Dongge Cave

Dongge Cave is a very long speleothem in southeast China, which is held to provide evidence on changes in the Asian monsoon.

It was most recently considered in Bao Yang et al (QSR 2007) in juxtaposition with the Dasuopu ice core, Oman speleothem S3 and the RC2730 Arabian Sea core showing G bulloides percentage. (This latter core was an important contributor to the Moberg reconstruction and has been discussed on several occasions – G bulloides are evidence for cold water and are interpreted as evidence of increased upwelling of cold water, which in turn is considered to be evidence of wind speed, and thus monsoon activity.)

Kim Cobb has also just reported a new speleothem in Borneo, somewhat the same part of the world, which I’ll try to look at soon. I’ve not previously looked at the Dongge Cave speleothem. Its results have been properly archived at WDCP, much facilitating examination.

First, here is a published image of the dO18 values from Dongge cave (this can be determined to be a plot of stalagmite DA). This shows a strong decline in dO18 values following the Holocene Optimum, which was reversed around 1400, after which there is an increase in values reversing the previous trend.

dasuop33.jpg

Next here is a graphic of the Oman speleothem S3, referred to in the Yang et al article. (This information is not archived; I’ve written to Stephen Burns, an associate of Raymond Bradley, suggesting that he archive his interesting data.) In this case, you can see a very distinct Holocene Optimum, with a very rapid increase in values emerging out of the LGM, followed by a long decline following the Holocene Optimum, with recent discontinuity in the records.

dasuop25.gif

The next graphic compares the Dongge DA speleothem shown above to the Dongge D4 speleothem, which goes back to the LGM and to another Arabian Sea G bulloides series (723A) which, unlike RC2730, goes back to the Holocene Optimum.

The longer Dongge speleothem (D4) shows the decrease in dO18 values as you go back to the LGM, which is very marked in this series as well as the Oman speleothem. However, the LGM transition to higher dO18 values is dated about 2000 years earlier for Dongge Cave than for Oman speleothem S3.

Is this a real difference in dates reflecting “regionalization”? Or is it an artefact of different dating? (and these speleothems are benchmarked with U/Th methods that can have pretty wide error bars?

I find it difficult to believe that there would be 2000 years difference in the dO18 regime changes and I dare say that the events are coeval, notwithstanding the author datings. (Obviously this can lead to the Crowley-esque perception of “regional” differences in timing, which may not be real.)

dongge37.gif

Dongge Divergence

One last point about the Dongge Cave spelothem series – that is not obvious from the first plot of this series shown above.

Here is a blown up version of the series for the last millennium (arranged in this case with the most recent value on the right.)

In this case, the maximum values were reached in the 1930s, and there has been sharp decrease in the last half of the 20th century – another “Divergence Problem” if you will.

I’m not saying that this shows that temperatures have decreased; or that this somehow invalidates the perception that these speleothem dO18 values show changes from the LGM to the Holocene Optimum and back again. I don’t know what it means – except that it deserves to be talked about.

My attention was drawn to this discrepancy by Yang et al 2007, who said that this series and Dasuopu were “very similar” – the sort of qualitative arm-waving that is so typical of climate science. Given that Dongge Cave dO18 goes sharply down in the last half of the 20th century, I don’t notice anything “similar” at all in the 20th century portion, let alone “very similar”.

dasuop35.gif

Obviously these time series show a very marked Holocene Optimum in both Oman (17N) and southeast China (25N), both of which give evidence on tropical conditions. They are interesting to juxtapose against statements from realclimate and NOAA, discussed previously here. NOAA stated:

Paleoclimatologists have long suspected that the “middle Holocene” or a period roughly from 7,000 to 5,000 years ago, was warmer than the present Image courtesy of Kerwin et al., 1999. Click here for larger viewing image. day. Terms like the Alti-thermal or Hypsi-thermal or Climatic Optimum have all been used to refer to this warm period that marked the middle of the current interglacial period. Today, however, we know that these terms are obsolete…

Obviously the speleothem data from Oman and Dongge shows that the Holocene Optimum was very marked in these areas and not just the NH extratropics.

If the Team wish to cavil that this information does not bear on annual temperatures, they must explain why, for example, the same type of data (e.g. percent G bulloides in the Arabian Sea, speleothem dO18) is permitted as a temperature proxy in Moberg et al 2005. All in all, the speleothem data certainly doesn’t support the claim that the term is “obsolete”.

51 Comments

  1. Posted Sep 30, 2007 at 11:34 AM | Permalink

    There is an abundance of evidence that lower lattitude isotopes are very complex in nature. Not only temperature but rain out, seasonality and rate of precipitation are important factors, which may dwarf temperature changes.

    some references:

    http://www.agu.org/pubs/crossref/2005/2005JD006022.shtml

    http://www.geo.umass.edu/climate/mathias/research/nsf99.html

  2. richardT
    Posted Sep 30, 2007 at 12:13 PM | Permalink

    If the science of palaeoclimatology had developed in tropics, rather than Europe, the Early Holocene would be remarked upon for its precipitation rather than its temperature. Evidence for higher summer temperatures abounds in Northern Europe, but is scant at lower latitudes. Indeed the Arabian Sea bulloides suggests cooling.
    The bulloides record was included in the Moberg compilation as a teleconnected response to warming in Asia, not to represent local conditions. I don’t see any contadiction.

  3. brian
    Posted Sep 30, 2007 at 12:18 PM | Permalink

    Where is the “remarkable similarity” quote in Yang et al. (2007)? I assume it is a direct quote from the paper since you put it in quotes?

  4. Posted Sep 30, 2007 at 12:20 PM | Permalink

    re #2

    If the science of palaeoclimatology had developed in tropics, rather than Europe, the Early Holocene would be remarked upon for its precipitation rather than its temperature.

    You may extent that to Siberia, where it led to the extinction of the megafauna.

  5. Don Keiller
    Posted Sep 30, 2007 at 3:43 PM | Permalink

    re#2 “The bulloides record was included in the Moberg compilation as a teleconnected response to warming in Asia, not to represent local conditions. I don’t see any contadiction”.

    Richard, can you explain this statement on non-Team speak please? Or are you just extracting the Michael?

  6. Steve Moore
    Posted Sep 30, 2007 at 5:43 PM | Permalink

    None of the trends shown in the speleothem directly relate to temperature

    The same can be said for tree rings.

    they instead relate to precipitation.

    No argument here. And I’ll take precipitation records as a proxy for temperature any day.

  7. Steve McIntyre
    Posted Sep 30, 2007 at 6:19 PM | Permalink

    #5. OK.

  8. Steve McIntyre
    Posted Sep 30, 2007 at 6:22 PM | Permalink

    #3.

    The comparison between the Dasuopu ice core d18O, the stalagmite d18O from Dongge Cave and the marine upwelling record offshore Oman is shown in Fig. 2. Decadal-scale variations as well as long-term trends of these three series are very similar.

    OK, they said “very similar” – not “remarkably similar”. I’ll edit accordingly.

  9. Steve McIntyre
    Posted Sep 30, 2007 at 8:39 PM | Permalink

    I have a few posts on this proxy: http://www.climateaudit.org/?p=650 http://www.climateaudit.org/?p=423 http://www.climateaudit.org/?p=93

    Moberg describes it as follows:

    A combination of two marine sediment records from the Arabian Sea in which the percentage of the foraminifera Globigerina bulloides reflects the extent of ocean up-welling, which is determined by the strength of monsoons, which in turn indirectly reflect both summer and winter large-scale temperature changes through the differential seasonal heating and cooling of the Asian continent and surrounding oceans. Although this record reflects temperatures only indirectly, it was included to improve the balance in the geographical distribution of proxy sites.

    If the purpose of the proxy is to measure high-latitude temperature by teleconnection, then why use it? There is no shortage of such proxies. Why insert a poorly dated splice with nonlinearity problems?

    Moberg’s justification is that it improves the geographical balance, and its tropical location on a map certainly gives the impression that it is a tropical proxy. But it isn’t.

    Also note the total inconsistency in the handling of this proxy by the Team. Soon and Baliunas cited G Bulloides from Cariaco as evidence of a warm MWP. David Black reviled them for this, stating that it was merely an indicator of wind speed and creating an issue of it. But when Moberg used it as a temperature proxy, Black said nothing, nor did any other paleoclimatologist. If it was no good for S&B, it was no good for Moberg; or if it was OK for Moberg, it was OK for s&B. SEe http://www.climateaudit.org/?p=650

  10. Harry Eagar
    Posted Sep 30, 2007 at 9:23 PM | Permalink

    My experience with proxies is limited mostly to historic price series, and I have plenty of problems with those.

    However, the professional economic historians have pretty loose standards about what they will accept as a proxy. So be it.

    But this proxy, whatever it is, wouldn’t pass their sniff test.

  11. brian
    Posted Sep 30, 2007 at 10:43 PM | Permalink

    …the sort of qualitative arm-waving that is so typical of climate science

    Is it the use of proxies in general that are gettin’ people upset? Or, just this proxy? Or just this location? Just this paper? The fact that qualitative evaluation is involved? All of the above?

    Given that Dongge Cave dO18 goes sharply down in the last half of the 20th century, I don’t notice anything “similar” at all in the 20th century portion, let alone “very similar”.

    But, isn’t their statement about similarity regarding decadal and longer-term trends when considering the entire record in question (last 1000 years), not just the last 50 years? (i.e., their Fig. 2, the figure they refer to for the ‘very similar’ statement)

    At the very end of their paper they discuss how their results are at odds with other work at much shorter time scales (w/r/t snow cover data), and how more work needs to be done to reconcile this.

  12. Mark T
    Posted Oct 1, 2007 at 12:33 AM | Permalink

    re#2 “The bulloides record was included in the Moberg compilation as a teleconnected response to warming in Asia, not to represent local conditions. I don’t see any contadiction”.

    Richard, can you explain this statement on non-Team speak please? Or are you just extracting the Michael?

    Teleconnected: the process by which certain conclusions can be drawn in climate science without supporting evidence. Teleconnections reveal themselves as a sort of spooky action at a distance with no other basis in science than furthering the accepted dogma. For example, the temperatures in Europe are often known to affect the growth patterns of BCPs that reside in Colorado, CO. That their ring widths do not correlate well with _local_ temperatures is immaterial because they are _teleconnected_ to the temperatures in other parts of the world.

    Mark

  13. Willis Eschenbach
    Posted Oct 1, 2007 at 2:31 AM | Permalink

    brian, you ask (quite reasonably);

    Is it the use of proxies in general that are gettin’ people upset? Or, just this proxy? Or just this location? Just this paper? The fact that qualitative evaluation is involved? All of the above?

    Don’t know about anyone else, but some of the things that get me upset regarding the use of proxies are:

    1) Lack of clear ex ante proxy selection criteria.

    2) Use of proxies that are known to have problems (e.g. stripbark tree rings).

    3) Use of unarchived, “gray” versions of proxies.

    4) Use of proxies with upside-down quadratic response curves to the forcing variables.

    5) Use of proxies supposedly connected to global temperature fields but demonstrably disconnected from local temperature fields.

    6) Use of a proxy for two explanatory variables (e.g. temperature and precipitation) without explaining how the confounding variable is removed.

    7) Proxy-based calculations with incorrect or missing error estimates.

    There’s more, but that’s a start. I have no problem with using a proxy, in fact, we do it all of the time — for example, we use the expansion of a column of mercury as a proxy for the temperature. I have loads of problems, however, with the way proxies are often used in climate science.

    w.

  14. Steve McIntyre
    Posted Oct 1, 2007 at 9:18 AM | Permalink

    the sort of qualitative arm-waving that is so typical of climate science

    I can’t think how often people say that two series are “remarkably similar” without quantifying the similarity or showing that the “Similarity” is statistically significant allowing for autocorrelation. Different issue than simply being a “proxy” – though there are lots of interesting issues there as well.

  15. moptop
    Posted Oct 1, 2007 at 10:03 AM | Permalink

    The funny thing about the exinctin of megafauna. Except in Africa, where megafauna evolved with humans, the rest of the world’s megafauna first encountered humans in our modern form complete with effective weapons. Megafauna extinction has no relationship to climate except where climactic conditions enable human dispersal.

  16. Don Keiller
    Posted Oct 1, 2007 at 10:17 AM | Permalink

    re#13, ah I see, Richard in #2 was taking the Michael.
    Thanks, Mark.

  17. richardT
    Posted Oct 1, 2007 at 10:35 AM | Permalink

    #10 Site specific calibration is essential when invoking teleconnected responses. The Cariaco bulloides record is negatively correlated with N. Atlantic SST during the instrumental period (and has zero correlation with local SSTs), so it makes no sense to ascribe the high thirteenth century bulloides concentrations to warmer temperatures. I assume that you agree that the Soon and Baliunas paper was pretty dire.
    The Arabian Sea bulloides record has a strong correlation with high latitude temperatures over the Holocene. Assuming that this teleconnection is stable, (which could be checked with reanalysis data), it is reasonable to use it as an indirect predictor of northern Hemisphere. It’s not the perfect record to use, but you would complain even more if they only included the old familiar dendroclimatological data.

    #15
    I agree that too often palaeoclimatic correlations are done by eye. I tend to ask what the r-squared is if I suspect this is the case, and don’t usually get an answer.
    A bigger problem is to assign a reasonable significance level to the correlation of two noisy, autocorrelated unpaired timeseries. This is the joyous task I have to do this evening. Any suggestions? A statistically valid solution should be publishable – somewhere like CP.

    #11
    Thank you for your useful contribution.

    #13
    Teleconnection: a simple climatological concept that some sceptics are unwilling to grasp.

  18. Stan Palmer
    Posted Oct 1, 2007 at 10:44 AM | Permalink

    re 19

    #13
    Teleconnection: a simple climatological concept that some sceptics are unwilling to grasp

    Teleconnection is usually used to explain the validity of a proxy that does not correlate with local temperature. For example, bristlecone tree rings not correlating with local temperature but being used as a proxy for North American temperature. This explanation is quite distinct from the explanation given in your link. Is the usual explanation incorrect or are there additional subtleties that have not been captured in the link’s description.

  19. Mark T.
    Posted Oct 1, 2007 at 10:52 AM | Permalink

    #13
    Teleconnection: a simple climatological concept that some sceptics are unwilling to grasp.

    No, it is nonsense that has no basis in nature, used to explain something that cannot otherwise happen. “Skeptics” do not grasp the concept because they have the force of will to not be persuaded by pseudo-science such as what is being used to explain tree-ring growth that does not correlate well with local temperature. What is explained in the link still requires direct connection between entities, and is an entirely different concept.

    Mark

  20. Mark T.
    Posted Oct 1, 2007 at 10:55 AM | Permalink

    re#13, ah I see, Richard in #2 was taking the Michael.
    Thanks, Mark.

    Of course he was. It is impossible to quote real scientists while making such nonsense claims.

    Mark

  21. Michael Jankowski
    Posted Oct 1, 2007 at 10:58 AM | Permalink

    The Cariaco bulloides record is negatively correlated with N. Atlantic SST during the instrumental period (and has zero correlation with local SSTs), so it makes no sense to ascribe the high thirteenth century bulloides concentrations to warmer temperatures.

    …yet this rule doesn’t apply when it comes to the divergence problem of temperature with tree ring proxies in the 20th century.

  22. David
    Posted Oct 1, 2007 at 10:59 AM | Permalink

    #19: Teleconnection: How clouds and animal shapes are connected.

  23. Michael Jankowski
    Posted Oct 1, 2007 at 11:03 AM | Permalink

    The bulloides record was included in the Moberg compilation as a teleconnected response to warming in Asia, not to represent local conditions. I don’t see any contadiction.

    So what is the physical mechanism for this “teleconnected response?”

  24. Steve McIntyre
    Posted Oct 1, 2007 at 11:12 AM | Permalink

    #18. Richart T, my beef is this: if pct G bulloides is a good proxy in one place, it should be a good proxy in another place. You can’t opportunistically pick them based on their 20th century performance.

    My beef with the dendros is only partly the dendro series, but it’s vehemently against their opportunism. When the Polar Urals Update didn’t go his way for the MWP, Briffa switched to Yamal. If high-latitude treeline spruce chronologies are supposed to be a temperature proxy, then you have to consider the entire population; you can’t pick after the fact based on ones that went up. Likewise with G bulloides.

    The other issue is that “teleconnections” is used as an excuse for data mining. Mann’s methodology, taken at face value, purports that the (say) the 6th PC series of Stahle’s SWM network is teleconnected to the 11th PC of NH temperature. This is absolute bilge. If you do thousands of correlations, you will get some that are “significant” if the data mining is not allowed for.

  25. Steve McIntyre
    Posted Oct 1, 2007 at 11:14 AM | Permalink

    The bulloides record was included in the Moberg compilation as a teleconnected response to warming in Asia, not to represent local conditions. I don’t see any contradiction.

    Except that he already had the Yang composite for this region. Also its location on the map makes people think that it is representing the tropics.

  26. Michael Jankowski
    Posted Oct 1, 2007 at 11:20 AM | Permalink

    Re#2, just a quick google, curious about your thoughts…

    (1) http://adsabs.harvard.edu/abs/2005AGUSMPP22A..03P – “…These results provide strong evidence for a basin-wide atmospheric teleconnection between Arabian Sea and North Atlantic climate on sub-Milankovitch timescales.”

    (2) http://drs.nio.org/drs/bitstream/2264/467/1/J_Geol_Soc_India_68_379.pdf – “…This suggests a link between high-latitude climate on the one hand and monsoon upwelling strength in the Arabian Sea and tropical climate changes on the other.”

    (1) found a teleconnection to the North Atlantic, and (2) found a teleconnection with high-latitude climate, specifically mentioning Greenland and the North Atlantic. But you would suggest there is no telecommunication with high-latitudes aronnd Europe, and that Moberg was justified in finding a telecommunication with Asia.

    What an intricate puzzle.

  27. Steve McIntyre
    Posted Oct 1, 2007 at 11:28 AM | Permalink

    PErsonally I think N-S movement of the ITCZ may be a very important organizing principle for centennial climate variations. I’ve previously commented favorably on such comments in (say) Newton et al 2006. Look, I thnk that it’s quite plausible that changes in precipitation such as those recorded here are associated with larger scale hemispheric phenomena.

  28. Mark T.
    Posted Oct 1, 2007 at 11:33 AM | Permalink

    The so-called “teleconnection,” as described in the link provided by richardT, STILL requires that the local weather/climate/whatever change accordingly. I.e., if the temperature in Asia is teleconnected to some other place, driving tree-rings for example, the temperature where the trees are located will change accordingly, too. The teleconnections as described by “climate scientists,” do not exhibit such behavior, spooky action at a distance but no evidence other than the desired result. Bilge is a polite way to describe it.

    Mark

  29. bernie
    Posted Oct 1, 2007 at 11:37 AM | Permalink

    “Teleconnection” is another way of saying “we have a grossly under-specified model” which is another way of saying “we don’t know”. The next thing is we will be hearing about “phlogiston” and “ether”.

  30. Michael Jankowski
    Posted Oct 1, 2007 at 11:42 AM | Permalink

    Re#29, ‘Zactly.

    Some folks would have you believe that in a house with a thermostat set to 75, that you’d get goosebumps or chills if there’s a cold front moving through your area, or start sweating if there’s a heat wave. Neither of these responses are going to occur unless the indoor (local) temperature is affected.

  31. David
    Posted Oct 1, 2007 at 11:48 AM | Permalink

    In its most extreme form, it sounds like “Teleconnection” is to Climate Science what the “Bible Code” is to Theology.

  32. mikep
    Posted Oct 1, 2007 at 12:06 PM | Permalink

    It’s the cherry picking of proxies that seems most worrying. Choosing proxies based on correlation with temperature during the period of instrumental record and not on the basis of prior well-established biological relationships seems designed to pick out spurious relationships. And if that is happening what you would expect to see is that the relationship with temperature would break down outside the period over which the correlation was established – i.e. a divergence problem.

  33. bernie
    Posted Oct 1, 2007 at 12:18 PM | Permalink

    Apropos my earlier comment:

    Here is the opening quote from Gribbin’s book on the founders of the Royal Society, the Fellowship:

    Men are deplorably ignorant with respect to natural things and modern philosophers, as though dreaming in the darkness, must be aroused and the uses of thing, the dealing with things; they must be made to quit the sort of learning that comes from books (i.e., theories), and that rest only on vain arguments from probability (i.e., computer models) and upon conjectures (i.e., value judgments) …(brackets added)

    William Gilbert, De Magnete, 1600

    Seemed appropriate.

  34. bernie
    Posted Oct 1, 2007 at 12:25 PM | Permalink

    It would help if I got the quotation right. Sorry.

    Men are deplorably ignorant with respect to natural things and modern philosophers, as though dreaming in the darkness, must be aroused and taught the uses of thing (i.e., data), the dealing with things (i.e., statistics); they must be made to quit the sort of learning that comes from books (i.e., theories), and that rest only on vain arguments from probability (i.e., computer models) and upon conjectures (i.e., value judgments) …(brackets added)

    William Gilbert, De Magnete, 1600

    Seemed appropriate.

  35. richardT
    Posted Oct 1, 2007 at 1:01 PM | Permalink

    Percent bulloides is always a good proxy. It just means different things in different settings. In the Norwegian Sea, it was high in the early Holocene, and is interpreted as an indicator of warm temperatures. In the tropics it’s often associated with upwelling, where the water is cool/productive enough for it to grow. But wind-driven upwelling in different regions is associated with different large-scale weather patterns (ie teleconnections). The interpretation has to be context dependent. This context can be informed by 20th Century climate (ideally independent of the proxy), or Holocene climate changes.

    #29
    Of course the trees or other proxies only respond to local conditions. But climate attributes other than temperature can be associated with large scale temperature anomalies. Take for example this map of temperature and precipitation anomalies associated with ENSO. There are plenty of regions that have a preciptiation anomaly without a temperature anomaly. A precipitation sensitive proxy from such a region could be used to infer the state of ENSO. And since El Nino events are associated with warm temperatures globally, this precipitation indicator would be a predictor for global temperature.

  36. Paul Linsay
    Posted Oct 1, 2007 at 1:23 PM | Permalink

    #36,

    Percent bulloides is always a good proxy. It just means different things in different settings.

    To paraphrase Humpty Dumpty, ‘When _I_ use a proxy,’ the climate scientist said in rather a scornful tone, ‘it means just what I choose it to mean–neither more nor
    less.’

  37. Michael Jankowski
    Posted Oct 1, 2007 at 1:50 PM | Permalink

    RE#36,

    Of course the trees or other proxies only respond to local conditions

    That’s not what some have argued.

    There are plenty of regions that have a preciptiation anomaly without a temperature anomaly. A precipitation sensitive proxy from such a region could be used to infer the state of ENSO. And since El Nino events are associated with warm temperatures globally, this precipitation indicator would be a predictor for global temperature.

    But it’s not ENSO every time that precipitation and temperature anomalies don’t go hand-in-hand in such regions, either. It’s tenuous enough to use a local temp proxy to represent global conditions. It’s even more difficult to use an inference from a local temp proxy to represent global conditions. And it’d be meaningless in non-ENSO years. Sure, one could estimate whether or not ENSO cycles are becoming stronger and/or more frequent, but then one would have to account for all the other factors (local and teleconnection) involved. Seems like a major introduction of errors.

  38. SteveSadlov
    Posted Oct 1, 2007 at 3:18 PM | Permalink

    Asian monsoon strength is a function of two things. Firstly, the nature and strength of the semi persistent high pressure systems in the subtropical western pacific and over the vast desert of Western China. Secondly, the degree of summer heating (and winter cold) in Central Asia. In so much as degree of summer heating in central asia may be a proxy for global temperature, then it is true, monsoon strength may be a temperature proxy. Now, let us assume this is true. So much for “warmest X in a millyunnnnnnnn years…..”

  39. Kenneth Fritsch
    Posted Oct 1, 2007 at 7:00 PM | Permalink

    Re: #36

    The interpretation has to be context dependent. This context can be informed by 20th Century climate (ideally independent of the proxy), or Holocene climate changes.

    That’s fascinating stuff RichardT, but you must realize how what say could appear as though percent bulloides could be spun into any interpretation given the imagination to make the connection. You indicate that perhaps these various effects can be identified/measured/calibrated in the near term present time. If so can you provide some links to these theories and attempts to validate them.

  40. Don Keiller
    Posted Oct 2, 2007 at 3:54 AM | Permalink

    Re “Teleconnections”, thanks to RichardT, I have got a much better understanding of the underlying mechanism. A truly wondrous concept indeed. Choose your preferred outcome (a hockey stick?). Wade through all the proxy data you can find (irrespective of type or location). Select (cherry pick) the data that support your desired outcome. Wave the magic “teleconnection wand” and Hey Presto! you get published in “Climate of the Past”, become a High Priest of the IPCC and lauded as a visionary “Climate Scientist” by Realclimate.
    Way to go!

  41. MarkW
    Posted Oct 2, 2007 at 4:57 AM | Permalink

    ENSO’s do more shifting of precipitation patterns than increase the total amount of precipitation.

    Depending on where you live, an El Nino event could cause precipitation to increase, or decrease.

    Ditto with generalized warming.

  42. moptop
    Posted Oct 2, 2007 at 5:42 AM | Permalink

    Use of a proxy for two explanatory variables (e.g. temperature and precipitation) without explaining how the confounding variable is removed.

    There’s an easy problem, all you need is a third factor which is known and relates to both. Here’s one, the propaganda requirements of the hockey team.

  43. Roger Dueck
    Posted Oct 2, 2007 at 10:31 AM | Permalink

    Does anyone have a peer-reviewed reference defining “teleconnection”?

  44. brian
    Posted Oct 2, 2007 at 11:29 AM | Permalink

    “Does anyone have a peer-reviewed reference defining “teleconnection”?”

    Sometimes it’s difficult to find journal articles that define terms like a glossary would. One paper is the Rial et al. (2004) review article on nonlinearities and feedbacks in the climate system. They define the concept of teleconnection by giving an example:

    “…large-scale atmospheric circulation patterns exert a major influence on local weather. Conversely, thunderstorm development exemplifies how small-scale climate processes can upscale to affect large-scale atmospheric circulations at long distances from the source of the disturbance (i.e., teleconnections).”

    This might not be a definition in the strictest sense, but is one example of how some climate scientists view the concept. I suppose one would want to do a thorough literature search to try and figure out when the first usage is.

    Rial, J.A., Pielke Sr., R.A., Beniston, M., Claussen, M., Canadell, J., Cox, P., Held, H., Noblet-Ducoudre, N., Prinn, R., Reynolds, J.F., and Salas, J.D., 2004, Nonlinearities, feedbacks and critical thresholds within the Earth’s climate system: Climatic Change, v. 65, p. 11-38.

  45. moptop
    Posted Oct 2, 2007 at 1:30 PM | Permalink

    Teleconnection is how GHG on Earth cause warming from Mars to Pluto.

  46. Sam Urbinto
    Posted Oct 2, 2007 at 2:21 PM | Permalink

    ‘cuz if’n it aint peer reviewed, it obviously doesn’t exist.

  47. Carl Gullans
    Posted Oct 2, 2007 at 4:34 PM | Permalink

    #18: This is far more general than this, but is relevant to say here: How can you build a confidence interval when you don’t know the distribution of the data? Why is everything assumed to have a linear relationship with everything else?

  48. Sam Urbinto
    Posted Oct 2, 2007 at 5:26 PM | Permalink

    Not just linear, but with time periods that match and equal effects.

  49. Mark T
    Posted Oct 3, 2007 at 1:22 AM | Permalink

    Why is everything assumed to have a linear relationship with everything else?

    You sir, have hit one of many nails on the head. There are plenty more if you look deep enough into the wabbet hole.

    Mark

  50. Posted Oct 3, 2007 at 3:27 AM | Permalink

    RichardT,

    #15 I agree that too often palaeoclimatic correlations are done by eye. I tend to ask what the r-squared is if I suspect this is the case, and don’t usually get an answer. A bigger problem is to assign a reasonable significance level to the correlation of two noisy, autocorrelated unpaired timeseries. This is the joyous task I have to do this evening. Any suggestions? A statistically valid solution should be publishable – somewhere like CP.

    Distribution of R (sample correlation between two independent autocorrelated series) is not necessarily unimodal. Even if we could compute var( R ) ( http://www.climateaudit.org/?p=903#comment-66605 ) the significance is a tough one to obtain. And we don’t know those autocorrelations, so maybe it is safe to assume random walk or something like that.

    See for example Granger and Newbold, Spurious Regressions in Econometrics, some important points:

    Thus a high value of R^2 should not, on the grounds of traditional tests, be regraded as evidence of a significant relationship between autocorrelated series.

    It is quite clear from these simulation that if one’s variables are random walks, or near random walks, and one includes in regression equations variables which should in fact not be included, then it will be the rule rather than exception to find spurious relationships.

  51. Willis Eschenbach
    Posted Oct 3, 2007 at 5:03 AM | Permalink

    RichardT, you ask an important question, viz:

    #15 I agree that too often palaeoclimatic correlations are done by eye. I tend to ask what the r-squared is if I suspect this is the case, and don’t usually get an answer. A bigger problem is to assign a reasonable significance level to the correlation of two noisy, autocorrelated unpaired timeseries. This is the joyous task I have to do this evening. Any suggestions? A statistically valid solution should be publishable – somewhere like CP.

    I use the method of Quenouille, which adjusts the effective N by an amount incorporating both the short and long term autocorrelation:

    I do not know whether this is the best way, but it has the advantage of being well established, being developed originally by Bartlett in 1935, and refined by Quenouille in 1952.

    There is also a discussion of the issue in Koutsoyannis, as well as in Cohn and Lins.

    All the best,

    w.

    REFERENCES:

    Bartlett, M. S., Some Aspects of the time-Correlation Problem in Regard to Tests of Significance, J. R. Stat. Soc., 98, 536–543,1935.

    Cohn, T. A., and H. F. Lins (2005), Nature’s style: Naturally trendy, Geophys. Res. Lett., 32, L23402, doi:10.1029/2005GL024476

    Quenouille, M.H., Associated Measurements, 242 pp., Butterworth Scientific Publications, London, 1952.

One Trackback

  1. [...] You can get up to speed on speleothem’s at Wikipedia and then travel on over to Climate Audit. This stuff is not for the scientifically faint (or is that feint?) of heart! Speleothems are [...]

Follow

Get every new post delivered to your Inbox.

Join 3,203 other followers

%d bloggers like this: