An open letter to Craig Loehle

26 11 2007
Dear Craig,
I wanted to explain the position of wariness i have publicly taken on a potential collaboration with you.

First, i should have thanked you for offering it, because that implies that you believe i could contribute something useful. Quite a few people know vastly more about this subject – but it could also true that they might not be so open to collaboration outside of the tried “teams”. Hence my inexperience comes as both a blessing and a curse, but i am honored that you would even consider a collaboration with me – especially after some rather unsparing critiques. This alone is reason enough to respect you.

I have been prudent, however, to get involved in collaboration, as i wanted to be sure that you are after the answer to the old question “Was the Medieval Warm Period (MWP) a global signal and was it warmer than now ?” – and not just another dissenting publication to give fuel to the AGW-denying side, throwing away the standards of scientific publications that you seem to have upheld during much of your career (if i am to judge from a quick Web Of Science search, and browsing through a few papers).

You seem like a very creative scientist and i would never mean to downplay the contributions you have made to applied mathematics. The issue is, as far as i could judge from that one Energy & Environment article, you did not seem genuine and serious about answering the aforementioned question. This, some flamboyant prose notwithstanding, is the essence of my review.

Also, you referred in the 2007 article to a discussion of the MWP in “Loehle (2006)”, whereas upon closer examination, the paper lists only 2 references of yours : one in 2004, one in 2005….

– Loehle, C., (2004). Using historical climate data to evaluate climate trends: issues of statistical inference. Energy & Environment, 5, pp. 1-10.

– Loehle, C., (2005). Estimating climatic timeseries from multi-site data afflicted with dating error. Mathematical Geology, 37, pp. 127-140.

I just read the latter article and found it extremely interesting. I fully agree that the field of paleoclimatology has failed to taken dating uncertainties into explicit account (p137, top line). This is something i have recently started to work on, and yours is exactly the kind of contribution that could change this. The last paragraph, however, left me quite dumbfounded :

it is not sufficient to simply average historical data to produce a global or hemispheric timeseries, because dating error afflicts virtually all extant pre-instrumental reconstructions. There are two options available for obtaining statistically valid global or hemispheric timeseries. If better dating methods can be developed to reduce dating errors in proxy records, then simple averaging of series is valid. Given the many sources of dating error, this is a challenge. If, on the other hand, multiple series can be estimated, the estimation methods developed in this paper can be applied to identify trends and cycles in the historical record, even with dating errors.

(p 139)

So, if “it is not sufficient to simply average historical data to produce a global or hemispheric timeseries”, why did you do so in the E&E article ? And why didn’t you apply the two-stage nonlinear estimation technique you describe in the Mathematical Geology article, since it was so readily available to you ?

Have i been unfair to you while overly indulgent with my colleagues ? I have been accused at and on this blog of not taking a “hard look in the mirror” – a fair criticism i mulled over this during the Thanksgiving break. As i’ve stated elsewhere, my criteria as a reviewer have been :

1) Is the result or approach original ? Does it recognize prior scholarly work on the topic ?
2) is the methodology described with enough detail ?
3) are all important choices substantiated by reference or discussion ?
4) are the uncertainties appropriately discussed ?
5) are the conclusions warranted by the analysis ?

(these would undergo minor adjustments according to the field of study, e.g. paleoclimate vs. recent climate vs. climate modeling. They are by no means universal, and i welcome suggestions from other adepts of peer-review to refine or correct them.)

Be that as it may ; in the case of you paper, i would answer by :

1) yes

2) not for me
3) somewhat (yes for excluding tree-rings, no for why certain other problematic proxies were retained)
4) no
5) no

So that would lead me to ask for major revisions, or plainly reject the manuscript if i felt that the author were not of good faith. As i said in the review, a few other minor flaws hinted to the latter.

How would these criteria apply to other published reconstructions ? Let’s take the canonical “Team” global reconstructions (Mann, Bradley and Hughes, [98/99] ; Esper et al. [2002] ; Moberg [2005], Hegerl et al 2007), that i have read in sufficient detail. I state here, that while all of them have their own limitations, the same criteria i applied to your article would lead me to accept them as valid contributions to paleoclimatology – but obviously not as the Climate Gospel. In particular, much remains to be done to address uncertainties in time and temperature (X and Y, so to speak) and estimate error bars in a manner that truthfully reflects the uncertainty in the reconstructed temperatures. It is a problem in our field as a whole, one where i can only hope to make some advances myself. I never demanded you to miraculously solve all the problems of paleoclimatology in one fell-swoop ; i merely demanded proof that the basic homework had been done. Since your paper gave such a displeasing impression of the opposite, i labeled it as “pseudo-scientific gibberish”. This seemed to get quite a few people angry at ClimateAudit (although this is not a very hard thing to do ;-), so I re-iterate my definition :

Is pseudo-scientific any text or speech that wears the attributes of science while lacking the necessary rigor in the essential fulcrums of its reasoning.

and stand by the claim that the adjective applies to the E&E manuscript in its present form. I concede that “gibberish” was a somewhat excessive substantive, and i apologize for this unnecessary strike.

So, after this clarification, what do we do ? If you are ready to meet and address these issues in good faith, i would be more than happy to help in whichever little way that i can. I do believe that much is to be gained from a insightful application of mathematics to climate proxies, and would be thrilled to take part in it if our objectives are common. If you are still interested, you know where to find me.

Of course, you are perfectly free to ignore this call – after all this is only a blog and there are more important things in life.

Julien Emile-Geay




14 responses

27 11 2007

Julien: Why the need to make this an “open letter”? As written, it reveals a rather unflattering persona.

27 11 2007

I would suggest you view the Loehl paper NOT as demonstrating the existence of an MWP.

What it demonstrates is how brittle Mannian reconstructions are to proxy selection and choices of methodology.

I have drawn two conclusions from my review of the literature:

1. The Mannian reconstructions are so dependent on choices of methodology and proxies that they provide us with no statistically meaningful insight into historical temperatures.

2. The Mannian methodologies were designed specifically to produce the hockey stick style graphs which they actually did produce. That they survived the peer review process cast a shadow over the efficacy of that process in reguards to papers that would appear to support AGW arguments.

Loehle adds to the literature by showing, in simple fashion, how the proxy data can be used to support a wildly different conclusion from what Mann arrived at.

Its ability to support this conclusion is not diminished by:

1. arguments which apply equally to more accepted reconstructions.

2. arguments which, if applied to the Loehle data do not meaningfully alter its conclusions.

27 11 2007
El Niño

Bernie :
open = transparency.

Perhaps it is unflattering because such is the article – i am not alone in saying that.

Craig’s argument is essentially that “simple is better”. I fully agree that simple methods are sometimes best because they :

a) require fewer assumptions
b) can be investigated thoroughly in every nook and cranny
c) are more transparent to the reader

But what bothered me was not that it was simple, merely that the paper made quite strong conclusions without adequately documenting the data, methods and uncertainties. In other words, it had the disadvantage of its simplicity without the advantage of being thoroughly explained and detailed (something that is much more difficult when big statistical machinery gets involved).

Now, i tried to make clear that i respect Craig’s part work and am just disappointed by this one, which motivated me to ask : “What are you after ? If you are after the answer then i want to work with you”.

That’s all.

27 11 2007
El Niño

Jason :

“That [the Mannian reconstructions] survived the peer review process cast a shadow over the efficacy of that process in reguards to papers that would appear to support AGW arguments.”

it may arguably cast a shadow on the reviewers’ ability in statistics (mine would be no better), but i don’t think that is enough to question their integrity.

Loehle adds to the literature by showing, in simple fashion, how the proxy data can be used to support a wildly different conclusion from what Mann arrived at.

the problem is that a few of the proxies involve subjective choices that are not described (e.g. which of Holmgren(1999)’s “speleothem data” did he use ? There are 2), certain statements like “Whatever temperature calibration issues exist with these proxies are not common across the different proxies.” are not justified, and overall, i am sorry : this is NOT rigorous work. He is a very smart guy, his publication record shows that he knows the ins and outs of peer-reviews, and could have done much better, had he wanted to be taken seriously.

in response to your points:
The arguments i made would lead me to accept previous publications (though i maybe a statistical dummy and hence should not be a reviewer, i’ll grant you that).

2. it may well be that once the dust settles, Loehle’s reconstruction turns out to be a valid one, perhaps the closest to the “Truth” (if we ever get to know it) that was published to date . I agree that his simple approach has advantages. Unfortunately it has limitations too, and they should be explicitly acknowledged in the paper : that’s all i am saying. Not doing so is wildly misleading.

So Loehle may turn out to be right, but as it stands, it takes a big leap of faith to deduce that from the manuscript – the faith that Mann is wrong, that the MWP was warmer than now, and that any paper illustrating this point is not worthy of the same skepticism as the rest of climate science.

If one is a true skeptic, one must be fair and generous in applying this skepticism to everyone…

29 11 2007

I greatly appreciate the thoroughness of your response and the verve with which you have discussed the Loehl paper here and on CA. But I think you missed the point I was trying to make: If a young lady asks you for a date, it is somehat bad form to say “No” in a letter to the editor!

30 11 2007
El Niño

Bernie : i do feel like i could have treated Craig Loehle more considerately… i did email him backchannel, but having started on the wrong foot, i understand his reaction (which is “No”).
I’m learning, though, i’m learning…

4 12 2007

Having just read the paper, the thing that struck me is that this is NOT a simple reconstruction, because the calibrations are disparate (hidden in the component papers). The simplicity is on the surface, the assumptions are buried.

One thing to be said for the MBH and MM papers, were that all proxy series were calibrated to the surface temperature record on a common basis.

4 12 2007
El Niño

Hi Eli,

it is true that some assumptions are unstated, and that is why i went ballistic : it looked like a deliberate attempt to fabricate a warm MWP under the pretense of ‘simplicity’. On the other, there is a lot to be a said for a simple approach, so the underpinning of Loehle’s work are very legitimate : i just wish he had been more thorough and transparent in the execution.

At least as thorough as the much-decried papers by the infamous “Hockey Team” …

One thing to be said for the MBH and MM papers, were that all proxy series were calibrated to the surface temperature record on a common basis.

Hmm, i’m not sure that’s true.
= Could you elaborate ?

5 12 2007

Take a look at the files at

MM used some different conventions but the idea on this issue was the same. To quote from MBH98

“We first decompose the twentieth -century instrumental data into its dominant patterns of variability, and subsequently calibrate the individual climate proxy indicators against the time histories of these distinct patterns during their mutual interval of overlap. One can think of the instrumental patterns as ‘training’ templates against which we calibrate or ‘train’ the much longer proxy data (that is, the ‘trainee’ data) during the shorter calibration period which they overlap. This calibration allows us to subsequently solve an ‘inverse problem’ whereby best estimates of surface temperature patterns are deduced back in time before the calibration period, from the multiproxy network alone.”

As far as I can see, Loehle takes the temperature reconstructions from individual proxys and averages them. He thus buys all the assumptions from each underlying paper which are not on a common basis and may, in the worst case be contradictory.

Anyhow good stuff on your part.

5 12 2007
El Niño

Hi Eli,
thanks for clarifying.

That is what i thought you meant, but couldn’t be sure.

Well, the problem in Climate Field Reconstruction (CFR) is that it assumes that teleconnections (or to use a more vanilla statistical term, spatial covariance structure in the temperature and precipitation fields) are constant through time.
I happen to think it is not completely insane, in fact quite reasonable as a first order approximation, but if you go over at CA you will see that the whole concept gets them climbing up the walls.

And after hearing them scream for a bit, i acknowledge that they have a point : it is important to see what happens when you relax that assumption and use proxy records that are calibrated to local temperature only.

This assumes that they are ‘good’ paleothermometers, and adequately cover the globe – something that C.Loehle did not bother to write out. In fact, it is rather apparent that Steve McIntyre and many other CA readers are sometimes very delusional about the qualities of such-and-such proxy ; and often are guilty of the EXACT same bias as the Team members in selecting them – only equal and opposite.

You have a very strong point that one great virtue of CFR is that it performs a calibration on an internally consistent basis. This can be preferable over multiple local calibrations, if and only if those are made using purely statistical relationships.
However, in some cases there is a good amount of physical reasoning that goes into the development of the proxy (e.g. Mg/Ca ratios in the carbonate shells of certain zooplankton species, Sr/Ca in a coral skeleton, etc).

In those cases, there is much to be said for “first principles” over statistics : though in practice, the development of the proxy will necessarily involve some constitutive statistical relationship.

Thanks for your thoughtful comments !
El Niño

6 12 2007

The principals over at CA can speak for themselves, but as I read the discussions the issues they have are not with proxies per se but with certain proxies (notably BCPs) that when combined with others and subjected to a problematic PC analysis produce misleading results. The procedure in MBH makes sense in theory, the question is how exactly is/was it operationalized.
Seems to me, Julien, that you use far more penetrating logic with the opposition than with “friendly” folks like Eli. Then, of course, there is the issue of data access and the point you hammered Loehle on, namely, the explicit description of methods that would allow replication. But that is for another time.
Bon nuit,

P.S. Eli do you think the MWP was not a global event?

14 12 2007
Hank Roberts

Science 23 February 2001:
Vol. 291. no. 5508, pp. 1497 – 1499
DOI: 10.1126/science.291.5508.1497

PALEOCLIMATE: Was the Medieval Warm Period Global?–Wallace S. Broecker

… Broecker discusses whether this warm period was global or regional in extent. He argues that it is the last in a long series of climate fluctuations in the North Atlantic, that it was likely global, and that the present warming should be attributed in part to such an oscillation, upon which the warming due to greenhouse gases is superimposed.

——–end quote——

Sounds likely; it’s a “Perspective” piece, opinion, from one of the big names. “Likely” is a statement of probability, recognize. Up to somewhere around the 1940s, I’d think, greenhouse gas warming hadn’t kicked in strongly.

14 12 2007
Hank Roberts

Oh, for those inclined to spend a bit of time at a library:

Related Content In Science Magazine

Performing your original search
> medieval warming period was not a
> global event

in Science will retrieve 209817 results.

16 12 2007
El Niño

Bernie said

Seems to me, Julien, that you use far more penetrating logic with the opposition than with “friendly” folks like Eli

Well, Eli does have the merit of asking clear questions, and so it is easier to answer. Also, he’s seemingly got for himself this thing of knowing what he’s talking about – so i don’t have to use as ‘penetrating’ logic.

But these days i think i’m pretty good at balancing between the mitigation ayatollahs and the global warming denialists… there’s just a whole lot more of the latter posting comments here… You know, “sampling bias”.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: