I suspect it would be more like 0.5 inches a year for a decade, followed by 10 inches a year for a couple decades and then decline.)

    Or have we had all the warning we are going to get ?


  4. Edward Greisch:

    Thanks for the article and the link to the pnas article. I see another hockey stick.

  5. CArsten:

    Nice read, but Stefan can you comment on the criticism appearing in spiegel.de today (http://www.spiegel.de/wissenschaft/natur/0,1518,769424,00.html) regarding that only this US site matches the data of the past ~120yrs? Their main claim is neglecting other non matching site data will make this research not looking very solid…

    [Response: spiegel.de got a number of things badly wrong; I responded to those this morning at KlimaLounge. Regarding your specific question: our Fig. 3 tries to give an overview over previous reconstructions, regardless of their quality. You will see that a couple of them are “all over the place” (i.e. the ones from Israel and the Cook Islands). They do not only mismatch the other data sets but are implausible in themselves, in requiring huge sea level jumps over short time periods. Apart from these two, the only one that does not match the new data, within stated uncertainties, is the one from Iceland. This warrants further investigation whether this is a local effect.
    (p.s. This response applies to the full time series, not just the past 120 years. For the past 120 years the proxy reconstruction nicely matches the Jevrejeva et al. (2006) and Church&White (2006) global tide gauge reconstructions – see above and Fig. 6 of paper.) -Stefan]

  6. Davos:

    “To get a continuous record of good resolution, we need a site with a rapid, continuous sea level rise. Kemp and colleagues used salt marshes in North Carolina, where the land has steadily sunk by about two meters in the past two millennia due to glacial isostatic adjustment. Thus a roughly 2.5 meters long sediment core is obtained. The effect of land subsidence later needs to be subtracted out in order to obtain the sea level rise proper.”

    How did you already know that this location was a site of ‘rapid, continuous, sea level rise’ if this is the sort of thing you were actually trying to test for with your study? And, if you are to “effect of land subsidence”, how do you know what this component is relative to either the post-glacial rebound or the rise in sea-level itself? It seems like at least 2 of the three need to be known in order to identify the magnitude of the 3rd. Is this study working off of assumed or expected/modeled values for certain variables in order to complete the reconstruction?

  7. Nick Barnes:

    Thomas@2: glacial isostatic rebound in North Carolina is essential to this research (the land there is sinking at ~1mm per year, which means that there is a steady background sedimentation rate). See the post and also the article.

  8. Paul S:

    Very interesting piece. Would the steady rise after 1000AD imply that the global average temperature may have been at a fairly stable “high” until about 1400, rather than dropping off after a short medieval peak?

    On that note is there any possibility of using this and similar sea level reconstructions in multi-proxy temperature reconstructions?

  9. chris:

    fascinating research. The link between global temperature and rate of sea level change provides a brilliant opportunity for cross-validation of these two parameters over the last several millenia (one might add-in the relationship between atmospheric [CO2] and Earth temperature in the period before any significant human impact on [CO2]). As these data sets expand (paleo-sea level/paleo-temperature) there’s every chance we can home-in on some really self-consistent interpretations of temp/sea level/greenhouse gas relationships going back several millenia which we be extraordinarily useful as targets for modelling and as predictors of future scenarios.

    I have a couple of questions about the PNAS study:

    (i) Why does the “equilibrium temperature” (the temperature at which sea level change is zero) vary with time (e.g. Fig 4c). Does the pattern of change (warming raises the equilibrium temperature, cooling decreases it), indicate a negative feedback on sea level change (e.g. as land ice melts it requires a little warmer temperature to continue to melt further land ice…and vice versa??).

    (ii) The sea level hindcasts in Fig S6a. This is really a Grinsted et al question, but how were the Moberg (2005) and Jones and Mann (2004) temperature reconstructions normalised to give a global (rather than N. hemispheric) temperature that is required for a valid temperature-sea level hindcast?

  10. Kevin McKinney:

    Thanks–yes, Figure 2A does allow one to see the detail much more clearly.

    The summary graphics of other proxy reconstructions around the world are also worth the time to take a look!

  11. Martin Vermeer:

    chris #7,

    (i) We found it necessary to introduce a component of sea level change that equilibrates on a short time scale (order several centuries) in addition to a “secular” component. This short-term component is essentially what Grinsted et al. used (but they left the secular component out). But, while they defined an “equilibrium sea level” Seq, as a function of temperature, we chose to do it the other way around (which is mathematically equivalent): define an “equilibrium temperature” T0(t) corresponding to a given sea level. You have to make it time dependent so that it, and the corresponding component of sea level, can equilibrate. This is expressed in Eq.2b in the paper. But basically your understanding is right: there is a component of the cryosphere that responds in this way, on a relatively short time scale tau, which is one of the parameters coming out of the hindcast.

    (ii) You would have to ask the authors… my impression is that they just took the NH reconstructions as if they were global. For this figure we just re-plotted their curves from data on Aslak Grinsted’s web site.

  12. J. Bob:

    Some time ago I started looking at sea levels, using tidal gauge info, to see how it correlated to temperature. The So. & East coast of the US was chosen, since that seemed to have the least seismic activity, including uplift. I would have preferred the East coast of S. America, but the data records were not as good as US records. Looking at the records, noted in the figure below:


    seven stations were selected, to form a composite anomaly. These included Galveston, Pensacola, Key West, Charleston, Baltimore, Atlantic City & New York. The composite was filtered with a 10 yr. Fourier filter, and compared a Trend line:

    It was noted that, after ~1915, the trend line held fairly close to the filtered composite, in spite of increasing CO2. The HadCRUT3 global anomaly was also included as a comparison

    An addition, some long term records were evaluated, comparing temperature to CO2. These were from stations that began recording prior to 1800:
    Central England – 1659-2010
    Debilt Netherlands – 1706 – 2010
    UPPSALA (LÄN)Swed. – 1722-2010
    BERLIN (TEMPELHOF), Ger – 1701-2010
    PARIS (14E PARC MONTSOURIS) Fr, 1757-2010
    GENEVE (NASA), Switz. – 1753-2010
    BASEL (BINNINGEN) Swiz.- 1755-2010
    PRAHA (KLEM.-RUZYNE) Czech – 1775-2010
    STOCKH (GML-LAN) Sw – 1756-2010
    BUDAPEST (Hungary) – 1780-2009
    HOHENPEISSENBERG, Ger – 1781-2010
    MUNCHEN, (RIEM FLUGHAFEN ), Ger – 1781-2010
    EDINBURGH (SCOTLAND), GB- 1785- 1993
    WROCLAW (SOUTH WEST), Pol – 1792-2010

    CEL & Debilt were from:

    The rest were from the Rimfrost site:

    The anomaly of each site was computed (1969-1999 base) & a composite average was formed for each year. The data set was then filtered with a 50 yr. Fourier Convolution filter, and compared to the CO2 Mauna Loa & Law Dome (DE08 & De08-2) ice core data.

    The result is shown below:

    Since all the long term temperature data was taken from central & western Europe, the HadCRUT3_NH anomaly was also included.

    A few items noted were:
    On a long term basis, there was little CO2 change, while Europe went through some temperature swings, comparable to the present.

    While the Ave14 & HadCRUT3_NH seem to follow each other (especially the post 1900 rise, 1940 dip & subsequent rise), CO2 seems to have little correlation.

    Ave14 seems to lead the HadCRUT3_NH curve by about 10 years, so we may be in for a NH dip, or are already in it.

  13. JCH:

    I’ll probably ask this the wrong way, but is Mitrovica’s theory – paper cited by mike’s response @ 2 – showing up in the observations?

  14. Martin Vermeer:

    JCH #13: unfortunately not. We actually looked at this; see the Supplementary Material, where we did a sensitivity analysis. Using Mitrovica’s numbers, one finds that in North Carolina, the observed variations in sea level may be, due to this ‘fingerprint effect’, as little as 83% of the global variations; but, dividing the NC record by 0.83 didn’t noticably affect the quality of the semi-empirical model fit. So no, you cannot see this (yet) in the observational data.

  15. Jan Sjoerd de Vries:

    From the link given on the 120m sea level rise after the last ice age it is easily derived that the maximum sea level rise was typically 150cm / century for several thousands of years. The sea level rise for the 21st century mentioned in the update to this paper is also app. 150cm, although given with a large uncertainty.
    A simple question, maybe: Is this a coincidence, or is there another, physical explanation?

  16. Nick Verge:

    From the article:

    “To determine this, we can exploit the fact that each level within the tidal range is characterized by a particular set of organisms that live there. This can be analyzed e.g. from the tiny shells of foraminifera (or forams for short) found in the sediment. For this purpose, the species and numbers of forams need to be determined under the microscope for each centimeter of sediment.”

    This research relies on being able to determine palaeo-waterdepth of a saltmarsh very precisely, and here lies the rub.

    Different species of foram do indeed live at different water depths, in this case in a salt marsh. However, once they die their test can be transported by water, wind and biological activity (other critters) to different water depths, sometimes shallower, from that at which they lived and are then deposited Moreover, foram tests are relatively robust, they survive in sediments and so if earlier sediments are eroded, and the resulting sediment redeposited, any contained forams will be too. Failure to identify contamination of the sedimentary record by allochtonous foram tests, will lead to false water depth estimations for the sediments that host the forams.

    I expect the authors of this research have taken all these complications into account, but this is not stated in this article. I dont disagree with their findings, but would like to have greater confidence and reassurance that they indicate is what the authors interpret them to indicate.

  17. Thomas Lee Elifritz:

    Going back a little further than 2000 years of sea level proxy analysis, here is a very recent paper that some of the readers here might find interesting, and going back a little further here is another quite interesting result. Interesting in that we haven’t quite yet figured out how that water could have gotten from Lake Agassiz into the Champlain Sea, since at the time the route seems to have been securely blocked by the Laurentide ice. Curious.

  18. Gilles:

    two questions :

    – is it right to say that this study doesn’t show any significative influence of anthropogenic, post -1970 warming on SLR, since the SLR reacts mainly with a very large time constant and averages the temperature over a time much longer than 40 years ? so that it is not really justified to associate the current rate with anthropogenic influence, but rather to the exit of LIA ?

    [Response: If you’re interested in what happens in the 20th Century, then this proxy study is not the way to go but rather you should look at the tide gauge data. These are consistent with the proxies (see our graph) but of course more accurate and with higher time resolution, and we have a global set. Thus you should look at the Vermeer & Rahmstorf (2009) study linked above, which correlates the tide gauge record with global mean temperature since 1880 and shows that the modern acceleration of sea level rise is closely related to modern global warming.]

    – if the model is so sensitive that the pre-1000 discrepancy could be removed by only a 0.2 °C change, what is the reliability of the agreement between 1000 and 1500 ? i don’t think that temperatures were known to this accuracy, so is the agreement just adjusted by a convenient choice of the “right” temperature reconstructions ?

    [Response: There was little (or even no) choice here since we needed a global land + ocean reconstruction (i.e. not just northern hemisphere, and not just land – both of these would have been useless). In any case we have not tried any other temperature reconstruction. -Stefan]

  19. CM:

    The auditors are sharpening their, uh, teaspoons. If you can stand the bile, comments over at ClimateAudit about the “preferential treatment” of this paper are good for a laugh. I think McIntyre manages to

    – get the affiliation, degree level, and supervisor of the lead author (tendentiously) wrong (as many of his commenters already pointed out);
    – make a false comparison between distinct PNAS submission routes, as Lindzen, enjoying the member’s privilege of sending a “contributed” paper could not also enjoy the pre-arranged editor that is an option for “direct submissions” (h/t Nick Stokes);
    – misstate the PNAS quarantine period for reviewers and editors due to previous co-authorship (two years, not four);
    – rail about unavailable data (foram dataset used to develop transfer functions) without even following a simple reference to the previously published work of the lead author (h/t R. Norvegicus); and possibly
    – suggest that a downweighting of the proxy data applies to the sea-level reconstruction, when it actually relates to the fit to that reconstruction of a model based on reconstructed temperatures (? – not sure I got that right, though, and would appreciate if someone could clarify).

    As far as I can tell, he’s right that the data from the foram assemblages used in the reconstruction isn’t available on-line – will it be?

    And about that weighing factor: is it important, is it subjective, and if not, can the rationale be explained to Bears of Very Little Statistical Brain like yours truly?

  20. Slioch:

    Just in case you have a libel lawyer handy who is looking for some work:


  21. Martin Vermeer:


    – suggest that a downweighting of the proxy data applies to the sea-level reconstruction, when it actually relates to the fit to that reconstruction of a model based on reconstructed temperatures (? – not sure I got that right, though, and would appreciate if someone could clarify).

    Yes, it only relates to the fit.

    And about that weighing factor: is it important, is it subjective, and if not, can the rationale be explained to Bears of Very Little Statistical Brain like yours truly?

    The reconstructed sea levels as provided by Andy Kemp have their own uncertainties, which we used in the computation; but as so often with these things, those formal uncertainties do not capture everything that is there in the way of error sources(*). This shows up in the Bayesian fit as all ensemble members getting very small posterior likelihood values.

    Furthermore, looking at the original sea level point data it probably contains a bit of auto-correlation. So we have less independent sea level data to work with than we think we do.

    Both circumstances can be accounted for — and yes, this is a bit subjective — by re-scaling the assumed uncertainties for the sea level data. You can play with this for yourself as the code we used should be with the paper (haven’t checked).

    (*) You see the same in geodesy: the per-km error of levelling as estimated from back-and-forth between benchmarks is always smaller than from loop closures, which again is smaller than from network adjustment. Same for triangulation networks and even modern GPS networks: the prior uncertainty of the observations is never the whole story.

  22. Alex Harvey:

    Dear Prof. Rahmstorf,

    This has two parts.


    I assume you have seen responses in the German media at Der Spiegel Online (translated from German by GWPF

    [Response: Is that a legal, authorized translation? It contains a section on supposed “deviations from previous studies” that is not in the English version of the article published by Der Spiegel (and which was corrected in the German version, because there are no such previous studies by Mike or me, as the original article falsely claimed.) -Stefan]

    The translation quotes you as saying, “The new study confirms our model of sea level rise – the data from the past sharpens our view in the future”.

    Then it continues, “But other experts doubt exactly this claim. They see a major problem of the new study in the fact that it is ultimately based only on the finding from the coast of North Carolina. That could be too limited for a statement regarding global developments.”

    I am surprised that no one has commented here on this objection or preempted it with discussion?

    [Response: That is discussed in great detail in the paper, but also in the post above. We estimate that the NC data should track the global mean sea level to within +/- 10 cm (on the time scales we resolve), and so far nobody has challenged this estimate. Specifically, Jens Schröter, quoted in the Spiegel article, has confirmed to me that he also finds this a reasonable estimate. -Stefan]


    The study has also been criticised on various blogs for using [edit] Tiljander lakebed sediment data series [edit]
    I don’t claim to understand the objection but given that it has spread like wildfire on blogs I am surprised that no one has mentioned it here.

    Is this objection valid?

    [Response: No. Just more of the usual deception from dishonest mud-slingers. More on that in short order. -Mike]

    Alex Harvey

  23. Ray Ladbury:

    Re: #18

    Is it right to say that Gilles is constructing his talking points so that they justify his preconceived notions?

    Is it true that he will torture any result within an inch of absolute falsehood just to support those same notions?

  24. ferd berple:

    200+ years ago the British Admiralty undertook one of the largest scientific studies ever conducted. They charted the oceans of the earth in such exquisite detail that except for minor corrections, these chart are still in widespread use today.

    For obvious safety reasons, included in these charts are tens of thousands of “drying rocks”. Rocks that were at the time underwater at high tide, but uncovered or awash at low tide.

    As such, all that one needs to do to confirm sea level rise is to take a British Admiralty chart for your area down to the beach at low low tide. The year the chart was drawn will be noted in the legend. Most likely yours was drawn before the industrial age, before CO2 could have been an issue.

    Look for any cross marked with 4 dots on the chart. Now look at the ocean. Depending on the scale, the rock may be drawn further offshore than it is in reality, for safety reasons to make it visible on the chart. Allow for this when looking at the chart.

    Most mariner’s can already tell you what you will find. The same British Admiralty charts from 200+ years ago, except for minor corrections are still in widespread use today.

  25. CM:

    Martin (#21), thanks.

  26. Kevin McKinney:

    “. . . included in these charts are tens of thousands of “drying rocks”.”

    Which means there are tens of thousands of data points to evaluate. “Your local beach” is hardly enough to confirm anything.

    To do that, you’d have to:

    1) identify when each rock was charted;
    2) identify local tidal extremes both when charted and now;
    3) inspect each rock, verifying its vertical position vis a vis tidal extremes now;
    4) determine if the rock is affected by subsidence or uplift, applying a correction to #3 as needed;
    5) analyze the data to determine probable sea level rise.

    (This would be a rather involved process, given the ‘messiness’ of the data set. How would you do it–probabilistic approach based upon proportion of rocks which don’t dry anymore? Infer mean SLR from the tidal magnitudes of ‘non-drying rocks’ versus the time baseline?)

    Seems quite a bit easier to go with the tidal gauge record; it has its complications, but at least you are starting with reasonably direct measurements of sea level.

  27. CM:

    PS. I don’t see any code with your paper on the PNAS site, though.

  28. CM:

    Ferd #24,

    Sea level rise over the 20th century was 17 cm, give or take a bit. Tidal ranges tend to be quite a bit more.

  29. Jim Bullis, Miastrada Company:

    24 ferd berple

    You seem to speak with certainty, but ‘low low tide’ varies a lot from day to day. So we might expect a lot of hype, both positive and negative from the test you suggest.

  30. Jim Bullis, Miastrada Company:

    I am curious about the dribbly dotted red line. I would have expected it to follow the medieval global warming period from 1000 AD back to 500 AD, or is this no longer believed to be credible? Also, I thought temperatures started cooling beginning about 1100 AD and continued dropping until around 1500 AD.

    [Response: You are confused about the relationship between temperature and sea level in the semi-empirical model. It is the rate of SLR change that is proportional to the temperature. The reason you see such a sharp upward slope in the dotted red line between AD 500 and 1000 is that the relatively warm conditions indicated by the temperature reconstruction is suggesting a fast rate of SLR increase. By contrast, the actual SLR estimate during that same interval is relatively flat, suggesting that temperatures were not as warm as indicated by the temperature reconstruction. That is why, as stated in the paper, the sea level reconstruction appears to suggest that temperatures during AD 500-1000 were not as warm as indicated in the M08 reconstruction (about 0.2C or so cooler on average). -Mike]

    Also, can there be a little more said about the simple model relating atmospheric surface temperatures to sea level?

  31. grypo:

    It appears McIntyre is continuing with the teaspoons, CM. This time he has a powerpoint presentation that may or may not have anything to do with the new paper (I dunno), which may or may not be comparable to the sea level numbers, and an assumption about downweighing (who knows!) He went far enough to know when the pdf might have been created. He’s a real sleuth. I also noticed the PNAS time limit for referees mistake. He actually relied on “free-market energy blog” for that information which came from a letter sent by the PNAS to member in 2008. Oops! He obviously doesn’t care about accuracy anymore, or even the appearance of having it. U Penn, Penn St? So Mann’s graduate student got preferential treatment from the PNAS by being ‘given’ a “prearranged editor” within the “prohibited” window of CoI rules. Yup. Too bad none of that is at all accurate.

    [Response: Hmmm. Given that no graduate student of mine, to my knowledge, had any involvement with this paper at all, its hard to see how said imagined graduate student could have received any hypothetical ‘preferential treatment’ let alone any treatment at all. Very curious indeed. -Mike]

  32. Deep Climate:


    That’s a mixup between U Penn and PSU no doubt, combined with McIntyre’s rather skewed world view. As it happens, I’ve just posted the following comment at my Open Thread # 10.

    Says Steve McIntyre, commenting on the new sea level reconstruction paper by Kemp et al at PNAS, “Climate Related sea-level variations over the past two millennia”:

    A few days ago, PNAS published Kemp et al 2011, a submission by one of Mann’s graduate students. …

    It was certainly generous of PNAS to give a “prearranged editor” to a submission by a graduate student at Penn State.

    Only problem is, Andrew Kemp received his PhD at the University of Pennsylvania, not Penn State, and his PhD advisors included Ben Horton and three others, but not Michael Mann.

    It makes one wonder what else McIntyre got wrong.


  33. CM:

    (Off-topic but totally cool paleo-temperature reconstruction: Body Temperatures of Dinosaurs Measured for First Time)

  34. deston:

    isostatics are one thing. subduction is another. isnt the the east coast sinking, aside from glacial isostatics>tectonic subduction? and maybe more direct to the study- would nutrient flow also affect mash growth- ie, eutrophication- and re…sult in bigger, ‘higher’? deposits? does sea level rise vs. variance of avaialble nutrient loads have different deposit profiles? I see stuff on west coast marshes that wouldnt grow, or grow bigger, if not for the golf course nitrogen drift….how does one correct for change in available nutuients to the salt marsh resultingfrom ag outflows? #freakology

  35. ferd berple:

    “low low tide’ varies a lot from day to day”

    [edit. Ferd a.k.a. “Greg Elliott”, take it elsewhere]

  36. Sean:

    I see code and data is online, and I hope that is an arguement avoided.

  37. Davos:

    “[Response: You are confused about the relationship between temperature and sea level in the semi-empirical model. It is the rate of SLR change that is proportional to the temperature. The reason you see such a sharp upward slope in the dotted red line between AD 500 and 1000 is that the relatively warm conditions indicated by the temperature reconstruction is suggesting a fast rate of SLR increase. By contrast, the actual SLR estimate during that same interval is relatively flat, suggesting that temperatures were not as warm as indicated by the temperature reconstruction. That is why, as stated in the paper, the sea level reconstruction appears to suggest that temperatures during AD 500-1000 were not as warm as indicated in the M08 reconstruction (about 0.2C or so cooler on average). -Mike]”

    What is the basis that makes the suggestion that the temperatures were not as warm as indicated during AD500-1000 more valid than perhaps the M08 reconstruction suggesting that the SLR estimate instead is incorrect?

    [Response: That’s a very fair point. Both alternatives, in my view, are equally viable. Future work will hopefully better pin this down. -Mike]

  38. KR:

    I did not see the code or data on the PNAS site (aside from the supplemental), although I’ll admit I’m not completely familiar with PNAS. Where would this code and data be located?

  39. Martin Vermeer:

    Davos #37, my perspective differs a little from Mike’s here. I do not consider it likely that the “knee” at 1000AD and the downturn before that are realistic. It would imply or suggest that sea level around year zero would be a metre or more below present — something that seems to be ruled out by the classical Roman fish tanks result, and so large that, if true, we would probably know about it. Also, we see no similar “knee” in any of the other sea level curves in Figure 3 that appear good enough to make such a statement — Massachusetts, Louisiana.

    There are alternatives to our proposition of a -0.2K temperature offset — which seems a bit large against the apparent quality of this reconstruction, even this far back — but they would be speculative at this point.

  40. Martin Vermeer:

    KR #38: there would be an obvious link in the main text, and it’s not there. We’re working on it ;-)

  41. Gilles:

    #23 Ray : I don’t think Stefan’s answer really contradicts my remarks ; that the conclusion that “20th-Century sea-level rise on the U.S. Atlantic coast is faster than at any time in the past two millennia.” has nothing to do with anthropogenic influence since it would already have been deduced from pre-1950 data, for instance. So if it proves anything, it’s only that natural variability CAN produce such variations in the SLR. And that the comparison with a model based on temperatures relies on inaccurate temperature reconstructions, and fails to reproduce the data at some point.

  42. Ray Ladbury:

    Gilles, thank you for proving my point. Yes, if you take any single datum, you might be able to explain it by natural variability. If you take all the data…not so much. It is why you focus only on cherrypicked results interpreted in isolation of physics and the rest of the evidence.

  43. Hank Roberts:

    >fred berple … British Admiralty

    “Accumulated sealevel rise since Bligh is less than 30cm
    – wave height on calm day.”

  44. JCH:

    One of the SLR papers that is making the rounds is Unal and Ghil, 1995, which found sea level rose an average of 1.62 mm/yr (+ or- .38) between 1807 and 1988 (averaged over 181 years.)

    Does that paper have a number for just the 19-Century years?

  45. folbec:

    out of topic for this tread but interesting :
    “Why volcanism isn’t the source of increasing carbon dioxide emissions”


    One year of volcanic activity releases the same CO2 as the following human source:

    A state such as Florida, Michigan or Ohio.
    ~13 times less than land use changes (3.4 gigatons)
    ~11.5 times less than light-duty vehicles (3.0 gigatons)
    ~5.3 times less than concrete production (1.4 gigatons)
    ~2 dozen 1000 MW coal-fired power plants (2% of the world’s coal-fired electrical generation)
    Or, roughly the same CO2 emissions as Pakistan, Kazakhstan, Poland or South Africa.

  46. Terry:

    Mike #37. “It is the rate of SLR change that is proportional to the temperature” How can that be from a purely physical point of view. Surely one would normally expect a linear relationship of SLR and temp.

    [Response: Ummm, no. One quantity represents an approximate integral of the other. That this is true empirically is obvious from a comparison of the historical observations. It isn’t that hard to review this sort of stuff on your own. Read for example the referenced papers by Rahmstorf, Rahmstorf and Vermeer, etc. -Mike]

  47. Noureddine:

    The big question arises about the maps of the famous seas: the Red Sea southern straight, the Black Sea at the Bosphorus, The mediterranean at Gibraltar. How about the Bering sea? Your findings will shed light on historical migrations issues.

  48. M:

    Terry #46: “Surely one would normally expect a linear relationship of SLR and temp.” I think the key point here is the difference between equilibrium and transient: in equilibrium, the rate of change of SLR will, by definition, be zero, and this state can be true at any number of temperatures. However, in the transient the rate of SLR will be roughly proportional to the difference between the current temperature and the temperature at which the current SLR would indeed be in equilibrium…

    (mind you, I’m a little surprised that this relationship holds over a period when there is both decreasing and increasing absolute SLR)

  49. l. david cooke:

    Hey All;

    This point may have already been made; however, I think it is important enough to reitterate.

    Though examiniation of old tidal gauges are good, many were painted wood and metal with a life of between 7-15 years. If anyone is truly interested in how much the N. Atlantic region has risen in the last 50-75 years they only need to inspect bridge pillars and concrete seawalls in protected bays, lagoons, marshes, river mouths and marinas scattered up and down the southern US East Coastal region. Going further, if anyone has an interest in surface acidification they can also inspect these same submerged concrete structures for calcium errosion.

    Generally, though the biota may change or the waters in these areas be subject to various changes in pollutants, generally the waterlines are clearly indicated and samples in the top 1/3rd can be compared to a basin control either at the “basin bottom” or via underground cores.

    That there are thousands of potential sites, the potential for robust evidence is overwhelming. This evidence would certainly reduce the breadth of concern over conflicting references.

    Similar to Dr. Mann’s tree ring analysis of the ’70’s (Popular Science ran the initial popular press article, if I recall correctly.) it should be possible to extend the modern era evidence via coastal chalk and limestone deposits. I’ll leave that up to the experts to discuss.

    [Response: Hmmm. During the 70s I was mostly focused on t-ball practice and comic book collecting. I think you must have someone somewhat…well..more senior than me in mind here. -Mike]

    Dave Cooke

  50. Septic Matthew:

    36, Sean: I see code and data is online, and I hope that is an arguement avoided.


  51. Karl Quick:

    Fifty years ago I worked at a lab that did submarine launched ICBM targeting. Calculations included tides, sun/moon gravity, air/water temps, and location at point of launch… virtually anywhere in the world’s oceans. The gravitational model of the earth required variations due to the density and depth of stone layers, bays, swamps, etc. The information was derived by analysis of thousands of orbits by hundreds of man-made objects. The results were proven to be quite precise.

    If you truly believe there is a significant rise in sea level (and compensation required by “glacial rebound”) then certainly these classified datasets should show the impact of 50 years of warming.

    You may want to submit a suggestion to NRL or other Naval R&D facilities to review this data and compare with modern ICBM targeting data/models.

    IMHO, you are talking about tracking nats in a huricane. The changes over geologic time are so much more massive, the REAL news is that we’ve seen rediculously little variation over the past 4000 (10,000?) years and THAT is what we should be surprised and shocked about.

    Such stability suggests some form of natural feedback mechanism that has compensated for far more severe changes in inputs than man-made CO2.

    Rather than a focus on detecting the impact of MM CO2, try to understand how the much greater stimuli of the past made so little relative impact on life on this planet.

    Humankind is far, far too self-centered and self-absorbed; nature does not dance to our tune, no matter how much it pumps up our ego to think so.

    [Response: I might suggest that whether humanity is affecting nature (and how could it not be?) is more a question of investigating the degree, than of a priori assumptions that it isn’t possible regardless of it’s effect (or not) on our (collective?) ego. – gavin]

  52. Neil White:

    Karl Quick (#51)

    The gravitational models used can’t have been that good, because the US Navy launched the GEOSAT mission in 1985 specifically to map the marine geoid (and gravity field) as the fields they had weren’t good enough for much of the world’s oceans.

    This data has now been declassified because gravity data from civilian missions (e.g. ESA’s ERS-1) made the continued classification of it pointless. There is no need to ask the NRL for this data.

    More recently, NASA’s GRACE and ESA’s GOCE gravity missions have improved gravity fields, both time-varying (GRACE) and time mean (GOCE).

    The GRACE data shows the increase in mass of the oceans from ice melt.

    Neil White

  53. Kevin McKinney:


    “. . . some form of natural feedback mechanism. . .” that was effective over the past 4000 (10,000?) years but not previously?


  54. Chris O'Neill:

    Karl Quick (#51):

    Rather than a focus on detecting the impact of MM CO2, try to understand how the much greater stimuli of the past made so little relative impact on life on this planet.

    Could you let us know what these much greater stimuli of the past (10,000 years, i.e. after the last ice-age) are? Strange that you’re so confident they existed while I’ve never heard of anything “much greater”.

  55. dhogaza:

    Karl Quick:

    Rather than a focus on detecting the impact of MM CO2, try to understand how the much greater stimuli of the past made so little relative impact on life on this planet.

    Don’t you like it when an amateur tells professional scientists what they should be researching. Talk about condescending.

    As Neil White points out, though, Karl’s time might be better spent getting up-to-date on the data that’s available rather than pontificating on how scientists should do their job …

    The changes over geologic time are so much more massive

    Yes, over a small fraction of geologic time we’ve seen the evolution of hominids and over a tiny, almost imperceptible period of time, the evolution of humans followed over an eyeblink of geologic time, the growth and flourishing of human civilization.

    Yet you’d have us marvel over the fact that pangaea once existed and stuff like that rather than concentrate on our human future.

    Based on falsehoods like:

    try to understand how the much greater stimuli of the past made so little relative impact on life on this planet.

    Little relative impact on life? You mean those huge extinction events like that which led to the demise of dinosaurs?

  56. l. david cooke:

    RE: 49 comment

    Hey Dr. Mann,

    My apologies, it was a 40 yr old memory. The images that come to mind are, a shaggy brown bearded researcher, (a grad. student, maybe, working with a European Professor/Scientist) standing in front of a large fresh redwood cutting. Then there was an image of a cross cut tree sample with reference arrows, maybe it was a rare fold out page… As it would fit with Dr. Furgeson’s C-14 dating paper, I would be more inclined to think the bearded man was a young Mr. Hughes, now Dr. Hughes, though I can find no references of their working together.

    Again, my apologies, though I am curious as to why the idea of CO isotope or calcium leaching was not considered in the prior research, are there too many degrees of freedom to establish a robust evidence base or is there no reliable means to track calcium errosion or the acidic signature responsible? Never mind, if it were worthy of analysis it would already have been done.

    Again my thanks and appreciation for your tolerance over the years, for an “old man’s” meanderings and rants…

    Dave Cooke

  57. JCH:

    Wow, it really is true. More than one author worked this paper! (Interesting comments about paper in the linked article.)

  58. Davos:

    Another thing I was thinking about…

    I was at a weather conference in New England a few years back, and we heard a talk/presentation there by a company who’s primary function was to forecast eddy-currents and such so that those involved in yacht races could take advantage of them as they swirl current toward (or away) from their intended directions.

    One of the identifiers for those eddies was noticeable water temperature perturbations within these swirls (sometimes as great as 2-4C as I recall).

    Anyway, could the presence of these eddy swirls (and the sea-surface temperature garbling that goes along with them) have an impact on the sea-surface temperature record by ships traveling unbeknownst through them over the last 50-100 years? Or, is it just assumed that all of this sort of thing evens out in the end such that it can be considered a negligible entity? (I acknowledge this may be slightly OT, since SST measurement is not specifically part of this post about SLR and Temperature corelation…But if there’s someone that can help, or point me to better place, that would be fine).

    [Response: These eddies are a big part of the variability seen by shipboard SST measurements and indeed of the SST variability more generally. They last longer than eddies in the atmosphere, so necessary sampling is not as high frequency, but in areas like the Gulf Stream extension, the Aghulas retroflection (off South Africa), or the Antarctic Circumpolar current, they are really important. – gavin]

    [Response: Getting at your question in a slightly different way, one thing that is peculiar about eddies embedded in the western boundary currents (there are analogies for the atmosphere too, e.g. the jet stream), is that they exhibit the property of “negative viscosity”. That is to say, the transport of energy is (over some range of wavelengths) from shorter to longer wavelengths, the opposite of what is typically seen in a fluid, where energy is dissipated by the small-scale turbulent structures. The eddies actually work to maintain the mean large-scale flow (e.g. the Gulf Stream). This was noted decades ago by my academic grandfather, Victor Starr of MIT, who in fact even wrote a book about all of this entitled “Physics of Negative Viscosity Phenomena”. So the short answer to the question is, the properties of these eddies are not only important, but in fact they are fundamental to the large-scale planetary circulation. And there are things about them we still don’t fully understand. – Mike]

  59. M:

    A further follow-up to Terry (#46) and my response (#48): actually reading the paper (what a concept) made me realize that the semi-empirical model contains two important terms: one that is relative to an absolute equilibrium T (T0,0), and the other than is relative to T0(t), where T0(t) depends on the historic temperature…

    T0,0 ranges from 0 to -10K (or more) based on a Bayesian analysis: if it is less than about -0.4 K, that would imply that the long-term term of the semi-empirical equation is contributing to SLR for the entire reconstruction. That would also imply that (T-T0(t)) must be negative during the pre-900 period when SLR=0… would a plausible physical explanation be that the deep ocean and ice sheets are still responding somewhat to the post-glacial temperature increase (eg, T-T0,0>0), but that the faster components of SLR like the surface oceans and glaciers were actually responding to the decrease in temperature since the early Holocene?


  60. Davos:

    #57 (If I may…)

    So does this sort of phenomenon affect the paleo-reconstruction of global SSTs? I’m assuming there’s no way to know if boat-bucket measurements were conducted in eddy anomalies back-in-the-day.

    Is it thought that the number of occurrences are probably evenly distributed (toward hot and cold) and thus isn’t worrisome? Or are these eddies particularly long-lasting enough to simply assume they ‘do’ or ‘can’ represent grid-smoothed temperature accurately for a particular time interval in the olden days?

    [Response: The presence of eddies is taken into account in the estimation of uncertainties in gridded SST products because they express themselves in the sampling variability estimated for a given location. The error bars in state-of-the-art SST compilations take into account such sampling uncertainties, and indeed they become larger back in time, especially the earliest decades (1850s-1870s) in part due to the fact that there is substantial eddy variance. This is especially true e.g. in boundary current regions, and other areas of large eddy variance as mentioned by Gavin earlier. A good discussion is provided by Brohan et al (2006). – Mike]

  61. Ryan:

    Looking at Figure S6 of the supporting document I notice that the decline in sea level 2000 back to 1900 is measured by tide gauges as 200mm. For the new proxy it is measured as about 300mm. This implies that the new proxy is 50% out in its ability to hindcast after just 100 years. In that case, would it not be reasonable to suggest it is likely to be 1000% out over a 2000yr period?

    [Response: Only if you don’t know what % means. (i.e. a linear trend that 50% off another trend in one period, remains 50% off if the trend is extended). – gavin]

  62. Andy:

    Looking at the sea level curve. It appears this study would preclude a sea level stand in the mid-Holocene that is equal to or higher than present. I can’t imagine that sea level would have been higher than today, dip 0.4m from present, and then rise again. That seems like too much variation.

    A mid-Holocene high stand is something that’s been debated for quite a while in my neck of the woods as it would say a lot about barrier island and tidal marsh formation and migration.

    The standard model used to be a slow, steady rise up to about 6,000 YBP and then steady till modern times. I’ve seen the 6,000 YBP date chipped away over the decades to about 4,000 YBP, but this study seems to chip it down to almost nothing (500 YBP). Of course, I work in the microtidal environment of the Texas coast where 0.2m of sea level rise is something we fight about at every conference after a couple of beers.

    Would you agree that the mid-Holocene high level stand is dead?

  63. Martin Vermeer:

    M #58:

    would a plausible physical explanation be that the deep ocean and ice sheets are still responding somewhat to the post-glacial temperature increase (eg, T-T0,0>0), but that the faster components of SLR like the surface oceans and glaciers were actually responding to the decrease in temperature since the early Holocene?

    Very perceptive, M. That is pretty much what I believe is happening.

  64. Filip Hondekyn:

    I wonder: have vertical tectonic movements been considered?

  65. One Anonymous Bloke:

    Eric #65, Luckily there are people who regard difficulty as a challenge rather than an excuse for inaction.

  66. Jacob Ballard:

    This is some great, and scary information. It obviously shows a correlation between our emissions and the rising sea level. Although people will always say that correlation is not cause and effect, the evidence is far to strong to ignore the fact that sea levels are rising at a rapid rate.

  67. Ray:

    Working with forams in UK salt marshes, the big problem is the whole teasing apart of isostatic/eustatic influences – I noted the correction for subsidence in the paper, but are you confident in not accounting for other sources?
    Can I also please ask what was your take on the transfer function error? Things like infaunal species, bioturbation, non parametric salinity tolerances etc all can produce quite significant error, and I am impressed that in this data set it was less than 10cm!