Craig Loehle has contributed a post to WUWT claiming that a new temperature reconstruction by Ljungqvist (2010, A new reconstruction of temperature variability in the extra-tropical northern hemisphere during the last two millenia, Geografiska Annaler 92A(3):339-351) somehow “vindicates” his own work (Loehle 2007, A 2000 Year Global Temperature Reconstruction based on Non-Treering Proxy Data, Energy & Environment 18:1049-1058).
How does he achieve this vindication? He compares his reconstruction to Ljungqvist’s this way:
I centered both on their respective long-term mean values (I did NOT rescale) and got the following.
He then proceeds to wax philosophic about the excellent agreement between the reconstructions.
For those who are a little unclear (and for WUWT readers who are a lot unclear), let me translate his procedure for you: “Make the Ljungqvist reconstruction a lot hotter — then compare.”
Suppose we don’t heat up the Ljungqvist reconstruction by centering it on its long-term mean value — then how would they compare? They’d look like this:
Gee. When compared honestly, Loehle’s so-called “vindication” becomes an indictment. What a surprise.
Open Mind 
Quick question: if this is a reconstruction of N. Hemisphere land temperatures, than why aren’t we using the N. Hemisphere land instrumental record on that end? That has increased ~ 1 C, not the 0.7 C seen globally (that appears to be what your red line is using, judging by its size).
reading from the data file at NOAA:
“Instrumental temperature data from the variance adjusted CRUTEM3+HadSST2 90–30°N data set”
Thanks for this post! Amazing that he thinks he can get away with that (well, on second thought, he probably can at WUWT)
Amazing, isn’t it, what a little context will do?
FWIW, the “agreement” up to about 1000 CE doesn’t look too good even in Loehle’s comparison–and I would suspect the point of Loehle’s work is to feed the MWP fetish, so that could be a problem for him, too.
Open Mind said:
“Make the Ljungqvist reconstruction a lot hotter — then compare.”
What is your justification for making it hotter? Recall that we are examining a record of temperature _anomalies_, not absolute temps.
absolute_temp = mean_temp +/- temp_anomaly
So we must align both plots on their mean value before we can compare the anomalies. No rescaling allowed because the anomalies already have the same units.
So (quoting Loehle) this is exactly how he did is comparison:
“I centered both on their respective long-term mean values (I did NOT rescale) and got the following.”
Again, how do you justify making one scale hotter? Perhaps to prevent the falsification of your alarmist AGW theory?
:-/
[Response: Spare us the nonsense. For proper comparison the records must be anomalies relative to the same baseline during the calibration period (the modern era). That means they should be left the hell alone. Your comment is nothing more than a pathetic excuse for Loehle’s sleight-of-hand.]
Another aspect of the Loehle 2007 and 2008 papers that is frequently overlooked by the climatologically challenged is that his data do not extend to the present. [Loehle does at least make this clear with the statement in Loehle 2008, “Accordingly, the corrected estimates only run from 16 AD to 1935 AD, rather than to 1980 as in Loehle (2007).”]
Yet they are frequently given, for example by Jo Nova, as evidence that “Temperatures were higher 1000 years ago” than “Today” in this appalling piece from WUWT last December:
in
Tamino – I put a link to your above article on the Loehle WUWT site less than an hour ago, but it has been removed.
Subsequent post now allowed.
Tamino,
Have a link to the individual datapoints you used to plot the second graph? I have a mind to plot them on the same chart as Mann et al 2008’s NH reconstruction, but I’m not sure where to find Ljungqvist and Loehle’s outputs in non-graphical form.
[Response: The Ljungqvist data are here:
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/ljungqvist2010/ljungqvist2010.txt
The Loehle data are here:
http://www.econ.ohio-state.edu/jhm/AGW/Loehle/LoehleMcC.txt
]
Given Loehle’s (lack of) correspondence with the instrumental record in the last 200 years, you could make a good argument that Loehle is consistently too warm throughout, and it should be reduced, thus achieving much better agreement with Ljungqvist, just like Loehle wants.
Does anyone have any insight into why Loehle is too hot? Is it just that the limited data he used happened to be warmer than the wider average? And why on earth does he claim that his result is “global”?
From all I’ve read, Loehle’s work is typical of what is expected at undergraduate level – I’m not sure why it was ever published.
Reminds me a lot of the Viau et al. 2006 reconstruction also and Moburg (2005). Pretty clear there is greater centennial scale variability than as originally produced in the late 90s and early 2000s but that we have warmed far beyond that regardless.
Zeke, the red line is what Ljungqvist put in, and is variance-adjusted CRUTEM3 + HadSST 30-90N. I think it is the HadSST that reduces the temperature increase, but I’m not 100% sure.
I’d be interested in tamino’s thoughts on this other recent paper on a S. hemisphere reconstruction giving similar dates for MWP & LIA to many N. hemisphere ones.
Ammonium concentration in ice cores: A new proxy for regional temperature reconstruction?
Kellerhals T et al. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES Volume: 115 Article Number: D16123
Abstract: We present a reconstruction of tropical South American temperature anomalies over the last similar to 1600 years. The reconstruction is based on a highly resolved and carefully dated ammonium record from an ice core that was drilled in 1999 on Nevado Illimani in the eastern Bolivian Andes. Concerning the relevant processes governing the observed correlation between ammonium concentrations and temperature anomalies, we discuss anthropogenic emissions, biomass burning, and precipitation changes but clearly favor a temperature-dependent source strength of the vegetation in the Amazon Basin. That given, the reconstruction reveals that Medieval Warm Period- and Little Ice Age-type episodes are distinguishable in tropical South America, a region for which until now only very limited temperature proxy data have been available. For the time period from about 1050 to 1300 AD, our reconstruction shows relatively warm conditions that are followed by cooler conditions from the 15th to the 18th century, when temperatures dropped by up to 0.6 degrees C below the 1961-1990 average. The last decades of the past millennium are characterized again by warm temperatures that seem to be unprecedented in the context of the last similar to 1600 years.
I did a write up on Skeptical Science a few months back regarding Loehle’s work titled Kung-fu Climate. In it I managed, with numbers provided by Dr Loehle himself, to show how current temps are higher than even his reconstruction (his original data ends in 1930, if memory serves). My sense from communicating with him is that he really tries to go out of his way (well out of his way) to come to the conclusion he desires.
In the game of temperature reconstructions we might call what he’s doing now “high sticking.” Surely Loehle knows his peers are going to send him to the penalty box for this.
Loehle’s method of comparing the two is SO obviously wrong, but what method did you use, Tamino? How should they be calibrated against each other? Is the instrumental record used in some way? It ain’t obvious to me, I need a bit elaboratin’ here.
How would these 2 reconstruction compare to the rather flat Mann reconstruction?
The original Mann reconstructions were rather flat. Current techniques show more centennial variation. This is a known fault (now) of the early Mann methods.
As Ljungqvist notes, his results agree well the Mann 2008 results.
The funny thing is when I saw that first figure, it struck me how poor the agreement was before year 1000…..and that was meant to be the figure vindicating Loehle. Regardless, the truth comes out thanks to Tamino.
Zeke makes a good point…is that red trace correct (i.e., N. Hemisphere land temperatures?)
Either way we are currently at least about 0.3 C warmer now than the (brief) peak of the MWP shown by Ljungqvist.
I’m sure that the “Auditor” will be auditing the Ljungqvist paper, and also Loehle’s, umm, comparsion. Not.
Tamino,
I am curious, would it be possible to compare Ljungqvist’s reconstruction with the latest N. hemisphere land SAT reconstruction, or add that to one of the graphs above?
“Fredrik Charpentier Ljungqvist says:
September 28, 2010 at 7:16 am
A comment from the author:
Some remarks have been made suggesting that the amplitude of past temperature variability are deflated. It is indeed true and discuss in length in the article. The common regression methods do deflate the amplitude of changes in the reconstructed temperatures. This reconstruction shares this problem with all others.”
Is this comment pertinent?
I think this is an important point.
Linear regression-based methods cannot detect large deviations from the linear model. So if there are positive or negative feedbacks that occur at the extremes (MWP & LIA), the (Ljungqvist and others’) reconstructions would not reflect them. Although Loehle’s reconstruction is not entirely objective, I think it’s a good lesson to think about how to properly capture the amplitudes.
I’m assuming the Ljungqvist anomalies are calculated against the 1961-1990 mean. What is the baseline period for Loehle’s reconstruction? If they are not the same is is sensible to compare them in this way?
Ljungqvist’s recontruction has as a baseline the 1961-1990
mean temperature level (as far as I can judge from the abstract).
This is what Loehle did in his reconstruction: “Data in each series were smoothed with a 30-year running mean. All data were then converted to anomalies by subtracting the mean of each series from that series. The overall mean series was then computed by simple averaging.”
So the two reconstructions use two uncomparable baselines.
If Loehle wanted to compare the two reconstructions and put them in perspective of the present climate, a better solution would have been to shift his own curve downwards instead of shifting the Ljungqvist curve upwards.
The best would have been if he had calibrated his own reconstruction against the thermometer record and converted it to the 1961-1990 baseline.
You know, I used to give Loehle big props. He did some great work in the 90s on plant life history, disturbance, and theoretical ecology. I read a lot his papers in grad school. But, unlike temperature, he peaked around 1998.
Oh, the last line of the Ljungqvist abstract?
“Our temperature reconstruction agrees well with the reconstructions by Moberg et al. (2005) and Mann et al. (2008) with regard to the amplitude of the variability as well as the timing of warm and cold periods, except for
the period c. AD 300–800, despite significant differences in both
data coverage and methodology.”
Fake graphs are so useful until someone comes along and checks, good one.
Somewhat related is the distortions of the 1990 FAR graph to boost the MWP in the Wegman Report. See Strange Scholarship in the Wegman Report. tamino gets quoted (on “Hockey Stick Illusion”) and see appendices W.4.2-W.4.4 about funny graphs and claims about them.
Nevermind, found the raw data. Here is a comparison of Mann ’08, Moburg, Ljungqvist, and Loehle including uncertainty bands: http://rankexploits.com/musings/2010/comparing-proxy-reconstructions/
The take-away is that Ljungqvist and Moburg are nearly identical; both running a bit warmer than Mann ’08 in the middle ages, but both considerably below the Loehle reconstruction.
I can still see your link. What’s the problem?
Johanus, in case you didn’t follow Tamino’s rather aggressive reply (I was confused a little at first):
You calibrate the temperature proxies based on records from today, and for the sake of argument call today ‘0’ (‘today’ might be a multidecadal calibration period). Then your datapoints for the past will have values of temperature relative to today.
Ljunqvist finds that temperatures are, on average, lower than Loehle calculates. So by plotting them against the same mean, Loehle has artificially warmed Ljunqvist’s report by about 0.2 C (from a visual inspection) at all points.
An example: Let’s say Ljunqvist finds that temperatures at 500 year intervals were -0.5 C, -0.2 C, -0.5 C, 0 C, but that Loehle finds 0 C, +0.5 C, -0.2 C, 0 C.
Ljunqvist finds that the latest time is definitely the warmest. Loehle finds that there is a warmer period.
The way Loehle would say this agrees with him is simply to move them to the same mean: which is mathematically equivalent to adding 0.3 C (in my example) to every yearly value that Ljunqvist calculates. He would then call this a vindication. I think that’s rubbish.
benpal
“I can still see your link. What’s the problem?”
My initial post at WUWT reappeared after I posted the above here at September 28, 2010 at 2:00 pm.
> Johanus, in case you didn’t follow Tamino’s rather aggressive reply (I was
> confused a little at first):
MarkR, thanks for the explanation. Yes, I made a little skeptical jab at the end, but I didn’t think the overall tone of my posting deserved such harsh moderation.
Johanus, the suggestion that Loehle’s results would somehow “falsify the alarmist AGW theory” is just plain stupid. Proxy reconstructions of temperatures over the past two millennia are interesting (scientifically and otherwise) but they’re not the reason why we are confident that greenhouse gases are the cause of modern warming.
In fact, a higher-than-expected Medieval Warm Period would truly be a cause for alarm, since it would imply a higher-than-expected climate sensitivity and thus more projected warming in the 21st century rather than less.
There’s nothing wrong with posting stupid and mistaken comments (I do it all the time :-) but if your stupid and mistaken comment also happens to be written in inflammatory language, don’t be surprised at the harsh response.
J’s first two paragraphs: BINGO!
The sad thing is that many contrarians do not realize they are shooting themselves in the sensitivity foot if they support Loehle.
I think the goal of the more intelligent contrarians is to keep beating up the hockey stick so as to discount climate scientists and the IPCC which has the effect of making the unwashed discount the science of AGW altogether.
Didactylos ” I’m not sure why it was ever published.” Well, it was published in E&E, so it was published only in the loosest possible sense of the word. So loose that it’s lost…
Well being a non-scientist, when I looked @ Dr. Loehle’s comparison, alls that I saw was a semi correlation between the dips and the rises. I did not take the example to mean that the anamolies matched, just the undulations. I think it is pretty clear that Dr. Loehle’s graph is way more “energetic” then the other re-creations.
Johanus, I’m confused. From what I understand of Tamino’s post, it was Loehle who “[made] the Ljungqvist reconstruction a lot hotter”. If that’s the case, why did you ask Tamino what his justification was for doing that, and why do you seem to think that an accusation of purposeful lying isn’t deserving of “harsh moderation”? Also, did your comment actually get moderated? What’s missing or changed?
BROKEN LINK – NOT FOR PUBLICATION
The “Climate Sensitivity” link on your page at https://tamino.wordpress.com/climate-data-links/ ist kaput.
MarkR,
Loehle doesn’t calibrate against the thermometer record, so one cannot tell how his reconstruction compares to the present period. His anomalies are only meaningful for comparing temperatures within the period of the reconstructiong.
However, if one looks at the latest part of the reconstruction and tries to calibrate against Ljunqvist (using the very imprecise eye-balling method) it seems that one would need to subtract 0.2 or 0.3 degrees from Loehle to adjust it to the 1961-1990 baseline.
Loehle says in a comment at WUWT
I was comparing the SHAPES [his emphasis] of my curve and the new one, which you do by overlaying them.
Well firstly, as has been remarked on above, before 1000 AD the shapes actually look quite different. But leaving that aside, I’m no statistician but even I figured out that using his method would give a comparison of the shape of the curve but not much else. And it’s not as if the “shape” of temperature changes over the last millennium is in doubt. Surely the point is that it is a chart of temperature anomalies from a given baseline, and if it shows temperatures which are consistently warmer than another chart when using the same baseline then one chart can hardly be said to vindicate the other.
Your offsets to the proxy- studies are purely arbitary. Proxies do not represent actual temperatures, but anomalies instead in an unknown y-axis by default and knowing your background you definitely know this. Therefore you need to offset both of the graphs as near to each other before to compare them and before making any thermometer comparison.
[edit]
[Response: Total unadulterated bullshit.
Proxy studies are anomalies relative to a known baseline. Therefore it’s straightforward to add the mean temperature during the baseline and get actual temperatures. If Craig Loehle wants to do paleoclimate work he should know this, he should state what his baseline period is, and he should really follow the usual convention of using the 1961-1990 baseline that’s just about universal in paleo studies.]
Ljungqvist entered the fray at WUWT. He was replied to shortly with potted political histories of Mann et al 1999, and he replied…
In summary, then: Loehle has no baseline, making his reconstruction absolutely useless for most purposes. Instead of adjusting his floating baseline, he rudely changed the baseline on Ljunqvist’s data. I’m prepared to accept that he did that out of rudeness and ego, but I’m fairly sure that Watts was conscious of the fact that this adjustment gives the false impression that Ljunqvist shows medieval temperatures close to recent temperatures, when in reality it does nothing of the kind.
Question(s): The Datasets used by both Mann and Loehle are an ensemble of different proxies that have somewhat nonsynchronous rises and falls caused by each proxy’s physiological response to temperature lagging by a unique time constant. Correct? So this would mean we would expect Loehle’s time series to be more sensitive to high frequency variations then Mann’s. Correct? Did Loehle’s original paper ever show a graph that did not use a running mean?
Tamino,
JeffId has entered the fray making some very serious allegations against you. Amongst them this:
“There is not much worse than a dishonest writer, Tamino demonstrated that characteristic pretty badly and due to the stupidity of his claims, boldly.”
I find that statement hypocritical coming from the source. But whatever. Could you please address this here or there. I am not a statistician nor a paleo person, but from what I have seen here and over at SS, this new paper is certainly not a vindication of Loehle as he is claiming. He also now seems to be claiming that the fact that the handle of the Ljungqvist is not flat, well neither are the handles of other recent reconstructions such as those of Moberg, Mann08,09.
Loehle also makes the claim that:
“but I eventually have remembered that my own reconstruction is set to a zero baseline for the entire 2000 yrs, so it is only possible to compare to other series that are centered likewise. Gee Tamino, didn’t you read my paper before calling me dishonest?”
What I sense here is obfuscation and weaseling by Loehle, but I’d feel a lot more comfortable if the maps were compared as realistically as possible. Not that I think it will change the comparison much, Loehle’s reconstruction is more like a mouse shape to me– a hemisphere or dome, with a small tail shown in recent times.
Anyhow, what must hurt for Loehle, and something which they choose to ignore is that current N. Hemisphere temperatures are warmer than those in the MWP in Ljungqvist’s reconstruction and even those in Loehle’s paper, the latter being dependent on which baseline one calculates the anomalies wrt.
Thanks for considering.
Mapleleaf wrote:
Seems to me this is the real clencher, isn’t it? If what is important to Loehle and rest of the WTFUWT crew is whether or not the Medieval Warm Period is warmer than the late 20th Century then Ljungqvist is really an indictment of Loehle’s work.
Loehle might stop his reconstruction well short of the late 20th Century but Ljungqvist does not — and in Ljungqvist’s reconstruction the late 20th Century is warmer than the Medieval Warm Period. This is in fact part of the Loehle paper itself:
And the attempt to argue that he is merely talking about anomalies? Just so much smoke.
CORRECTION: my lead up paragraph to the quote from the Loehle paper should have included word contradicts and have been been:
… where Loehle concludes that the Medieval Warm Period was slightly warmer than the later part of the 20th Century. It should also be noted that this was considered by Loehle and coauthor central enough to their Energy and Environment paper that this conclusion was actually stated as first part of the concluding sentence of the leading paragraph of their paper:
In the Ljungqvist reconstruction the late 20th Century is decidedly warmer than the Medieval Warm Period — and in contrast to Loehle’s small 0.07 °C difference, Ljungqvist’s difference considerably bigger at ~0.2 — nearly 3X the size of Loehle’s self-admittedly insignificant 0.07.
The mean for ‘anomalies’ Loehles 2000 years to 1935 is -0.0077
The mean for hadcrut3vnh from 1850 to 1935 is -0.3048
The mean for Loehles ‘anomalies’ from 1850 to 1935 is -0.0843
So given Loehole’s claims about the need to compare by averaging, to compare with the instrumental period and the baseline used by everyone else, he needs to take off 0.2205 degC from all his data points.
For what it’s worth, Loehle says the baseline is the mean of the entire series. He says that was spelled out in the paper (which I haven’t read).
Tamino states, “Proxy studies are anomalies relative to a known baseline. Therefore it’s straightforward to add the mean temperature during the baseline and get actual temperatures. ”
Loehle says, [September 28, 2010 at 8:20 am at http://wattsupwiththat.com/2010/09/28/loehle-vindication/#comment-494473 says],
“My data end in 1935 due to too few proxies after that. How could I zero on 1961-1990? To repeat–these are NOT absolute temperatures but anomalies. Is one period warmer than another? This is what you can ask, not what was the temperature in the year 890 AD. ”
Seems to me those two statements sum up the source of the disagreement.
Moreover, Loehle’s statement above is contradicted by his own work in Loehle 2008 where he extends his proxy record using NASA GISS to conclude that “While instrumental data are not strictly comparable, the rise in 29 year-smoothed global data from NASA GISS (http://data.giss.nasa.gov/gistemp) from 1935 to 1992 (with data from 1978 to 2006) is 0.34 Deg C. Even adding this rise to the 1935 reconstructed value, the MWP peak remains 0.07 Deg C above the end of the 20th Century values, though the difference is not significant.” See p.98 in:
Click to access Loehle_McC_E&E_2008.pdf
I see the thread over there has devolved into an intense discussion about grapes. How incredibly fascinating.
Does Watts get some kind of royalty for every time he makes a snide remark about “hiding behind pseudonames”?
Probably not, but I rather suspect he does for pseudoscience.
There is also contribution about sundials and churches!
This is making me feel good about NOT spending much time at WUWT. . .
Crackpots…they find a nice home at WUWT.
“This is making me feel good about NOT spending much time at WUWT. . .”
“Crackpots…they find a nice home at WUWT.”
Maybe. But crackpots still have a vote and can still influence governments’ policies.
Regarding those who are deluded/confused/in denial as not worth talking to seems a pretty good way of ensuring that the gulf between you and them remains in place. Your sort of comments will be music to ears of Sarah Palin and her like.
I know it is not to everyone’s taste – particularly those at the more professional end of the spectrum – to engage in the tedious business of explaining for the umpteenth time what is and what is not the case with respect to some aspect of climate science, but unless some do so then ignorance and denial flourish unhindered. So, at least, please don’t disparage the effort to stem this tide: just as Tamino’s erudite (and sometimes irascible) efforts have their honourable place, so too, to some extent, I would hope, do those who wade in at WUWT and similar sites.
Slioch, I have considerable experience discussing with people who simply cannot be reasoned with. At a place like WUWT it is even impossible, since he WILL be defended, for the sole reason that he denies AGW. I realise these people also have a vote, but there is no chance in the non-existing hell that I will ever change his mind. Or anyone else at WUWT for that matter. Outside WUWT, perhaps, and then only some. In the horde…no way.
I sympathize, Slioch, but prefer to concentrate my efforts in a more neutral forum, as there are more folks who are susceptible to reason. I’ve been debating for years, and haven’t, to my knowledge, convinced a single denialist yet.
But I’ve countered a whole lot of disinformation, and presumably have had some positive impact on normal folks.
That said, there is–or ought to be!–a special reward for those who do have the stamina and patience to represent reason in an unreasoning environment.
Horatio’s not dumb, but he can’t understand
Why Craig talks like a statistician and centers on the mean
Oh my Loehle lo-lo-lo-lo Loehle lo-lo-lo-lo Loehle
Well said Slioch. I’m not a scientist, yet I feel compelled to “….wade in at WUWT and similar sites.” Those who’s minds are made up by the well funded professional “skeptics” are largely a lost cause. However there are many unmade minds who search those sites for homework answers, or are generally seeking information. These people are valuable to you and your work(-I’m speaking of the professionals in climate sciences). I make it my job to “enlighten” them as much as I’m able. To that end, I come here to learn from the “experts” and to glean information and insight. The discussions are enlightening, but as Slioch explained, you are shooting yourselves in the foot by being dismissive, or in giving up in the face of ignorance.
So Loehle still doesn’t get why centring everything on the mean is a futile endeavour?
The problem with Tamino’s final graph is that it’s also meaningless. Comparing a properly baselined series with a floating series isn’t possible unless you find a way to add a baseline to the floating series. Yes, Tamino only did this to underscore the point that Loehle’s “vindication” is based on crap – but in so doing, he opened himself up to needless sniping.
But Loehle’s data has so many flaws and issues that really doing anything with it is an exercise in futility… and that includes adding in a proper baseline.
Did, can I request your permission to post your post at Bart’s place? JeffId is singing the praises of Loehle (and himself). Or maybe you could engage him there. He seems very taken with “variance loss”.
What do you think of efforts made by Zeke and those at John Cook’s place?
PS My above comment timestamped October 1, 2010 at 12:13 am was supposed to be a response to Didactylos‘ comment timestamped September 30, 2010 at 5:57 pm.
Sure you may. OTOH, I have no desire to talk to JeffId myself. I’m always worried that stupid is contagious….
“What do you think of efforts made by Zeke and those at John Cook’s place?”
Commendable. Futile. :-)
They do have the benefit of highlighting the real flaws with Loehle – it’s not really a global reconstruction, for example.
One result from Loehle that doesn’t get talked about enough is this: no matter how stupid your reconstruction, or how dodgy your proxies, you still get the same hockey stick. All Loehle did is show that monkeys at keyboards can produce a hockey stick, but monkeys are no more capable of recognising a hockey stick than they are of recognising Shakespeare.
Actually Loehle gives his reconstruction a baseline.
Slioch wrote:
He has it float in his “vindication” over at WTFUWT because that’s the only way that he can be “vindicated.” Ljungqvist’s reconstruction shows the late 20th Century roughly 0.2 °C above the peak for the Medieval Warm Period.
Read more carefully, Tim…. his “method” is a way to add the instrumental record explicitly without baselining anything.
He completely glosses over the extreme inadvisability of splicing two datasets by tying together two endpoints.
Oh dear.
Fredrik Charpentier Ljungqvist says:
September 28, 2010 at 7:16 am (Edit)
A comment from the author:
Some remarks have been made suggesting that the amplitude of past temperature variability are deflated. It is indeed true and discuss in length in the article. The common regression methods do deflate the amplitude of changes in the reconstructed temperatures. This reconstruction shares this problem with all others.
What is your take on this, Steven? —
… or Loehle’s statement:
Judging from Ljungqvist’s statements, Michael Mann’s 2008 and 2009 are a better “vindication” of Loehle’s work than Ljungqvist’s — at least as far as a warm Medieval Warm Period is concerned.
With considerable measure of (unintended) irony, Loehle states at one point:
Looking at Zeke Haufstadter’s graph “Comparison of Loehle, Mann, Moberg, and Ljungqvist Northern Hemisphere Reconstructions” in “Comparing proxy reconstructions”, in 1930, when Loehle’s original data came to an end, Mann and Ljungqvist differed by perhaps 0.07 °C. But Ljungqvist and Loehle differ by roughly 0.4 °C. (Link, Archive) As such it would appear that despite the “similarities” between Loehle and Ljungqvist’s data, Ljungqvist has a “huge blade” and a “huge recent warming jump” similar to Mann et al. (2008).
Furthermore, in Mann et. al (2008) tree-data makes very little difference over the past 1,300 years.
From the abstract:
… but in contrast to Loehle’s views and in support of Ljungqvist’s:
Mann et al. (2008) gets a Medieval Warm Period that is warmer than Ljungqvist’s — and it is warmer because it uses the tree-ring data that Loehle regards as faulty. And it is this tree-ring data that Loehle blames for the “huge blade” and “huge recent warming jump” over which it actually has almost no effect.
Does the expression “shooting oneself in the foot” ring a bell?
Zeke — my apologies for getting your last name wrong. I was going off memory but intended to check it before I posted — then forgot.
And your point is? I don’t think that anyone disputes this (well, except idiots).
Mosher, as I told Patrick Condon, this a no win for Loehle:
“You believe that the Ljungqvist paper “vindicates” Loehle, but it is a closer match to Mann08 and Moberg, you cannot deny that. But you suggest that Moberg and Mann08 suffer from “variance loss in the precalibration period”. Thus, using your logic, Ljungqvist’s reconstruction likely suffers from the same problem (it does agree very with those supposedly flawed reconstructions after all). Now here is the kicker, you and Tony and Loehle and others are claiming that this “flawed” (again, applying your logic) reconstruction vindicates Loehle’s work.
So, if Ljungqvist is right, then Loehle is clearly not vindicated, in fact, Loehle is shown to be an outlier. If Ljungqvist’s reconstruction has the same flaws as Mann08 and Moberg (as your logic/reasoning above clearly suggests), then it can not be used to vindicate Loehle.”
[http://ourchangingclimate.wordpress.com/2010/09/28/open-thread-2/#comment-8292]
Do you happen to know, Steve, if that is so on the decadal scale, rather than re interannual variability? I’m wondering if it makes a difference to the general finding that the recent decadal temperatures (last 20/30 years) are likely warmer than any similar period in the millennial reconstructions.
I’m still trying to plow through all of jeff’s approach to demonstrating the issue and without a counter view of things to compare it to, I’m suspending judgement.
Holy….. batman!
This means that Loehle was perfectly aware that his method for adding the instrumental series was inappropriate. It also means that Loehle was unaware of better methods of calibration – which means that he either didn’t do any background reading, or that he’s incompetent. My money is on both.
At WUWT, Loehle says about the (upward-shifted) Ljungqvist graph (shown by tamino above):
Does Loehle actually believe that relatively high correlation (r=.86) means “excellent agreement” between graphs?
Is Loehle really unaware that the correlation would have been precisely the same if the Ljungqvist graph had not been shifted upward? (although, in that case the overlaid graphs might have given a slightly different perception as indicated by Tamino’s bottom graph)
In fact, Horatio can show you two in-phase sine curves shifted vertically relative to one another by a constant cool billion that nonetheless show perfect “correlation” (r=1) even if one of them is also “scaled” by a billion (has an amplitude 1 billion times that of the other).
Would those graphs show “excellent agreement”?
Is there some sort of disease in the blogosphere that leads people to equate “high correlation” with “agreement” between graphs?
Our old friends Carter and McClean showed symptoms of this very same malady back when they published their paper on ENSO.
And, at WUWT, Correlationsheimers seems to be especially prevalent (and virulent — and ,unfortunately, totally unresponsive to Dr. Tamino’s Statistical Sauce).
Horatio,
Did he provide r values for the other reconstructions for comparison? And shouldn’t he have been using r^2 (variance explained)?
Maple:
I’ve said nothing whatsoever about Dr. L’s reconstruction, his vindication or lack of vindication, or anything about his series. I posed an open question over at Lucia’s, basically asking for a a discussion of the ways to attack this problem WITHOUT refering to personalities on all sides. what are the various ways in which the series could be lined up or not.
My reason for posting about “variance loss” is merely this. jeffid isnt the only one who recognizes the issue.
What i would like to see, cause I want to understand the issue better, is a discussion about the issue. jeff has had his say, I would be interesting from my perspective to see Tamino do his take on the whole question. It doesnt even have to be in the context of a particular reconstruction ( which always causes food fights) but rather something more methodological. Much as I disagree about some of the things Tamino says, he does explain methods very clearly.
And for the record, I don’t put much weight on any reconstruction, and believe in AGW regardless of what they may or may not show. Nothing in the record of the past can overturn the basic physics which tells us that more GHGs will warm the planet. The arguments that crop up around tangential evidence are more illuminating about the sociology than they are about anything else. I think Gavin called them scientifically uninteresting.
we agree. So, I’m more interested in the methods, as the mud slinging bores me, after all these years. And I’m more interested now in what I can learn from say Nick stokes and Ron B and Zeke, and Tamino than I am in their opinions about what to do about AGW. basically, focus on the areas where we can discuss methods and not men.
At the end of the day, if there’s a clear exposition of why some approach is wrong, or questionable, or potentially miseleading, or whatever, I would not hesitate to say it. Primarily because none of these arguments has much to do with the central issue.
[Response: If you really want to get closer to the truth, then you’ll have to be honest with yourself.
Loehle says:
Either you’re fooling yourself about the lack of “a clear exposition of why some approach is wrong,” or you’re kidding us about really wanting it.]
Steve, look at the data (zip), that was used for Loehle’s reconstruction. Two problems:
1. Each proxy more or less have different number of points at different intervals. So averaging out the whole series for the baseline, means that you’re averaging out at inconsistent time ranges and length of time series.
2. The series is not calibrated against the instrumental data, and the reconstruction is just a simple average (!). You can calibrate after the average, but without fixing problem 1 above, problem 2 is moot.
Ljungqvist said that regression methods don’t detect high amplitudes, but I don’t think Loehle’s method gets the mean function legit.
As I said I am more interested in the problem in general rather than the specifics of this case. At some point when I have a general understanding I would look at the specific case and render a judgement. I have no trouble whatsoever saying a skeptic is wrong, saying it on their sites.
WRT 1. Yes I’m aware of that and yes I’m aware of #2
what I was looking for was a clearer exposition, one that I could use to explain to people. I do spend a considerable amount of time trying to debunk skeptic arguments and the clearer the exposition the easier that task. Oh well
Steve,
Unfortunately, the clearest expositions really do require some basic mathematical understanding. Without the math, you are left with analogy and handwaving. I have had some luck in the past making arguments based on the Central Limit Theorem, which most engineers and even social scientists will vaguely remember from undergrad. However, we are dealing with an argument based on correlation, and that is something most people are not comfortable with. It’s why the house wins.
Isn’t it standard to interpolate such series so as to regularize them? See for example how you handle series like that to do FFTs
Steven,
“I’ve said nothing whatsoever about Dr. L’s reconstruction, his vindication or lack of vindication, or anything about his series”
I have no idea what you are talking about. I cannot recall mentioning you when speaking to Ljungqvist’s paper or Loehle’s paper. But then again, it has been a busy week….so I could be wrong.
Slioch wrote on September 30, 2010 at 1:33 pm:
[blockquote]I know it is not to everyone’s taste – particularly those at the more professional end of the spectrum – to engage in the tedious business of explaining for the umpteenth time what is and what is not the case with respect to some aspect of climate science, but unless some do so then ignorance and denial flourish unhindered. So, at least, please don’t disparage the effort to stem this tide: just as Tamino’s erudite (and sometimes irascible) efforts have their honourable place, so too, to some extent, I would hope, do those who wade in at WUWT and similar sites.[/blockquote]
Please keep on engaging. It is always useful to see arguments refuted in the comments sections on those sites, so even if the hard core deniers can’t be swayed, lurkers may think twice before believing what written in places like WUWT.
Given the interconnectedness of the issues and science, I would argue that an evolution or climate “skeptic” can’t really be actively involved in the debate of anti-science vs. science for very long before they become corrupted. The process of debate and the cognitive activity that it requires are a process of discovery and either what you discover through such interaction forces you to leave or it changes you. This is part of what Religion and Science is about.
However, there are those that you never hear from, those who may to some extent be taken in by the propaganda and who listen to the debates, or the people who are just starting out. You can’t really know how many of these quiet ones you change. And at a certain level I believe it is worth it even if you help only one person who might otherwise have gone the other way. Your words and actions might not save the world, but then again, according to the rabbinic Talmud,
“Whoever destroys a soul destroys an entire world, and whoever saves a life saves an entire world.”
Unfortunately, there are many whose minds are made up and don’t want to be confused by the facts.
But not all, fortunately.
The problem is not that we have deluded fools. Much of humanity has always been deluded and foolish. The problem is that their foolish delusions are treated with respect rather tha derision. We have allowed our fellow citizens to remain so ignorant that they cannot distinguish between delusion and science–and perhaps more important, face no consequences for such mistakes.
People labor under the misapprehension that belief is purely a matter of volition–to the point where people wear their beliefs in demonstrable absurdities as a badge of honor. I really don’t know how you pentrate such stupidity other than by derision.
Nice discussion. I am minded of the story of The Starfish. We fight a battle that can be won but one way: a mind at a time.
Let us go and do likewise.
The Yooper
It was ever thus.
I remember when I was very young, possibly in middle school, I debated the subject of abortion. I took the pro-life side. My opponent was strongly pro-choice, and nothing I could say would budge her.
But a few years later (I had completely forgotten about it) she told me that she had seen some of those awful shock and horror graphic abortion pictures, and it had changed her mind completely.
It is really difficult to foresee what can change a mind.
I am still pro-life. I have always been pro-life. But in my old age*, I find that everything is a matter of degree. I have always been in favour of contraception and the morning after pill, so everything after that is about where to draw the line. I agree with the West Wing writers: abortion should be safe, legal and rare.
* I feel old. 30 is old, right?
[Response: Absolutely no further mention of the issue of abortion.]
Ray Ladbury | October 2, 2010 at 12:00 pm — In my limited experience derision works not.
Are there any citizen science efforts to create a good paleo reconstruction? I’m not aware of any (although Loehle really falls into the category, since it’s clearly not of a professional standard).
Given how much fuss the naysayers make about the “hockey stick”, it’s odd that there isn’t. But then, they always did prefer to sling mud rather than contribute anything substantial.
A citizen science project could be interesting. Not just to replicate Mann (we’re not that paranoid, and others have already done so), but to explore other ideas, trivial questions and simple models that real scientists wouldn’t usually bother with.
> I feel old. 30 is old, right?
You have no idea… 30 is young for your wisdom ;-)
> citizen science efforts… paleo
Actually not a bad idea… perhaps I should point to a recent paper by McShane and Wyner offering nice code in R for doing this, and others (like apeescape, and Schmidt et al. in an upcoming comment on the paper) elaborating on it. The paper itself is not very good (euphemistically speaking) but the code base may prove useful… one problem with Mann’s code, while otherwise fine, is that it is Matlab, which few citizen scientists can afford.
…one problem with Mann’s code, while otherwise fine, is that it is Matlab, which few citizen scientists can afford.
We’ve come a long way but proprietary software’s got us imitating alchemists, attempting to communicate using vanishing ink and mirror writing. Secrets kept for enrichment. In this case the lead-gold conversion is in the software itself, with the actual useful information being an innocent bystander in the resulting ongoing shakedown.
[Response: Matlab is a fine product, and has contributed to the advancement of science.
But there’s great value in software which is freely available to as wide an audience as possible. Like R.]
There’s always Octave.
I need some free software which is capable of rendering color images of 3-d plots. That is, z=f(x,y) rotated for best viewing.
The last time I did this I used Matlab (ugh) to covert the raw data into a form I could shove into Mathematica, which had decent graphical support (in those days Matlab’s graphs were too ugly). But now I no longer have acess to Mthematica and would even have to work some to acquire the right to use Matlab, although locally available.
Surely there is freely downloadable software now available which would read in the z=f(x,y) and do my rendering into a .eps file of at least the quality that Mathematica had in the previous century.
gnuplot?
Doug, my experience is that using free software doesn’t help the mirror writing problem much. Yes, it makes it possible to share freely – but the dream and the reality aren’t even within shouting distance.
David, I have used AutoCAD in the past. Not remotely free, but you can similarly use any 3D graphics package. Just a quick script to turn your plot into a mesh, and away you go. If you want high quality raster output, PovRay is an option.
R can also do what you describe, but the ease of use is lacking, and the quality isn’t great. It’s probably comparable to Mathematica, though, and if you’re already using R then the ease of use might not be an issue.
Science turned down my paper. Not sure what to do now.
[Response: “Science” is one of the hardest journals to get into — you certainly aimed high.
Take the advice, and comments, of the referees seriously — they can probably help you improve it dramatically. And consider submitting a revision to a journal that’s not so exclusive.]
Happens to all of us (well, I never aimed so high; I meant having papers rejected).
AGU has a variety of specialty journals. Try submitting to the best fitting of those.
Barton, join the club. Science rejected a paper of ours that eventually got into Infection and Immunity, which is a fine, but more specialized, journal (different subject matter, of course). We had no reviewer comments from Science since it was not sent out for review, but getting into one of the AGU journals would be an accomplishment.
BPL,
Pardon my ignorance, but what is the paper on. I might be able to suggest a suitable journal once I know the subject matter.
As for software. Grapher (GoldenSoftware) comes to mind, not free (~$350), but a lot cheaper and more user friendly than Matlab.
Journals like Science and Nature will reject 95% or more of the papers sent to them, most are never sent for review. It doesn’t mean your paper isn’t good or interesting.
Next thing to do is pick the journal you next want to submit to. Seeking the advice of other researchers in the field might be a good idea if you’re not sure where to send it.
“I need some free software which is capable of rendering color images of 3-d plots.”
You could try matplotlib (http://matplotlib.sourceforge.net) if you’re happy to do some python scripting.
Vapor can create stunning 3D, but is probably over the top http://www.vapor.ucar.edu
Or have look through here:
http://en.wikipedia.org/wiki/Category:Free_plotting_software
D.
I WISH I’d had referees’ comments. It never got to peer review. They turned it down before that, saying it wasn’t of general enough interest. I would have thought the proposition “if global warming is allowed to continue, human civilization will collapse in the 2050s” would be of interest to everybody, but apparently not.
Sorry Barton, I like and agree with a lot of what you write. And you may be entirely correct in your proposition, but you are predicting the collapse of civilisation in 40 years, and nobody else has noticed?
I fairly certain that the Science editors have but a single form letter for rejections. You might consider one of the Nature Publ. Group publications (which now includes Scientific American). Another possiblity might be PLoS One.
Those brought up on Matlab and Mathematica might like to have a peek at Scilab and Maxima.
Thanks to all for the graphics software suggestions.
I liked the Mathematica rendering as particularly useful. It filled the surfaces between the mesh points and colored the surface using the six lights recommended to make it easy to userstand the shape of the surface. Maybe on of the freebies does much the same; I’ll check.
BPL,
Science accepts maybe 1% of the articles it receives. I think you shouldn’t get too hung up on it. If you don’t mind me asking, what subject matter was the paper relating to and so on?
RC: you are predicting the collapse of civilisation in 40 years, and nobody else has noticed?
BPL: Yes. Exactly that. Except I would make it “has not noticed yet.” Somebody has to be first.
Barton why not submit it to E&E, just to see how they react?
Funny–but a bit wicked.
BPL,
Why don’t you try this (relatively) new AMS journal “Weather, climate and society”:
http://journals.ametsoc.org/toc/wcas/2/3
Is that global civilisation, or just bits of it? Was the paper based on some insight that you haven’t shared in blog comments many times already?
BPL, you might make a side-by-side comparison to Lovelock’s book, for example, and talk about where you differ; yours will seem less of an outlier.
BPL,
Have you published a pre-print copy anywhere?
I couldn’t find it on your site and I’d be interested to read your paper. Although at the outset I should declare that I’m not convinced AGW alone will lead to the collapse of civilisation (Add in Peak Oil/Coal/Fisheries and I think it’s a bit more of a realistic risk, if you define “civilisation” as the “global machine” we have constructed).
Didactylos,
It’s based on a statistical analysis of drought I made using the database created by Dai et al. at NCAR. If my proposed expression based on other factors is correct, then, following the IPCC SRES A2 scenario, areas in severe drought grow to cover 70% of Earth’s land surface by 2050-2055, at which point human agriculture collapses.
That’s an interesting area to explore.
However, I think your analysis must have a fatal flaw: the A2 scenarios are predicated on a continuously growing world population, and a continued gap between developed and developing regions. The growing population must conflict with the collapsing civilisation and collapsing agriculture. Large scale famine will severely curtail population growth.
This means that while it is an interesting line to follow, the assumptions must break down long before you reach a global catastrophe. Also, the divide between developed and undeveloped will mean that the effects will be strongly regional. Developed regions will have the benefit of aggressive mitigation and water management strategies unavailable to poorer regions.
How are you handling the changing situation (and feedback in your own model)?
PS: I think part of your “Science” problem is that “Nature” pipped you to the post: “Global threats to human water security and river biodiversity”, Vörösmarty et al. (2010)
The paper also addresses mitigation and water management, which you haven’t so far talked about (and which, in my opinion makes the paper much more useful and of general interest).
As of yesterday, I revised the article based on observations by tamino and K.E. Trenberth of NCAR, and submitted it to J. Clim. MapleLeaf, I’ll try your suggestion next if this one fails.
It doesn’t help that I can’t pay page costs, having been unemployed since February ’08. Science estimated $2,700 to print the damn thing. I’ve cut it severely, but only from 33 pages to 29, counting figures, tables, etc.
29 pages, ouch! Still I await it’s publication with interest, but not trepidation, because 2050 will be towards the end of my optimistic tenure.
You can find a journal that does not charge page costs, or will waive themfor research not supported by grants.
Twenty nine pages? And you’re surprised Science passed on it. When was the last time Science published any paper that long?
Frankly BPL I’m amazed that someone who seems to be generally pretty au fait with the current state of science can appear so naive as to its practice.
I can only assume it is 29 pages in manuscript form. You know, double spaced, loads of white spaces, figures and tables each on their own page. Still gives a way too long paper in the end, but not 29 pages…
…which is still, being generous, twice as long as a Science submission should be. My latest manuscript is 31 (MS Word) pages plus 6 figures and will take up perhaps 4-5 pages of a standard journal. Even then we’ll probably have to move some stuff into online supplemental information.
All power to BPL getting his work out there, but Science? Seriously?
Speaking of the long-term mean related to the variations in temperature and CO2 levels, some people may be interested in this webpost:
http://open.salon.com/blog/kanuk/2009/11/24/global_warming_a_figure_is_worth_1000_words
Tamino et al.,
Posting this at Bart Verheggen’s place. JeffId has just finished replicating Ljungqvist’s reconstruction. See here:
“Jeff,
Well I’ll give credit where credit is due, good job in replicating Ljungqvist.
Unfortunately, the spin and rhetoric throughout your post makes is obvious that you are pushing an agenda. Are you, for example, suggesting that Mann08 and Moberg05 were not mathematically honest, even though they used similar techniques to Ljungqvist and obtained very similar results? You also say something along the lines of “Now we have two reconstructions”. Really, only Loehle and Ljungqvist count, even though Ljungqvist is better agreement with Mann08 and Moberg05 than it is with Loehle? Not to mention the many other paleo reconstructions out there. Interesting…..you seem to be indulging in McIntyre’s antics of feeding the skeptics fodder– would it have pained you so much to mention even Moberg?
Earlier, “skeptics” were buoyed by Ljungqvist’s statement that his reconstruction suffered from variance loss, now you say it does not have much, but do not put a number on it (well, not that I could see). And if it does not have much variance loss (which is possible, not discounting that) how come then Ljungqvist’s reconstruction agrees so well with Mann08 and Moberg05? Obviously the various analysis techniques do not make a huge difference to the final result….someone it seems is making mountains out of molehills.
So, to summarize you conclude that Ljungqvist’s reconstruction is “mathematically honest” and does not suffer from much variance loss. If Ljungqvist is good enough to allegedly “vindicate” Loehle, then Ljungqvist’s (inconvenient) conclusion also holds true:
“Our temperature reconstruction agrees well with the reconstructions by Moberg et al. (2005) and Mann et al. (2008) with regard to the amplitude of the variability as well as the timing of warm and cold periods, except for the period c. AD 300–800, despite significant differences in both data coverage and methodology.”
Not only that, but the graphic you shows also indicates that currently N.H SATs over land are warmer than they were during the MWP (as have others). Good luck trying to convince Craig of that.
I do not have time to discuss this more with you, so I am cross-posting this at Tamino’s. And either you can go there and chat about it, or others can come here to chat about it.”
Chris, the 29 ms pages include title page, abstract page, figure captions page, 6 figure pages, and 6 table pages. Ms pages are not the same as issue pages.
Barton, I suspect you’ll want to have a look at this new paper.
Egads Steve, that *is* disturbing…
Matches what I saw earlier this year:
Ouch.
Sure hope that’s variability, not trend. . . .
Regarding the change in evapo-transpiration in around 1998. After about the same time the difference in temperature anomaly between N and S hemispheres started to exhibit a substantial difference.
