Readers have inquired recently about how tropical speleothems are calibrated to temperature. Judd Partin, Kim Cobb (both of Georgia Tech) and associates wrote an excellent article last year (Cobb et al, EPSL 2007) on detailed O18 observations near a Borneo speleothem about which they had published. They introduced the article with the following observation that will resonate with most CA readers:
Detailed on-site analyses of the relationship between large-scale climate and local rainfall d18O are critical to accurate climatic interpretations of many terrestrial paleoclimate reconstructions based on ice core, tree ring, or stalagmite d18O, but few such studies exist.
They report a pronounced variation in northern Borneo dO18 values as follows (see graphic below):
Rainfall d18O values range from −11.5‰ to −2.5‰, with a mean of − 6.7 ± 2.8‰. A ~ 6‰ seasonal cycle is visible, with lighter values (− 10‰) occurring from August to October and heavier values (− 4‰) occurring from December to March.
Fig. 4. Interannual variability of northern Borneo precipitation (gridpoint centered at 3.75°N, 113.75°E,CMAPdata) (Xie and Arkin, 1997) plotted with the Southern Oscillation Index (http://www.cpc.noaa.gov/data/indices/). Both timeseries have been filtered with a 2–7 yr bandpass filter
They observe that these substantial variations cannot be explained by local temperature variations since:
Temperatures lie between 26 and 27°C year-round, as recorded by on-site temperature loggers.
Indeed, they note that this viewpoint is also held by other students of tropical speleothems:
In the tropics, stalagmite d18O records are largely interpreted as rainfall d18O reconstructions, with a minor role for relatively small temperature changes that occur in the tropics.
However, this seemingly plausible explanation cannot be transposed directly to northern Borneo as precipitation amount is not nearly as variable as the dO18 fluctuations. Cobb et al:
The ~ 6‰ seasonal cycle in rainfall d18O cannot be ascribed to seasonality of precipitation amount, which has a weak semi-annual nature, with relative rainfall maxima occurring in late boreal spring and from September to December (Fig. 2). In fact, a low correlation between Mulu rainfall d18O and Mulu precipitation (R = 0.05) suggests a limited role for a local “amount effect” stemming from fractionation in a local convective event. Likewise, the observed seasonal variability cannot be explained by variations in source water d18O, which are less than 1‰ across the Warm Pool (Brown et al., 2006).
An interesting conundrum. Strong annual variations in dO18, but not explainable either by temperature or precipitation amount.
The authors conclude that the strong variations are caused by changes in seasonal wind direction, with low dO18 coming when the wind direction leads to a water source further from northern Borneo and thus greater rainout.
Rather, we hypothesize that the observed rainfall d18O seasonality is caused by the increased rainout, and hence greater isotopic fractionation, that occurs during late boreal summer, and vice versa during late boreal winter. In late boreal summer, the ITCZ has reached its northernmost position, and mean southeasterly winds carry moisture from the Java Sea to Gunung Mulu, leading to significant rainout as moisture is carried long distances over the mountainous interior (Fig. 3). The remaining water vapor would be significantly depleted in d18O via both the “degree of rainout” and orographic fractionation mechanisms. Conversely, during late boreal winter, when the ITCZ lies south of Borneo, northeasterly winds carry moisture from the Sulu and South China Seas to Gunung Mulu (Fig. 3). The heavier rainfall d18O values measured during late boreal winter are consistent with the relatively short moisture pathway during this time of year.
All of which is fairly sensible.
People who like Hubert Lamb (of Little Ice Age and Medieval Warm Period fame) can hardly cavil at someone using wind direction as a proxy. So readers should not immediately start piling on to the obvious point that these dO18 records merely indicate wind direction, as Hubert Lamb placed great importance on prevailing wind directions as indicator of centennial climate fluctuations.
Equally, however, climate scientists who want to use such information in Mannomatics and the like need to remind themselves that the best information that they are getting from such records is information on wind direction. If you had detailed measurements of wind direction in northern Borneo (or south China or Yemen), how helpful would that be in determining global temperature?
Reference:
Cobb, K. M., J. F. Adkins, J. W. Partin, and B. Clark. 2007. Regional-scale climate influences on temporal variations of rainwater and cave dripwater oxygen isotopes in northern Borneo. Earth and Planetary Science Letters 263, no. 3-4: 207-220. http://shadow.eas.gatech.edu/~kcobb/cobb07.pdf
Climate Audit 
49 Comments
Priceless.
Dontcha know wind teleconnects?
======================
So with the Bristle cones gone, and now speleothems knocked out, what’s left?
So with the Bristle cones gone, Finnish lake sediments discredited and now speleothems knocked out, what’s left?
Re: MarcH (#4),
Faith!
Re: MarcH (#4), How about thermometers. FWIW, Charlotte, NC, had the coldest November in 38 years. 5th coldest in the last hundred.
Could someone please explain why with an “R = 0.05” we bother to describe that as a “low correlation”, a “limited role”, why can’t we just say, heck it meaningless, diddly squat, not worth another paragraph, sentence or punctuation mark, in practice my last weeks lottery numbers will probably be more useful at this point. So for the benefit of Mann09 with a 5 year sample rate, the average temperature here is 35, 21, 43, 08, 17 deg C, didn’t win me a fortune, but doubtless the professionals know how to milk it better than I.
Ian
Re: Ian (#7),
Please avoid circular logic. Where do lottery numbers come from, do you think?
Hold on here. Steve was analyzing the behaviour of tropical speleothems in recovering climatic information.
Much further away from the equator, the speleothems do record much more interesting climatic information.
John A: perhaps you should say that “Much further away from the equator, the speleothems may record much more interesting climatic information”
Steve,
Thank you for heeding a small cry for some “back to basics”. I realise you gave examples from rainy, Tropical Borneo, where your rainfall graph shows variation of 1000 mm a month or more. It is likely that this is heavy, almost daily rain. I doubt if the subtle signals from delta-oxygen would remain unconfused and unmixed, even if the rain followed an immediate, repeatable path into the location sampled instead of (probably) lagging and diverting. If there was a deep Tropical laterite weathering of the soil, lateral transport of groundwater would be a candidate for mixing.
This appears primarily to be yet another cautionary tale. Specifically, there are so many factors involved in any natural process that we need to keep in mind the very basic point: there is no reason to think that a particular natural process can yield a meaningful temperature signal unless it has been firmly established that the process (a) in fact depends on temperature, and (b) the temperature effect is independent and unaffected by other processes or can be effectively teased out from the other processes.
Mere correlation during some geologically-brief period, or within a particular, narrow temperature range, should not provide us any confidence that the natural process is an effective proxy over long periods of time or over broad ranges of temperature.
John A (#8), if equatorial speleothems can show large dO18 variations in the absence of varying temperature, there’s no good way to suppose the dO18 variations of temperate speleothems reflect temperature where it does vary. Temperate dO18 may reflect temperature, of course, but how would anyone know?
Some random thoughts
Tree ring
It’s no good sampling trees in their optimum environment = (max growth). variations +- in temperature from the optimum will reduce growth. Need to sample trees where environment has not gone through optimum during growth (obviously!). Is this the case?
Speleotherms
growth ring must be affected by temperature (increased disolved solids, increased evapouration) effects may well be masked by much larger signals
Isotopes may be affected by temperature (pos or neg?) Temp changes may again be masked by other inputs.
Grape harvests (only go back to 1300s) There must be a late date and possibly early date limit where harvest will not be made (sunlight hours preventing ripening etc)
Steve: many tree ring collections have been sampled at sites that on a prior grounds were thought to be temperature limited e.g. treeline sites. Mann disregards such niceties and simply takes every series at ITRDB and puts it into the mannomatic.
Apologies meant to add more!
… I have compared grape harvest dates and measured temperatures from 1850s onwards and the broadly agree (both Pinoyt noir and swiss harvest dates) (obviously high temp -> early harvests).
ASSUMING that 1300 to 1800 grape harvest dates are similarly aligned with temperatures then a reasonable european record can be taken back to 1300s.
Speleotherms can then be adjusted to fit this record (x1 or x-1). Sufficient records should show an overall common factor if one exists. (rainfall, cave airflow, etc.) should be minimised with enough samples. GLOBAL temperature shifts should they exist will be highlighted.
The future of this planet seem to hang in the balance of 2 groups One having submitted a hockeystick and anther having submitted nothing but who have tried to destroy said hockey stick.
Is the really important question “what is the climate doing, and is whatever caused it to do “this”, controllable. I think most of us here would not believe a couple of meters sea-level rise acceptable to western “civilisation”.
Can anyone suggest a way of extending the temperature record backwards?
Mike
Wind as a proxy of temperature, why not ? Ask Allen et al (for example). Hey, it’s climate change.
I do not find Reductio ad absurdum very persuasive as to your description of the future of the planet. It is important to understand these issues so that good decisions can be made. It would seem that rushing to spend hundreds of billions of dollars and reconfiguring economies based on what is probably flawed science of the Hockey Stick may not be a good idea.
If the science is so certain, then it should be able to withstand severe scrutiny and the scientists that are so certain about its conclusions should welcome all challenges confident that their results will withstand such scrutiny.
Based on a reasonable reading of this and other sites it’s my opinion that 1) the hockey stick science is far from settled and that there are problems with the data used and the statistics. 2) It’s probable that the current period is not as warm as other earlier periods and that what we are experiencing can be partly explained by a natural cycle. 3) It’s probable that man is affecting climate by changes in land use and these effects may be independent of increases in CO2 levels.
Einstein’s theory of relativity looked great on paper but he knew it was not worth the paper it was written on until there was evidence to support it. His axiom was 200 scientists could disagree with his theory but it only took one to prove it. This required exacting measurements taken of stars viewable during an eclipse. Campbell obtained some data and his data calculations would have disproved Einstein’s theory. Campbell however, had not been able to take a very good sample and the equipment he used to take the measurements had been cobbled together.
Eddington used better equipment and got very good data and ultimately his calculations provided the proof of Einstein’s theories. Campbell never published his alternative results. Perhaps researchers in the Climate community should take a similar approach before declaring certainty regarding the end of Western Civilization as we know it based on the “Hockey Stick”. In this instance it only takes one scientist to disprove it.
Re: Ed (#18),
I did not mention AGW on purpose. GW is not just the hockey stick many other indicators are there for all to see. The shaft of the pathetic stick may be questionable pre 1370 figures post 1370 are more certain. I support any valid criticism, and indeed some are valid. But I was requesting that someone provide an altenative to the stick. If temperatures were hotter in the historical past then I assume sea levels would have been much higher and this would be evident somewhere? Certainly there are no vast pertubations in the grape havest dates from 1370 onwards
What causes water to deposite disolved substances onto stalagtites/mites. What modifies the deposition rate? – Does water flow rate? Does air movement?, does temperature?(how does water temperature in the cave reflect the air temperature?), does humidity. Many speleotherms seem to have missing/indicipherable years. are these caused by no water flow <- high temperature? freezing?, no rainfall.
Which of all deposit-rate effects causes most change? Any reading suggestions?
Mike
Steve: why don’t you email the authors of Mann et al who used these proxies in a temperature reconstruction? I’m sure that they will be very forthcoming with references for you.
Re: thefordprefect (#21), “But I was requesting that someone provide an altenative to the stick.” There are two answers to this. One, a vast literature on the Medieval Warm period (visit the CO2 Science web site) indicates that this period was probably warmer than today. Two, the autopsies on this site suggest that many attempts to obtain reliable proxies are fraught with difficulties of such a magnitude that the entire enterprise of reconstructions seems dubious. If you think the proxies are not reliable, then it does not make sense to create an “alternate” reconstruction.
Re: thefordprefect (#21),
Actually you come to the very crux of the whole situation:
There is no evidence in any literature that I am aware of that sea levels were much higher during the MWP, and scant evidence that sea levels were problematic during the Roman Warm Period either. Alarmism about GW rests upon two things 1) that the amount of warminig we are seeing is unprecedented, and 2) that because it is unprecedented we can only “guess” at the consequences to human life.
You see, if the MWP and/or RWP were as warm or warmer than our current period, and humanity did not suffer catastrophic effects, but in fact PROSPERED during these periods, then the lesson from history would be that mankind would likely prosper during the current warming. This, of course, goes right in the face of all the doom and gloom “projections” of climatic armageddon.
So if current GW has parallels to either the MWP or RWP, point #1 fails, and with it point #2.
My reply was in reference to the Fordprefect #16
Fordprefect #16
Would you have supported Campbell if he had used questionable math to force his observations to support Einstein’s Theory and then refused to share his proof or critial portions of it with others in the science community? I’m not a scientist but the Hockey Stick “concensus” does not survive rigorous review and the case for it is not strengthened by the authors being obstinate about sharing information to facilitate it’s scrutiny.
Steve: enough about Einstein. Too far afield.
There is plenty of research, some good, all too much bad and indifferent being done in many areas of climate science, so the hockey stick hardly determines our furture.
Moreover, no one has ‘tried’ to destroy the hockey stick. It has been destroyed repeatedly but is reanimated by its creators like some B movie zombie who seem to have built the future of their careers on it.
thefordprefect:
Do you have a reference to the data source for the grape harvest dates? Have you written this up in a more formal way?
Re: bernie (#23),
data is here do not know what happened above!
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/historical/europe/
Re: thefordprefect (#27),
One reads the table for France, giving the number of days after Sert 1 that the vendange occurred. Here are some of the figures provided:
DATA:
1. Pinot Noir grape harvest dates for Burgundy, France.
Column 1: Year AD
Column 2: Pinot Noir grape harvest date, number of days after September 1.
Year Harvest date
1370 27.0
1371 25.0
1372 28.1
1373 20.7
1374 28.2
1375 20.2
1376 25.2
1377 21.0
1378 25.5
1379 24.7
1380 22.2
1381 23.5
1382 15.7
1383 7.5
1384 6.9
1385 9.9
1386 20.0
1387 23.0
1388 27.5
1389 25.5
1390 15.1
1391 16.2
1392 39.7
1393 3.4
1394 38.0
1395 20.0
1396 26.0
1397 22.0
1398 25.0
1399 26.0
1400 11.6
1401 19.9
1402 16.9
1403 19.1
1404 29.9
1405 28.7
1406 26.7
1407 29.2
1408 29.7
1409 19.7
1410 17.2
1411 37.2
1412 19.1
1413 15.9
1414 30.0
1415 22.9
1416 26.4
1417 22.8
1418 11.7
1419 20.9
1420 -1.1
1421 25.7
1422 12.5
1423 26.6
1424 12.7
1425 16.0
1426 14.9
1427 25.0
1428 36.0
1429 24.0
1430 15.0
1431 19.0
1432 18.0
1433 12.0
1434 1.0
1435 25.0
1436 56.0
What on earth does it mean to tell us that in, say, 1424 the harvest occurred “12.7” days after Sept 1? We are told we know the date of the harvest to within 0.7 of a day? Did our good French yeomen peasant farmers of 1424 Burgundy have stopwatches?
Re: thefordprefect #21
snip – editorializing
pinot noir
Swiss
Re: #27
Grape harvests are used in a reconstruction of European Spring and Summer temperatures.
Tell me again what relationship this has to global, annual temperatures?
#29: I’d guess that this was an average of multiple vinyards…
The authors conclude that the strong variations are caused by changes in seasonal wind direction…
Where the highs and lows set up determines the wind direction, and the highs and lows positions depend on the oceans and their oscillations. Seems almost too simple and logical to be true, but there it is.
1436 was a very interesting year for Burgundy. Apparently they harvested the grapes in the last week in November — over 2 weeks later than any other time in the preceding 67 years. This suggests to me – not being an expert on viniculture – that there might be something wrong with the data set.
Re: bernie (#32),
Bernie, we can all see that but none of us can arrive at a definitive cause by guesswork. It is this unfortunate (though understandable) ad hoc interpretation approach that helps make climate science so unreliable. Steve and others would go back to the original records to see if there was a scribal error or recorded metadata. The AGW crowd might try to use it to bend the shape of a curve, or to invent a new stats variation. A fringe worker would dismiss it from the data set, while a dedicated one would search for rich extra information in the anomaly.
As more climate econstruction methods are put under the blowtorch on sites like this, the more the failings and poor science are coming out of the woodwork. I personally resent the demeaning of science by green enthusiasts and I also regret that so many fairy story schemes are floating around to deal with global warming.
If I had my druthers, on present knowledge, I’d put more emphasis on the grape harvest implications than on speleothem reconstructions.
Oops, last week in October–still way out of the norm but at least less chance of a killer frost.
It is hazardous to venture into areas where we do not have expertise, however to steal shamelessly from http://www.grapes.umn.edu/frontenac/enology.html (the ubiquitous first google hit I came to, so there is a lot more to be found on the issue with even minimal searching) a couple of points to give pause:
* Proper vineyard sampling and the testing of acidity is the best way to determine the right time to harvest. (my note: this means whatever affects acidity changes the harvest date – overall temperature might be one factor, and also perhaps rainfall, the occaisional outlier frost, or other things a good plant biologist could suggest)
* Good canopy management practices are essential to reduce acidity (my note: this means that you would want to also have records about any changes to the canopy practice, or to outbreaks of plant diseases et al that may have affected the canopy)
* As with any grape cultivar, the desired style of wine must be determined (my note: vintner styles may change [perhaps a new generation, or some new competitor on the market] and so change the harvest date accordingly)
* Depending on the fruit, sugar levels from bone dry to moderately sweet have shown appropriate balance and customer acceptance (my note: customers! – the markets where the produce is being sold might also change and may have different tastes than was previously the case)
There is a lot more to be done than just to look at the harvest dates and proclaim a temperature proxy. On the other hand, if the hard research yards are done then there might be a temperature signal to be extracted in there somewhere.
Geoff:
I agree — all I meant was that this type of data needs to be carefully explored and audited, the outlier just puts an exclamation point on the issue.
Re 32/34
It may well be correct. Historical sources show that there were some extremely cold years in the 1430’s.
As for using wine harvest dates as climate proxies, there has been a lot of work done on it, particularly in France (of course). See for example Le Roy Ladurie’s “Times of feast, times of famine: A history of Climate since the year 1000”. I recommend it, it’s written by one of the worlds greatest medievalists, and completely free of mannomatics.
Re: tty (#37), and Bernie as well.
Actually, weather may not have been the only reason the grape harvest was late during this period. As this timeline shows, there was several wars being fought in France during this period. Might not the harvest have been delayed because the peasants who would have done the work had been drafted to go fight?
Harvest dates are not great temperature indicators because other outside forces might affect the date the crop is available to harvest. A war in the country, a plague decimating the population, a drought where insufficient rain fell early in the year to provide early plant growth, or even a plant disease that wiped the crop out early requiring replanting; any of these could push a harvest date back. Without some form of archival reference explaining why the date occurred when it did, the date itself is meaningless.
tty:
Interesting. The book is now looks like a collectible at http://www.abebooks.com
For those of you interested in using speleothems as a proxy for climate then you might be interested in these two papers:
The first describes a method for extracting the palaeo-groundwater/precipitation composition from trapped fluid inclusions in speleothems. The second describes a new method of using clumped isotopes as a palaeothermometer.
Both these represent two developing methods aimed at improving our understanding of past climate. Measurement of the calcite oxygen isotope composition alone will only give a signal that is a complex function of precipitation and temperature, and which is difficult to resolve in terms of temperature alone. Moreover, whilst understanding the modern relationship between drip water composition, amount, calcite composition, temperature and synoptic weather data will help in understanding the controls on speleothem isotope compositions, there remains the problem of the relative importance of these as you go back in time, particularly into different climatic regimes such as ice ages.
The recovery and isotopic measurement of water from fluid inclusions in speleothems
P. F. DENNIS, P. J. ROWE, and T. C. ATKINSON, 2001, V65, 871-884
Glacial/interglacial temperature variations in Soreq cave
speleothems as recorded by ‘clumped isotope’ thermometry
Hagit P. Affek ,Miryam Bar-Matthews, Avner Ayalon,
Alan Matthews, John M. Eiler, 2008, Geochimica et Cosmochimica Acta, V72, 5351-5360
Happy reading!
Re: Paul Dennis (#39),
Thanks, Paul. In other words, a lot more work needs to be done before we can be confident about extracting a temperature signal. Therefore, we should not be pretending that we have a temperature signal until this is sorted out, and ergo, these “proxies” should not be used (certainly to affect policy decisions) until that time.
Re: Eric Anderson (#40),
Eric, I find it very difficult to extract robust temperatures from speleothems. I’ve yet to come across many that I’m confident have grown in isotopic equilibrium with the parent drip water. The degree of disequilibrium can vary depending on CO2 degassing rates, water film thickness, temperature etc. They may be useful in indicating general patterns of behaviour, for example a cooling or warming trend but putting absolute temperatures and rates of temperature change to these is very difficult.
#43. Paul, nice to hear from you again. Mann uses C13 as well as O18 in his reconstruction. Are you aware of any suggestions that C13 can provide a temperature proxy? Holmgren, for example, say that it is related to C3 versus C4 vegetation.
Is it something that would be directly affected by rising CO2 levels (as opposed to indirectly through temperature)? It seems like the sort of proxy that would be fraught with difficulties.
Re: Steve McIntyre (#44),
Hi Steve, I think it’s hard to justify the use of C13 as a proxy, especially in speleothems. I can show you C13 records from UK speleothems that move over a significant range of several or more per mille through periods of relative climate stasis. The reason for this is that the C13 record in speleothems can be dominated by the effects of CO2 degassing. In effect the lighter C12 is preferentially lost to the cave atmosphere leaving behind a dissolved inorganic carbon pool enriched in C13.
I have a beautiful example from the same speleothem that I describe in my paper. Unfortunately we didn’t describe the C13 record as the work was about fluid inclusions. The C13 and water content of the speleothem track each other almost perfectly in an inverse relationship. The lower the water content, the higher the C13 content of the speleothem and vice versa. I wasn’t sure how to interpret this and still am not certain. However, I think what we might be observing is an effect of the cave and groundwater plumbing system. Low water contents of the speleothem are linked to slow growth rates and reduced activity of the underground plumbing system. This allows more opportunity for CO2 degassing in cavities etc. and a greater evolution of the groundwater to more enriched C13 compositions.
We also need to bear in mind that CO2 in groundwater is largely picked up in the soil zone of the rechage area. Here the C13 composition is determined by the dominant vegetation, C3 or C4. On entering a limestone aquifer the acidic water will dissolve up the carbonate adding relatively C13 enriched carbon to the DIC pool. The exact path that the isotope composition of the DIC follows as the aquifer dissolves depends on the nature of the hydrologic system and whether it is open or closed with respect to the soil zone gas reservoir. Further, if the aquifer is undergoing diagenesis then further changes to the C13 composition are to be expected.
Where I live in Norfolk, UK, the groundwater in the chalk aquifer has a very wide range of C13 compositions depending on location of the borehole in a single area. By wide range I mean from as low as -22 per mille in active recharge areas to -8 per mille in areas of older groundwater. This is wider than the range of C3 to C4 plants!
I think Holmgren and co-workers are ok to use the C3/C4 argument when discussing, for example, ice age to inter glacial transitions but even there I would be very cautious.
I can’t see any justification for using C13 in the way Mann does. From what I have written above it should be clear that it simply does not respond to temperature in any predictable or easy to understand manner.
There are 7 C13 series in the Mann network and they make up 20% of the data in his longest AD0 network So they are not incidental in the ealy networks. The reported correlations to gridcell temperature have both signs – either sign being interpreted opportunistically as being “significant”.
id code start.rec r1850_1995 rtable.r1850_1995 rtable.r1850_1995lf
199 burns_2002_d13c 6000 1218 0.27 0.27 0.48
202 burns_2003_socotrad13c 6001 0 0.51 0.38 0.51
204 burns_nicoya_d13c 6001 0 NA 0.15 0.28
381 curtis_1996_d13c 4001 0 NA 0.13 0.26
382 curtis_1996_d13cpyro 4001 0 0.40 0.33 0.40
481 lee_thorpe_2001_c13 6000 0 -0.55 -0.55 -0.76
483 li_1998_d13c 6001 1526 NA -0.25 -0.39
Re: Steve McIntyre (#46),
Steve,I’m happy to go back and look at these speleothems again and provide my own commentary on the C13 signal they contain. I’m not optimistic that my views will be favourable but I’ll refrain ’till I’ve had a chance to look in detail.
Paul, here’s an example of a C13 series with a BIG hockey stick shape. This particular series ended up being double counted because of Mann’s fence-post error. It is an important ingredient in the non-dendro Hockey Stick.
That’s a hockey stick??
Jeff Alberts #49 writes,
It goes way off scale at the end, so it’s quite possible that when the vertical scale is compressed to put it all on scale, it will have a HS shape. As is, however, the graph admittedly does not show a clear HS.