Climate Audit

steve mosher

Glad gavin posted on it.

I asked gavin this.

1. could he give me a list of the 5 proxies that he thought could be used
2. What would his opinion be of reconstructions that used any one of the remaining 13.

I posted it a while back, but I think I’m still in the RC penalty box.

we will see.

Larry

4, classic strawman.

henry

From RC:

“How to you calibrate low-resolution data? That is still an open question. – gavin]

Refresh my memory – were the BCP’s considered a low res or high res data?

Cliff Huston

RealClimate on Moberg:

http://www.realclimate.org/index.php/archives/2005/02/moberg-et-al-highly-variable-northern-hemisphere-temperatures/

This paper takes a novel approach to the problem of reconstructing past temperatures from paleoclimate proxy data. A key result is a reconstruction showing more century-scale variability in mean Northern Hemisphere temperatures than is shown in previous reconstructions. This result will undoubtedly lead to much discussion and further debate over the validity of previous work. The result, though, does not fundamentally change one of the most discussed aspected of that previous work: temperatures since 1990 still appear to be the warmest in the last 2000 years.

These results are bound to stir up interest beyond the scientific community, since the “hockey stick” shape of previous reconstructions has become so totemic (although just about everyone agrees that there is no need for this “totemising” ). We hope that press reports about this paper that mention the increased variability will also emphasize the other key result: that there is “no evidence for any earlier periods in the last millennium with warmer conditions than the post-1990 period – in agreement with previous similar studies (1-4,7)” where (1)is MBH98, (2) is MBH99, (7) is Mann and Jones ’03. The “News” article in Nature explicitly rejects the idea that this means we’re not causing the current warming. And it quotes statistician Hans von Storch (who has been quite critical of the earlier work): “it does not weaken in any way the hypothesis that recent observed warming is a result mainly of human activity”.

There are a couple of technical concerns with the paper that are worth discussing, and which the compressed space of the Nature publication didn’t leave the authors space to address fully.

First, its not too clear that the use of wavelets has added much to the mix. Moberg et al. use the wavelets to merge the high-frequency data from the tree rings with low-frequency data from the other sources, which have lower temporal resolution. But that means the low-res proxies are doing all the work, and the tree rings are just adding a fringe of noise that your eye reads as sort-of error bars.

Second, because they have used the wavelets, they end up with a non-dimensional signal which has to be normalised against the instrumental record from 1859 to 1979.

http://www.realclimate.org/index.php/archives/2005/02/dummies-guide-to-the-latest-hockey-stick-controversy/

Firstly, there are more and better quality proxy data available. There are new methodologies such as described in Rutherford et al (2005) or Moberg et al (2005) that address recognised problems with incomplete data series and the challenge of incorporating lower resolution data into the mix.

http://www.realclimate.org/index.php/archives/2006/06/national-academies-synthesis-report/

The basic conclusion of Mann et al. (1998, 1999) was that the late 20th century warmth in the Northern Hemisphere was unprecedented during at least the last 1,000 years. This conclusion has subsequently been supported by an array of evidence that includes the additional large-scale surface temperature reconstructions and documentation of the spatial coherence of recent warming described above (Cook et al. 2004, Moberg et al. 2005, Rutherford et al. 2005, D’Arrigo et al. 2006, Osborn and Briffa 2006, Wahl and Ammann in press), and also the pronounced changes in a variety of local proxy indicators described in previous chapters (e.g., Thompson et al. in press). Based on the analyses presented in the original papers by Mann et al. and this newer supporting evidence, the committee finds it plausible that the Northern Hemisphere was warmer during the last few decades of the 20th century than during any comparable period over the preceding millennium.

http://www.realclimate.org/index.php/archives/2007/01/stalagmites/

Given the complexity of the signal transfer from the surface to the cave, the only approach is to: (1) work with stalagmites that were actively growing when sampled (so we know the precise age of the top of the archive, and that they were deposited over the period of instrumental rainfall and temperature data), (2) that we use stalagmites which have annual growth rings (in the same manner as tree rings, many stalagmites have annual rings too) so that a precise chronology can be obtained, and then (3) to analyse whichever proxy we are interested in at highest resolution possible to be able to calibrate that proxy against instrumental climate records over the period they exist.

Some caveats are now needed. Firstly, all three points listed above have to be fulfilled – any two of the three just won’t give the precision in terms of chronology or climate calibration necessary to inform the past climate variability debate (a problem with the Norwegian record that featured in Moberg et al is that it failed all three tests). Secondly, depending on the local site, some stalagmites just won’t record high frequency (e.g. annual) climate variability – that signal could be lost if waters are mixed in with a large reservoir of ground water. But they may be excellent at recording the low frequency (decadal to centennial) climate signal, something that proxies such as tree rings have more of a struggle to do. Thirdly, it is quite possible that stalagmites will have a seasonally biased climate signal. For example, if the cave climate varies seasonally such that stalagmite deposition only occurs for part of the year, or if the rainfall only comes during the wet season, then it can only record that season’s conditions.

http://www.realclimate.org/index.php/archives/2007/11/les-chevaliers-de-lordre-de-la-terre-plate-part-i-allgre-and-courtillot/

There is also a bit of nattering about Moberg’s take on the Hockey stick, the supposed considerable warmth of the Medieval Warm Period, and some supposed millennial solar variability which supposedly accounts for why the present warming sort of looks like Moberg’s take on the Medieval Warm. Even leaving aside evidence that Moberg’s method exaggerates variability (see Mann, Rutherford, Wahl and Ammann 2005, available here ), the “blame the Sun” mantra falls apart because neither the Sun nor cosmic rays have been exhibiting any trend that could conceivably account for the recent warming, as we have discussed in many places on RealClimate (most recently here ).

Cliff

Michael Jankowski

Gavin’s first sentence: “Many people hold the mistaken belief that reconstructions of past climate are the sole evidence for current and future climate change.”

Someone should let Gavin know that people hold the belief that reconstructions of past climate are evidence of PAST climate change.

The way RC’s statements in post #9 (particularly the first two paragraphs) seem to emphasize the importance of reinforcing the hockey stick, “showing” how warm temps since 1990 are, etc, you would think maybe RC finds that reconstructions of past climate are the most important evidence for current and future climate change. The inclusion of the “News” quote from von Stroch futher emphasizes how important RC seems to link the reconstructions with AGW.

It’s pretty amusing to see RC initially downplay a connection between the two, yet spend two paragraphs linking the two together.

Mike B

Mosher:

I asked gavin this.

1. could he give me a list of the 5 proxies that he thought could be used
2. What would his opinion be of reconstructions that used any one of the remaining 13.

I posted it a while back, but I think I’m still in the RC penalty box.

we will see.

Don’t hold your breath. I’ve been running robustness tests on the “loehle proxies”, and every combination of 11 or more that I’ve tested so far shows MWP > CWP.

I’m not an expert on all the proxies available, but from what I’ve seen so far, a hockey stick is very difficult to create. Any sensible sampling and weighting of worldwide temperature proxies appears to show MWP>=CWP.

UC

In some cases, RC non-linking to climateaudit is mere pettiness, but, in this particular case, they cite information on Loehle that is only available at climateaudit – for example, the exact data citations are provided in a CA post here and not in the original article.

We need Granger causality test. CA post is useful in forecasting next RC post 🙂

Larry

16, I’m seeing a pattern emerging. They will say [fill in the blank] doesn’t matter when cornered, but will toil endlessly to buttress their original assertion regarding [fill in the blank]. Same for surface temperature vis-a-vis ocean temperatures, same for the hottest temperature of the 20th century. I don’t think this is how scientists are supposed to behave.

Patrick M.

I’m interested in what CA thinks of this comment at RC:

“This one actually sounds like you could use the old Extended Kalman Filter to assimilate the various series; the individual natural or artificial smoothings peculiar to each series being the nonlinear parameters that the EKF would identify. The point of that is then you could have a single probability model with which all the individual series treatments were consistent.

The reason the EKF could be expected to work well is that the proxy series appear all to have at most weakly nonlinear processing that needs to be “undone”.

I don’t know the paleoclimate reconstruction literature enough to know if this has already been tried. It would be an old idea to meteorologists though.

[Response: Francis Zwiers and a student of his have been looking at this, the tricky part is finding the right model for the temporal dependence. Mark Cane’s group at Lamont Doherty/Columbia has looked at this as well, though the primary applications have involved sparse early historical data rather than proxy data, see e.g. here and here. -mike]”

Bob KC

I enjoyed this recent slip of the pen from Gavin:

…And in a decade! The statement was/is/remains ridiculous – and it has nothing to do with my lack of an open mind. – gavin]

Craig Loehle

I have gotten as far as responding to the point that I used the wrong data column for the Farmer data: mea culpa. Fortunately, since it is a linear transform to temperature and both proxy and temp have a similar range of values, the corrected plot is almost identical to the naked eye to the one in my paper. This error was found by Eric Swanson. Thanks Eric! (bet you don’t hear that very often!)

steve mosher

Hey, he answered my question!

Interesting criteria. But he let my comment through. Good for Gavin!

Pat Keating

20 Hu
Good work!

MarkW

Actually the RC post caused the CA post. It’s just another example of effect preceeding cause. Not at all unusual in post modern climate science.

Larry

If that’s the worst they can come up with, I think you’ve survived your trial-by-blog.

Alan D. McIntire

Re #22- David Holland’s argument appeared to be lethal. There must have been a lot of heat
going into the melting of those glaciers prior to any temperature rise. Most of the additional
heat would have already gone into the system well before the CO2 started rising 800 years later.

Chris

#38- best they can come up with? Loehle used a screwed up method, with little understanding of proxy-paleoclimatic work, and published in a screwed up journal, and it will be ignored for all practical purposes by serious academic community. Please. As for Moberg, screwing up Farmer and inconsistent concept and series usage are solely on Loehle. There were some I wouldn’t use if I was Moberg (e.g. Nyberg et al) but this doesn’t matter much now with the growing archive, but the criticisms on Loehle can’t just be superimposed on Moberg. For saying “best they can do,” the the best this blog can do is still jump all over a 1999 paper and run missions to make people think everything in the NAS report or IPCC or dozens of other studies is wrong. You should create a mission statement for this site- “we will continue to try to get a MWP no matter what.” Some of the comments above can add and then try to get relevance to the attribution effort for today’s warming. Actually, I’d be a bit more worried if Loehle is right: it would hint at higher climate sensitivty if causes are unmodified. If past fluctuations turned out to be a bit bigger than we suspect, this in no way affects attribution of recent warming to CO2: we know that the sun is not changing, cosmic-rays are not changing, heat is going into the ocean not coming out, enhanced downward infrared flux, stratosphere cooling, and other things that allow attribution with very high confidence. For posters who want “proof,” you are in the wrong field- try math.

#32- you’re right, it isn’t increasing. The worldwide glacier retreat, increased ocean heat content, sea level rise, TOA radiative imbalance, satellite and ground and radiosonde data is all an illusion. [snip[ Then, we can ignore the rather basic radiative physics behind CO2 because spectrometry and greenhouse basics is not longer important. In fact, in the physics-free world, we can just hope for a hypothetical scenario where the sun is doing some weird thing we don’t know about, or just hope that all our current ways to get a meaningful global temp increase are off, despite consistent and coherent responses from biological activity, ice masses, snow cover, etc. Wishful thinking won’t get you too far, except a quick comment on the blogosphere.

Also, are people still taking the CO2 lag argument seriously? I do hope that one posters husband understands Milankovitch cycles, and Henry’s law, and feedback processes, and the timescales in the glacial-interglacial cycles.

steve mosher

RE 42. Hu. “interesting criteria” was somewhat sarcastic. My fault.

My sense is now we can go through every paleo paper and cull out the proxies that doent extend beyond
1970. And next year those that dont extend beyond 1971 etc etc. and in few years gavinology will result
in the deletion of all BCPs.

IN fact, what a fun excercise. what happens when we use a 1970 cut off date.. or 1975.. or 1980.

Why 1970? did gavin peek at the land record and decide that 1970 was a good cut off date.
Why 1970? why not 1971? 1972? why not jan 7th 1974?

Sorry, being sarcastic again

John V

I think the problem with Loehle’s proxies ending in 1970 is that Loehle2007 makes explicit comparisons between the MWP and the *current* temperature. If his conclusions were restricted to mid-century comparisons, then proxies that end mid-century would not be a problem (with me at least).

Larry

49, so you seem to be suggesting that reconstructed temperatures aren’t real temperatures, if it’s only fair to compare proxies to other proxies.

John V.

#50 Larry:
The numbering is messed up so I’m not sure if your post was meant for me.

so you seem to be suggesting that reconstructed temperatures aren’t real temperatures, if it’s only fair to compare proxies to other proxies.

I don’t understand your point. Loehle2007 used proxy measurements (only) to conclude that MWP was warmer than *current* warming. Bender and others have made it very clear that the instrumental record can not be used in a comparison with proxies, for numerous reasons.

Steve McIntyre

No CO2 lag discussions on this thread please.

Ellis

Jeez Steve, you make one comment on RC and you shut the whole site down. What did you do?

Eli Rabett

Gavin got the data file from Eric Swanson who got it from Craig Loehle. You might want to alter the first sentence. See RC comment 13 for details.

Since the current global surface temperature is about 0.5-0.6 C above 1950 eyeballing Hugh McCullogh’s graphs puts the current value well above the peak of the European Warm Period ~1000 AD

Chris

Sorry, I missed 42. Not that the proxies are the same or not the same. I said in my earlier post- Screwing up the dating, inconsistent treatment of concept and series are Loehle’s alone, and the screwing up of Farmer and DeMenocal and other studies are Loehle’s alone. I am astonished that #30 Loehle gets off as easy as “I screwed up the method, but the conclusion is all that matters.” Sounds kind of like another paper I hear a lot about on this site 😉

M. Jeff

re: #50, Chris, December 7th, 2007 at 9:11 pm and #24, Chris, December 7th, 2007 at 4:22 pm

My impression after reading these posts is that Chris is not familiar with CA.

And, I second the opinion expressed by Ian Castles, December 7th, 2007 at 5:47 pm, (#25 in the quote below is now #24).

Re #25. Our host introduced Loehle 2007 on 14 November with the statement “Thread for discussion of Loehle 2007. I’ll try to comment later but have some other obligations right now.” Steve’s prompt posting of the material sent to him by Craig Loehle was eminently sensible and he was no obligation to “double-check” the paper before opening it up for discussion by others. Your comment that he “didn’t seem to care to double-check this [paper] for errors and we all had to wait for RC” is an outrageous misrepresentation.

M. Jeff

re: #58, Chris, December 7th, 2007 at 9:48 pm.

Just the fact that you did not seem to be familiar with, “Thread for discussion of Loehle 2007. I’ll try to comment later but have some other obligations right now.”, is enough evidence for me. I’m relatively new to this blog and somewhat aged, but I recalled that Steve McIntyre had made that statement.

Steve McIntyre

I said that the criticisms of Loehle alsop apply to the closely-related Moberg reconstruction.

In #44, Chris said:

the uncomfortable truth is you cannot make long, well-calibrated instrumental records appear at the sites where you decide to collect an ice core or lake core or tree ring, and the long instrumental records are almost never collected in places suitable for ice-coring or lake-coring or tree-ring coring, so some long-distance or otherwise relations must be sought.

Be that as it may, that does not justify Moberg relative to Loehle (or vice versa.) Phil, you say that Loehle is blameworthy because he drew conclusions about the MWP vis a vis the modern warm period. Moberg drew opposite conclusions. But that’s beside the point. The issue is that Moberg’s proxies low-freq proxies overlap with Loehle’s. So that my point that the Loehle critique also applies to Moberg still remains.

Chris, I think that I’ve been pretty consistent in saying that Loehle and Moberg are pretty much peas in a pod and that I don’t see how you can objectively prefer one to another. Yes, there are some errors in Loehle, but there are some errors in Moberg as well. Moberg inadvertently used one bristlecone series twice in two different versions. It also looks to me like he got something wrong in the Lauritzen series which ens in the 19th century in the original article and in the 1930s in Moberg.

To a considerable extent, I’ve been interested in the critiques of Loehle – originally JEG’s and now realclimate’s. I’m not arguing that their criticisms are invalid – as I noted, Gavin can be a good “climate auditor” when he turns his mind to it. My beef is their failure to be consistent and to apply similar diligence to the other studies. Imagine the reaction if Loehle had mislocated proxies as grotesquely as Mann did, and then refused to correct the locations.

You suggest that I should cover other lines of “strong evidence for AGW that crops up in the literature”. Look, I’m just one person and there’s only so much that I can do. As it is I’m covering a lot of ground. I’ve done some posts on modeling and even a few on water vapor NIR, but it’s impossible for me to cover the topic in the detail that I do things without restricting myself to that topic and leaving other topics unfinished.. I’ve been looking for some time for an article that clearly derives 2.5 deg C from doubled CO2 – one which does not merely report on a GCM run, but which describes in engineering-level detail the key parameterizations for the major feedbacks. So far I’ve received some citations to articles from the 1950s, but I’d like something that represents modern thinking. If you can identify such an article for me, I’ll try to discuss it some time in the next few months.

tetris

Re:49 and 57
Eli Rabett and SteveM

Proxies here and proxies there; MWP vs. CWP, etc., etc.. It’s all getting to be quite academic, navel gazing, and quite frankly truly boring.

Reality check: based on best available data for the year and a fact recently [very reluctantly] acknowledged by NOAA, and [very red faced] by the British Met Office and HADCRU [who pls note the date: on Jan 4, 2007, stated that “2007 is going to be the warmest year on modern record”] in actual fact 2007 is in the process of going into the history books as the coldest year globally since 1995… The R squared value between CO2 values and temperature fluctuations for North America for the period 1895-2005 is 0.29 or therabouts. Which begs the question: what in all hell are we talking about.

We have ample evidence other than tree cores or sea bed deposits or what have you, to tell us that there were periods over the past 2000 years in which temperatures were at least as warm if not warmer than we are seeing today. There is indisputable evidence other than thermometers and the like that clearly correlates these temperature fluctuations with the ups and downs in terms of economic growth and societal development [warmer=better; colder=worse, it’s that simple].

We knew that RC was going to have a go at the Loehle 2007 study, that was easy to predict.
That they took quite this long possibly tells us that the Hockey Team didn’t like its implications. Not much more.
That RC committed the sin of plagiarism as outlined above is only a marker of how low this “discussion” has sunk.

Peter

If you put it all together, do you not arrive at the following: that using the proxy record alone, MWP was about at the same levels as modern warming for the period the proxies cover. This is assuming the proxies really do measure temperature. So warming and cooling can both happen without any assistance from CO2. You are still left however with the question, whether there is anything remarkable about the last 10 or 20 years compared to the 20c record

Then, if you look at McKittrick’s stuff about the measured temp record, the divergence between ground temps and satellite measurements, the divergence between US and ROW ground temps, you come inevitably to the conclusion that the recent temperature record is not remarkable in 100 year terms. Not only that, but it seems that the last few years show that stable or falling temps are compatible with rising CO2 levels.

Surely, there is just nothing left after you this? It simply is not possible to argue for impending doom or even AGW and take account of this evidence? You are left with the null hypothesis, climate fluctuations about some mean over periods of decades or centuries, due to causes not yet established.

Philip_B

Jon, I assume this is a reference to the practice of using (the theory of) teleconnection patterns to justify data mining (data fishing) for climate correlations. Here is a CA thread on RegEm.

JamesG

The thing that most struck me about the individual temperature series used in Loehle’s reconstruction is that none of them looked remotely like each other. Doesn’t that confirm that any warm or cold spots were likely regional?

Steve McIntyre

#77.

The thing that most struck me about the individual temperature series used in Loehle’s reconstruction is that none of them looked remotely like each other. Doesn’t that confirm that any warm or cold spots were likely regional?

This is a very important point and one that applies to the other studies as well.

Do these plots show that the warmth/cold was regional or that the proxies aren’t necessarily “proxies” for temperature, but might be reflecting something else. I’ve observed a spaghetti inconsistency between different reconstructions in the Urals. But I don’t think that Yamal was regionally cold in the MWP while the Polar Urals about 60 miles away were warm.

It is the lack of a consistent “signal” that really causes the dispute over methodology. If you had a consistent signal, you’d get pretty much the same answer no matter what multivariate methodology you used. Because there is no consistent signal, changes in weights yield quite different results. The Team gets one set of answers by picking the same proxies over and over and applying slightly different methods to them.

Jim Clarke

James G,

It would be strange indeed if all of the proxies from the different parts of the globe looked identical. Consider that the warming of the last 100 years has been regional and mainly concentrated in the high northern latitudes. Antartica has been generally cooling and most areas in the tropics and Southern Hemisphere have no, or barely discernable trends over the time period. Even in the Northern Hemisphere, there are many instrument records that indicate cooling over the last century. The Idso’s have had a good time posting a different station record ever week for many years now. In total, however, the evidence is very strong that the ‘global temperature’ has been warming.

In both the modern era and in the paleoclimate studies, the global temperature is determined by combining all of the data to get the global trend. Regional climates can and do vary significantly.

Boris

I like the gratuitous plagiarism charge.

welikerocks

#72 Hu, you can use a place like http://www.tinypic.com to upload and host your graphics for free, and it will also give you all the code to copy and paste where needed after you do the upload. It’s really easy.

steve mosher

boris you will note that I asked gavin very nicely to cite his source and he did.

Steve: Watch the pea under the thimble. He didn’t say that he didn’t consult CA, only that he had an alternative source for the numbering which he happily provided. However he also provided exact data citations for Loehle’s versions and the only place in which Loehle gave exact digital data citations was at CA. It’s possible that Gavin ignored the data citations provided by Loehle here, foraged around the internet and located digital sources for Loehle’s versions and then went to the trouble of verifying that these digital sources matched Loehle’s. Do you believe that he did this? It’s possible, but it would be more logical for him to use Loehle’s actual data citations provided here and that’s what I think that he did. If Gavin says explicitly that he did not consult the CA posts, then I would accept his explicit statement, but he’s not said that.

Kristen Byrnes

The main point that Steve Mc was making in this thread was that the standards used by RC in criticizing Craig Loehle need to be applied to Moberg. For those who usually want to agree with RC and disagree with Steve seem to be avoiding this rather valid point.

Mike B

#62 bender

The arguments for either Loehle 2007 or for Moberg et al 2005 are highly specious. Special pleading on either side (a logical fallacy) is wholly unjusitified. The reality is that these are only two flavors of millions that could be drawn from this set of proxies. This is the lesson taught to us by Burger and Cubash. Nobody but Steve M (and the CI gang: UC and Jean S, and now Hu McCullough) seems to understand this. If you were to do an honest-to-goodness spaghetti-o-gram that included the kinds of uncertainties that a young and honest policy wonk like Susann is asking for, you will simply not find a difference between MWP & CWP. Who wants to bet $100 in the CA tip jar? Chris? I’ll give you 10:1 odds.

The “special pleading” bit for Moberg vs Loehle sounds an awful lot like a strawman to me. I’ve commented primarlily on Loehle because the data was more readily available and I was working on my on take on his paper (not Moberg). And the two papers share many of the same series anyway.

The approach I’ve taken to the analysis is to examine subsets of the proxies. This allows detection of sets of proxies that exert undue influence on the overall combination, on any particular era, or any particular region. It provides for an approach for calculating confidence intervals (bootstrapping, which I’ve mentioned before). It doesn’t seem like this type of analysis is standard part of paleoclimatee studies, although it should be. But as Wegman stated, since there is little contact between the paleos and mainstream statisticians, this isn’t surprising. It also isn’t shameful, or at least it wasn’t at first.

So Bender, the members of your special little club are not the only ones who understand what the implications of these proxies are, and I can’t
understand what your purpose is for claiming some secret special knowledge or insight for yourself. Can we please leave the insufferable condescention to other climate websites?

Hu McCulloch

UC (#70) writes,

Yes. For example, in #20 figures, zero-line is outside the confidence limits some 5 % of the time

Good point, if we had no historical suspicion of an MWP or LIA. However, the MWP (according to Wiki in any event) supposedly was c 800-1300, while the LIA was c 16-19th centuries. The fact that Craig’s data shows significant warming during most of c 850-1050 and significant cooling during most of 1450-1650 is therefore not just data mining.

I would have expected the LIA to have set in earlier, given that Greenland was prospering in the early 13th century, froze out during the 14th century, and was depicted as permanently connected (by sea ice, presumably) to Norway or Russia in early 15th century maps.

Hu McCulloch

Try, try again:

Notice that a few of the series, after Craig’s approximately 30 year rolling average, appear as broken step functions.

Bruce

Anyone notice that Nov 2007 Southern Hemisphere HADSST2 is down to .061 above the 1960-1991 mean?

It hasn’t been that low since 1992.

ICE AGE!!!!

http://www.cru.uea.ac.uk/cru/data/temperature/

Hu McCulloch

Clayton #85, Thanks! I at first tried using 15 different colors, but the graph came out like mud, so I went back to cycling through 4 bright ones. It’s still a lot more legible than the B&W graph posted on an earlier thread in this category.
Individual series can be studied in the preliminary paper I linked at #20, before and after Craig’s rolling mean procedure. These may be copied out as WMF, and then pasted into Photoshop and exported as GIF, if desired.
A few of the horizontal steps may be out of range — I just set the y-axis to (-4, 4), which picked up most of the action.

Hu McCulloch

Gavin (on RC) knocks Craig for including the Calvo series, which he says doesn’t start until 1440. Infact, that’s about where it ends:

(Red X’s are raw data, blue steps are Craig’s extensions).
(I hope this one works…)

Hu McCulloch

Anyway, the hyperlink on #88 works OK.

Larry

Is that calendar year, or years in the past?

Hu McCulloch

Re Craig’s series 12, Calvo, the source data at
http://doi.pangaea.de/10.1594/PANGAEA.438810?format=html
gives its first reading as 0.51 kyr BP (11.1 dC). This would
be 1950 – 510 = 1440 AD, unless they have changed the BP scale
while I wasn’t looking. So I think Gavin is right on this one,
and LoehleData12.csv posted on CA is wrong. Since I was able to
replicate Craig’s results quite closely using this file, it is
apparently the version he was using.

Correcting this won’t change the big picture, since Craig’s resolution
is only 30 years at best anyway. However, it ought to be redone, just for the
record, after checking the other series.

Pat Frank

Steve M. wrote: “The post is a “climate audit” type of post and shows that Gavin can be a pretty good “climate auditor” when he turns his mind to it.”

Leading to the obvious question of why Gavin doesn’t invariably turn his mind to it when evaluating all climate-scientific claims. I think this is the most telling illumination of all on the standing of mainstream climatology these days. Opportunistically directed criticism is the mark of r..g…n and politics, but not of science.

Any fair reading of the peer-reviewed climate literature shows that GCMs can not at all reliably predict future climates, that proxy so-called science has become little more than tendentious assertionism (thank-you M&M), and that the canonical surface temperature record is so error-ridden as to be climatologically meaningless (thank-you Anthony Watts, et al., and Douglas Keenan).

If Gavin were to forthrightly apply his pretty good climate auditing skills to the supporting corpus of AGW claims, he’d invariably conclude the claims are scientifically insupportable through-and-through.

And yet to our public knowledge he does not so apply them and, presuming his RC statements reflect his truly held opinions, he does not conclude AGW bogosity. This constitutes a prima facie case that Gavin is not doing science. The alternative explanation is that he is evidently doing consciously partisan politics.

Steve McIntyre

#96. I’ve noticed that inconsistencies can arise in how authors handle their top-of-core dating. Sediment authors often are dealing with very big picture. I discussed this in connection with Black (Cariaco) and Richey (Pigmy Basin) where the evidence seemed very strong to me that Richey’s top is dated 50 years or so earlier than Black’s but are plausibly much closer,

steve mosher

RE 99 needling? Moi? I haven’t the slightest Idea what you are talking about.
I merely asked a question about the “other 13” proxies and gavin assumed I’m talking about comparisions to
Moberg? I didnt mention Moberg. And then when I ask him to cite the source for his numbering of Loehle series,
he says that butter won’t melt in my mouth. That’s gettin mighty personal, cowboy. Anyway. I think there is fruitful
ground to plow by asking questions about the role of recons in GCM calibration and C02 sensisivity estimates.
Gavin has been nice enough to answer my questions so moshpit is in the penalty box. He’s watching game films of
Gillies and McSorely

Jamesg

Imagine this scenario out of it’s current context. That is, if for some other reason a group of us gathered temperature plots from 18 parts of the world and none of them showed any similarity to each other, would we see any point at all in combining them into one plot? Probably not. It’s just the idea that there is a global signal that forces these combinations. But clearly neither in the past nor in the present do we have such a signal. At best there are many different regional signals. Comparing very poorly sampled dissimilar, combined past data to poorly sampled, dissimilar, combined present data – at different sampling locations even – seems meaningless to me. What might be useful is to find modern temperatures at these 18 places and compare them, or perhaps just to concentrate on the few places that keep being mentioned by professional climateers as vulnerable – ie the Arctic, Antarctic, Bangladesh, Florida, Australia.

On that note, I was astounded recently when I looked at an official action plan for Bangladesh which showed a local temperature history plot that had virtually flatlined for the last 50 years, at the end of which was tagged the GCM climate projectons of 1.4 degrees C for the next 50 years. Another one of those obvious GCM absurdities that everyone just ignores.

steve mosher

Sod:

” Steve, your article does not seem to mention the mistake about dating the series. do you have some evidence that it is false?
or do you think that 50 years are not important, when someone is making claims about the end of the 20th century?”

Sod, craig already thanked eric for doing the audit and finding the error. Now craig will correct it.
Do you think you could write a nice note on RC asking Dr. Mann to correct his latitude/longitude errors?
I’m trying to check the 5o year error. Could you tell me which series had the error, and what the error
was? at the start date or end date?

“Hu McCulloch, the error graph looks pretty good to me. why have so few people commented on it?
http://www.climateaudit.org/?p=2496#comment-173463”

There are probably less than 6 people here who understand what Hu has done and none at RC.
If they have an issue with it they will speak up. Then again, we could 1 million people comment
on it if that would give you a feeling of consensus.

“the main difference between Loehle and Moberg,
are the wild claims that Craig is making about a period, that he has little data about.”

That’s a testable claim: Go read Moberg and list his claims. Go read Loehle and list his claims.
And then tell us what constitutes wild. I’m just curious.

theduke

Mosh #102, “he says that butter won’t melt in my mouth. That’s gettin mighty personal, cowboy.”

Yeah, he better watch that kind of talk or we might have to conclude he’s less than a manly man. (just kidding, Gavin!)

I was just struck by the sullen air over there. It’s as if they know bender’s ABCs and XYZs argument on multi-proxies in #102 is the real meaning of Loehle but they can’t talk about it because it’s too horrible to contemplate.

Phil.

“the main difference between Loehle and Moberg, are the wild claims that Craig is making about a period, that he has little data about.

it can t be said better than this:

Only if both goose and gander said:
“Even keeping in mind that Figure 1 shows 30-year running means, it would indeed seem to show the MWP to be warmer than the late 20th century.””

Yes although Steve Mc wants to gloss over that. While some of the criticisms of Loehle also apply to Moberg the dating errors don’t necessarily. The claims by Loehle relating to the late 20th century for which he has little data certainly justify criticism.

Steve McIntyre

#111, Phil, I agree that these particular dating errors don’t apply to Moberg; I agreed that the dating errors were worth noting and even said that Gavin could be a pretty good “climate auditor ” when he wanted. However you notice that he didn’t comment on the dating error in Moberg’s Lauritzen series or Moberg inadvertently using one series twice; or the various errors like that in MBH.

I reject yor efforts to draw a line between Moberg and Loehle. You say that Loehle made “wild” cliams of a sort that Moberg did not make, citing:

Even keeping in mind that Figure 1 shows 30-year running means, it would indeed seem to show the MWP to be warmer than the late 20th century.”

Compare that statement to Moberg’s statement:

We find no evidence for any earlier periods in the last two millennia with warmer conditions than the post-1990 period.

Moberg’s claim is more categorical than Loehle’s. Now let’s look at the network of series that contribute to the late 20th century results and Gavin pointed out quite well the very few contributors to the late 20th century in Loehle. As I’ve said repeatedly, these are almost exactly the same roster that are used by Loehle:

GRIP borehole (match);
Chesapeake (match);
Shihua=Beijing stalagmite (match);
Yang China (match);

The differences are :
Moberg uses the very questionable Arabian Sea G Bulloides series;
Loehle uses the Ge and Holmgren series.

As noted repeatedly, the G Bulloides series was not calibrated to temperature and is evidence of colder upwelling water.

The “active ingredient” in Moberg’s different conclusion is supposedly more cold water offshore Oman. And this series BTW has been spliced.

Loehle needs to correct the dating of a few series, which I’m sure that he will do. Once that’s done, I see no objective basis on which Moberg can be accorded superior status to Loehle.

Note that I’m not saying that I think that one can put much confidence in Loehle’s reconstruction either. I did a similar calculation for the NAS panel with similar results and described it merely as “apple picking”.

Going back to Moberg’s claim, in Loehle’s shoes, I would probably have presented these results in a counterfactual way – showing that they disproved Moberg’s allegation that there was no evidence of a warmer MWP. There is some evidence of a warmer MWP; there is also some evidence the other way. Moberg is wrong to say that there’s no evidence and Loehle’s reconstruction shows this rather effectively. I don’t think that it shows the opposite any more than I think that Moberg proved his thesis.

And as to Gavin caviling about the inability of a couple of proxies to show modern-medieval relationships. I’m OK with that comment. But again, he needs to look in the mirror and ask himself how a few bristlecone pines chronologized by Graybill, in a chronology that Ababneh could not replicate, can show the opposite.

It’s the hypocrisy that bothers me.

Steve McIntyre

Gavin commented on the Moberg problem as follows:

[Response: These things aren’t equivalent. Moberg et al was an attempt to incorporate lower resolution data as well as high resolution data using a new methodology based on wavelet analysis – that is why it was interesting. The specific records they used were not so much the point, it was more a proof of concept. If they were to update it, they would likely have a larger sample and leave out the least well-dated records. More to the point, they seem to be aware of what they were doing and what they were dealing with. With Loehle, there is no new methodology to speak of, and so everything depends on the records and their treatment – you get that wrong, there’s not much left. – gavin]

Again, Gavin’s hypocrisy is, as always, astounding, as well as his total failure to understand how you establish a new method.

If the significance of Moberg is simply the introduction of a novel statistical methodology, then Moberg should publish in a statistical journal on data sets with known properties and show that his method works better than the competitive methods. And having done that, move on to data sets where empirical results are being derived.

As to knowing “what they are doing”, people introducing novel methods in climate articles don’t know the properties of their methods – witness Mannian PCA+ inverse regression, Mann’s first RegEM,… Does Moberg understand the properties of his statistical method any better than Mann (“I am not a statistician”) understood his methods. Of course not.

By using a simple method, Loehle has a fighting chance of understanding what he’s done – surely that’s an improvement over Wizard of Oz arm-waving. I think that the Mannian style efforts to develop even more unintelligible multivariate methods by publishing these new concoctions as statistical innovations in weakly reviewed (statistically) climate journals such as Journal of Climate are pointless. IF Mann wants to play statistician, publish the method in a proper statistical journal.

Gavin says:

With Loehle, there is no new methodology to speak of, and so everything depends on the records and their treatment – you get that wrong, there’s not much left

Well, I agree with this, but again, the hypocrisy is breathtaking. The Moberg and Loehle collections are essentially the same. As I’ve already shown, if you apply Moberg methodology to Loehle, you get an answer that looks like Loehle.

Yes, “everything depends on the records”. OF course it does. As I’ve said endlessly, if there was a common “signal” in these proxies, then the methodology wouldn’t matter. Because there’s no common signal, cherry picking is rife. Thus the Team’s addiction to bristlecones and foxtails.

As to what Moberg would do now, Gavin says:

they would likely have a larger sample and leave out the least well-dated records.

Unfortunately this isn’t as easy as all that. I think that the development and utilizaiton of very high resolution sediments is a very worthwhile effort and I’ve been keeping an eye out for over a year for new high-resolution series and have tried to post on them whenever I see one: e.g. Black at Cariaco, Richey at Pigmy Basin. But there’s not as common as one would like. they would likely have a larger sample and leave out the least well-dated records. There are other high-res series e.g. Rein’s grey scale that have terrific resolution, but it’s not clear what they measure. But it’s got a large medieval anomaly. But I haven’t seen all that many that are both well-dated and calibrated to temperature.

In any event, it’s nice that Gavin is finally coming to appreciate the relevance of data quality – a prominent theme here at CA – in his new-found role of Gavin Schmidt, “climate auditor”

Peter D. Tillman

Steve, this is a hard thread to follow, since (I presume) you have deleted some posts and changed the numbering of later posts.

Could you please, please leave the post numbers undisturbed when you moderate. The simplest method would be a simple “post deleted by moderator” or some similar boilerplate left in place, with the original post number.

Thanks in advance, Pete Tillman

Steve: In the software available, if I’m going to edit posts one by one, it takes me a long time to tidy posts one by one. I can move posts to Unthreaded by ticking them in a Move function and save a huge amount of time. I dislike the invasion of post after post by people arguing issues unrelated to the post. I’d ask people to simply not respond to off topic posts or to do so on Unthreaded. Please spend a little energy asking the thread invaders to disciplie themselves. I can’t do everything.

Phil.

Steve stop dodging, the fault I’m addressing of Loehle is not equally applied to Moberg. His reconstruction shows a peak temperature in 1966 higher than any other since 1300 and yet he explains away his ‘downtick’ in the 1970s as due to “The series ends with a downtick because the last set of points are averages that include the cool decades of the 1960s and 1970s.” According to his reconstruction those decades are the hottest for over 500 years, what’s he smoking? If Mann had made a nonsense statement like that you’d have been all over it, but it gets a pass!
As for a more nuanced way of expressing his conclusion, to paraphrase he said ‘even though I screwed up the reconstruction of the middle of the 20th century I still think the MWP was warmer than the end of the 20th century’, I suppose ‘since I screwed up the reconstruction of the middle of the 20th century it’s not possible to make any meaningful comparison with the present’ would do.

Peter D. Tillman

Re: #117, Steve said

In the software available, if I’m going to edit posts one by one, it takes me a long time to tidy posts one by one. I can move posts to Unthreaded by ticking them in a Move function and save a huge amount of time. I dislike the invasion of post after post by people arguing issues unrelated to the post. I’d ask people to simply not respond to off topic posts or to do so on Unthreaded. Please spend a little energy asking the thread invaders to disciplie themselves.

Thanks for the explanation. I don’t hold much hope for disciplining “thread invaders”, and don’t want you to waste time on admin. The rest of us can help by quoting a bit more than just the post number. I’ll add this to the list for the new FAQ, and maybe we need a page of suggestions for posters too.

Cheers — Pete T, who wishes blog software would pick up more pointers from Usenet software

Phil.

Re#122

Since he frequently does post here Loehle will doubtless be able do defend his illogical argument.

sod

I reject yor efforts to draw a line between Moberg and Loehle. You say that Loehle made “wild” cliams of a sort that Moberg did not make, citing:

Even keeping in mind that Figure 1 shows 30-year running means, it would indeed seem to show the MWP to be warmer than the late 20th century.”

Compare that statement to Moberg’s statement:

We find no evidence for any earlier periods in the last two millennia with warmer conditions than the post-1990 period.

Steve, i still am not convinced.

even by just eyeballing your graph, i think that Moberg has a higher ratio of “active” proxies around at the end of his time (1990) than Loehle has for the “end of the 20th century”.

it would be great, if someone could update your graph:
http://www.climateaudit.org/?p=2405#comment-167079

Hans Erren

RealClimate has a good discussion of problems with Loehle 2007: link, linking to JEG’s discussion but not to discussion here.

Now why doesn’t that surprise me? Did Michael Mann ever answer scientific questions by peers on a conference?

Steve: Well, he did tell the NAS panel: “I did not calculate the verification r2 statistic as that would be a foolish and incorrect thing to do”.

John Norris

re Steve McIntyre #115

Steve McIntyre

By using a simple method, Loehle has a fighting chance of understanding what he’s done … Steve McIntyre

Steve McIntyre quoting gavin

…With Loehle, there is no new methodology to speak of …

Loehle’s methodology was to keep it simple and transparent. For Climate Science, that is a new methodology.

re Steve McIntyre #128

In our Almagre case, we didn’t wait and peek at the results. I reported that we had re-sampled Almagre and committed myself to archiving and reporting the results whatever they showed. …

Some in the Climate Science are undoubtedly working on a new methodology that makes a HS out of your new data.

Hu McCulloch

Al, #118, wrote re my #20,

Those confidence intervals appear sensible from a y-amplitude standpoint, but they would appear mighty tightly fitting along the time axis. That is, let’s consider the sharp peak at around 1220. The peak on the upper-bounds line and the peak on the lower-bounds line are also sharp, closely fitting peaks. Seemingly identical peaks with just vertical displacement in fact. If there is any error in the time-axis data, there should be some spread in the peaks on the bounds-lines… yes? A five-year error (0.25% of x-axis scale) would appear to break the confidence interval.

Were both x-axis and y-axis errors included? Or is the dating that accurate?

I just took the time-axis as “true” and computed y-axis standard errors for each point in time, after applying Craig’s approximately 30-year rolling average. If some of the series had uncertain timing, that would tend to flatten out the signal, so if any signal comes through after averaging, it is probably an understatement of what was really going on.

It is probably more meaningful with this kind of data to look at even longer intervals than Craig’s tridecades — perhaps even centuries. But for now it’s interesting to go with what he chose (after corrections of Farmer and any BP problems).

Stan Palmer

re 128

Steve MacIntyre wrote:

A point to keep in mind in all these studies – and one which I repeat endlessly with few people getting the point. The problem is that there is no common signal. I’ll illustrate this in a slightly different way.

Can you comment on the robustness tests that Lohle and posters here have reported on – removing proxies and noting that the results persist.

Hu McCulloch

Phil (#120) wrote, re my #20,

The link to your approach doesn’t work for me, how did you deal with the very sparse data, did you use Loehle’s approach of spreading the data over 31 years or did you use the actual data? I’m surprised the error bars don’t get wider at the end of the series (~1980) since there are only ~5 data sets at that point. Thanks.

Try this
link to ?p=2505#comment-164663. If that doesn’t work, go to the 11/21 thread “MBH98-Style…”, Comment #11.

I took the raw data as posted on CA, and then used Craig’s approach of spreading each point over approximately 30 years. (Because we handled non-integer years differently, there were some slight differences, but nothing important.)

My errors bars do get wider at 1980, where his reconstruction ends, since there were only 8 series still active there, vs 14-16 typically in the middle. My OLS se for 1980 is 0.2280 dC, versus about 0.16 dC typically in the middle. See the graphs of my OLS and WLS se’s and the number of included proxies on the paper I linked successfully above at #20 and perhaps also here. The WLS se for 1980 is 0.1650 dC, vs about 0.10-0.11 typically in the interior.

(I’d insert the graphs themselves here, but am not having much luck with either imageshack or esnips — the latter uploads my image, but then doesn’t give me a URL to it. The former gives me lots of URLs, but they are either thumbnails or don’t work. I may try picasa or tinypic when I get around to it.)

Just to see what would happen, I tried extending Craig’s reconstruction forward to 2000 (see plot comparing my replication to his reconstruction in paper linked above). The little downtick at the end of his series after 1975 or so turns out to reverse itself around 1985, so it is nothing to get excited about.

Luboš Motl

Dear Steve,

correct me if I am wrong but the particular individual criticisms of the proxies by RC that I have looked at so far seem incorrect. For example, Gavin claims that Calvo et al. 2002

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VBC-45Y4JGC-2&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=e80a4ceb6072ae1e15412268cf2f6707

shouldn’t have been included at all because it starts in 1440. As far as the abstract above says, they go 8,500 years before the present. I even seem to think which related paper goes back only to 1440 – it is Zhao et al. 2000 (also sea cores)

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V7P-412RWCX-C&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=7b13733debd09cdc695a237c0119a653

but this Zhao paper is not used by Loehle. I am also suspicious about the claim that Loehle has confused the BP and BP2000 symbols because he carefully avoided them in his paper and in my opinion, these symbols shouldn’t be used for timescales shorter than many thousands of years.

Could you please be more specific which Gavin’s criticism of proxies used by Loehle (or Moberg) you agree with (or checked) and why?

Unlike you, I tend to agree with those who say that in the paleoclimate reconstructions, it is ultimately positive work that makes the real progress. Criticizing someone or even (sometimes) finding legitimate errors is OK but their work is tough anyway, isn’t it? If you think that you can make a better and more controlled reconstruction than Moberg and Loehle, why don’t you do it? Based on other reasons, I think that Moberg and Loehle might be pretty close to reality anyway and work that I could view as superior to Moberg or Loehle is just a better reconstruction, not just a criticism created for the sake of it.

Of course that I agree that one can invent many types of criticism and the same one may apply to many papers but still, one must be trying to compare different methodologies (and proxies themselves) relatively to each other. Which existing reconstructions do you consider superior over Moberg and Loehle and why? Click my name for my text about Gavin’s reaction to Loehle.

Best
Lubos

Bob Koss

Another possibility is you’re not using the Quicktag Img button to have the image display in your post? The link button won’t display it.

Steve McIntyre

Luboš, I did a post on Calvo core MD95=2011 here , showing plots of various indices from original data. MD95-2011 is a piston core and values after about AD1440 or so are lost. There is a nearby box core JM… that comes to the present. While dO18 values have been archived so that a reasonable splice can be made, this is not the case for alkenone SST where results from the JM core supposedly didn’t make sense and haven’t been archived (from a CA reader familiar with the matter.)

One publication on core MD95-2011 expresses dates in BP2000 (Risebrobakken et al Paleooceanography 2003), but the archive at pangaea.de is denominated in BP1950. Loehle incorrectly converted results expressed in BP1950 to BP2000 – it was an error that he promptly acknowledged; it occurred inadvertently and the original authors didn’t help much by their own varying usage. It’s a very high resolution core, worthy of being considered and it would be nice to locate the unarchived alkenone SST data for JM97-948/2A (done in the 1990s). The dO18 values in this high-resolution core are not exceptional in the 20th century.

I presume that this series was included even though it ended in the 15th century because it had enough values to meet Loehle’s criterion for inclusion even without the later portion. Is this a horrendous error? If so, it’s one that was also made by Esper without him being criticized for it.

Zhao has nothing to do with this and you should take this off the table.

I tend to agree with Gavin on the need for better resolution data from the ocean sediment people. I’ve been following this a lot longer than Gavin and have attempted to write posts on any new high-resolution series that I come across (Black on Cariaco, Richey at Pigmy Basin, Newton in the Pacific, McGregor offshore Morocco,the Arabian Sea series, Rein offshore Peru, Khim in the Bransfield Basin, as well as some lake sediment series – Pallcacocha,.., Mt Logan ice core,….) They are aide-memoires as much as anything. Not all of these have been calibrated to temperature by the original authors – no doubt for a good reason.

Kenneth Fritsch

One can criticize the entire concept of temperature reconstructions based on showing that the statistical methodologies are insufficient or incorrect, that the proxies were cherry picked to met an end result and that the proxies used for temperature do not appear to have been validated with a reasonable physical model. One could more readily correct the methodology deficiencies and overcome the problems of data snooping the proxies to be used, but, unless one has proxies that one can use with some reasonable level of confidence and give evidence for that confidence, why would anyone want to present their own reconstruction(s).

The Loehle reconstruction would better have been presented as an example of how a very different temperature series can be reconstructed using real world proxies and how that indicates not so much any thing about the comparison of the Modern to the Medieval WP, but rather an apparent weakness in the entire reconstruction process.

On the other hand, a defender of past reconstructions and the implied evidence they provide for AGW and the degree of AGW, being aware of the unstated implications of Loehle reconstructions, might well be motivated to wring some differences between Loehle’s and their favorites. That wringing appears to this observer to be currently coming more from the hands than the little gray cells.

Walt Bennett

Re: #144,

Can’t handle the occasional poke? (I mean you, not Steve…)

🙂

Just having a bit of a chuckle…

bender

#412
Maybe Walt can’t tell the difference between a critique of a paper vs the flawed methodology on which said paper is based? Walt, nobody’s “moved on”; the flawed methodology of MBH98 is still being employed as though nothing went hopelessly wrong. Mann is not a statistcian. And neither is Rasmus. And neither is Tamino. And neither are you. And it shows.

yorick

Here is an intersting survey of studies with conclusions re the relative amplitudes of CWP and MWP.

Phil.

Re149

Why do you find it interesting and which studies are surveyed?

sod

Here is an intersting survey of studies with conclusions re the relative amplitudes of CWP and MWP.

you are not serious about this? co2sciemce?

http://www.co2science.org/scripts/CO2ScienceB2C/data/mwp/qualitative.jsp

in comparison, the Loehle paper is amsterpiece in accuracy and methodology!

steve mosher

RE 152. Actually I am just correcting Walt. If I wanted to be silly, I’d compliment you on your intelligence.

SteveMc never accused them of plagarism
and neither did I. We were both very circumspect in our observations. You will note, as I stated,
that I asked gavin very nicely to cite his source, which he claims to have done.

The bottom line. Gavin, Mann, and Hansen in general refuse
to acknowledge SteveMc or CA in any way shape or form. I’ll call this tactic stalinist, but understand
that’s not a pejorative statement.

One thing people never seem to grasp. When the king characterizes his opponent as a jester, as a fool,
as someone who has nothing of interest to say, the king simultaneously creates an image of himself,
like it or not, that is the inverse. And he create expectations of perfection in everything he does.
Like it or not. Ask Nixon.

Consequently, when the jester has something of interest to say about something, the king must
avoid acknowledging him. And when the king makes a mistake, he cannot admit it.

Then comes the special pleading.

This is why we have the rule of law; that is, this is why we have the scientific method.

Susann

When reading both RC and CA over the past week, I am forced to think of an episode of Seinfeld when Banya comments, “This is GOLD, Jerry! GOLD!”

I’m trying to think of a good hockey / war related title for my possible climate policy blog. I’m taking suggestions and will offer full credit to the winner. 🙂

J.C.H.

Susann says:
December 9th, 2007 at 4:42 pm

“I’m trying to think of a good hockey / war related title for my possible climate policy blog. I’m taking suggestions and will offer full credit to the winner. …”

Went to a war and a hockey game broke out?

Jim Edwards

#157, Susann:

“Slapshot”

catchy, reminds one of snapshot – which is timely / journalistic, plus a great movie with the Bash Bros.

steve mosher

RE 157 Take this quiz first

http://www.youthink.com/quiz.asp?action=take&quiz_id=794

“You just don’t care, you will punish your opponent untill they can not move.
You have no problem throwing down the gloves. You will push a guy through the glass,
you will shove him into the ice. Your only mission is to inflict as much pain on your enemy as possible.”

Hmmm. these quizes are never right.

“laying on the lumber” would be my choice.
I also had “mental dentistry” once upon a time ( but thats not hokey or war )

Terry

Re 157

“I went to Stockholm and all I got was this lousy hockey stick”

Larry

Doesn’t this all belong on unscrewed…..er….unthreaded?

steve mosher

RE 167. True. Let’s return to the discussion of why humans lost their body hair, present
company excepted.

Seriously. The Issues with Loehle have already been hashed out. Gavin ( actually eric )
added a couple dating issues. This will improve craig’s paper. Hu’s work will improve craig’s
paper. Maybe JEG will throw in an improvement or two. In the end, I would hazard, that we will
still be in the same position, more or less: Can’t rule the MWP in, can’t rule it out.

Those speculations dont undermine the work that craig has done and the contribution that Hu
has recently made.

Jimmy

#141 states

Not all of these have been calibrated to temperature by the original authors – no doubt for a good reason.

Is that b/c

a) there is not enough overlapping SST data with the sediment core to calibrate the proxy data
b) the researchers believe that the lab experiments where they grow these critters under controlled conditions and measure trace metal incorporation are adequate
c) the sediment core proxy data actually diverge from measured SSTs, and this divergence is not reported as a cover up…

Jimmy

#168

I think everyone can agree that there was a warm period around 1000AD in some areas (MWP). The question is: How widespread was it? Can one say it was “global”… I don’t think averaging a bunch of curves will give you the answer to this – regardless of your statistical geek factor. People can even decide on how to calculate a US or global temperature these days with tons of datapoints. How can you do it with much fewer paleoclimate reconstructions???

A better answer would be draw a line on the a map around those points which show a MWP. Does it cover most of the globe or not?

Phil.

Re# 167

Hu what I was referring to as “gobbledegook” was the following:

“His reconstruction shows a peak temperature in 1966 higher than any other since 1300 and yet he explains away his ‘downtick’ in the 1970s as due to “The series ends with a downtick because the last set of points are averages that include the cool decades of the 1960s and 1970s.” According to his reconstruction those decades are the hottest for over 500 years, what’s he smoking?”

How would you define that statement?

theduke

Phil: “I see Gavin acknowledge SteveMc frequently. . . ”

Let’s check the list of links over there at RealClimate. Let’s see we have:

# Climate Change Education
# Climate Science
# Climate Science Watch
# ClimateArk
# Climatedenial.org
# ClimateEthics.org
# ClimateProgress

I don’t see any link to Climateaudit. Why? Did the mention of it give Mann the vapors?

Gavin might acknowledge Steve Mc. CA does not.

Yancey Ward

It is pretty obvious that RealClimate should list ClimateAudit amongst it’s links; that they don’t speaks loudly about the true depth of their committment to the integrity of science itself. It is unethical and unscientific to try to deny the existence of critics of one’s scientific work, especially critics as thorough and exacting as the handful found doing the work here on a regular basis.

Hu McCulloch

Phil (#174) wrote,

Hu what I was referring to as “gobbledegook” was the following:

“His reconstruction shows a peak temperature in 1966 higher than any other since 1300 and yet he explains away his ‘downtick’ in the 1970s as due to “The series ends with a downtick because the last set of points are averages that include the cool decades of the 1960s and 1970s.” According to his reconstruction those decades are the hottest for over 500 years, what’s he smoking?”

How would you define that statement?

What Craig wrote was,

Thus it is not possible to compare recent annual data to this figure to ask about anomalous years or decades. The data show the Medieval Warm Period (MWP) and Little Ice Age (LIA) quite clearly. The weries ends with a downtick because the last set of points are averages that include the cool decades of the 1960s and 1970s.

So he says that although his series shows big events like the MWP and LIA, it shouldn’t be used to make statements about “anomalous years or decades” As a specific example of such an “anomalous year or decade”, he cites the downtick after 1975.

Obviously, this inconsequential yet eyecatching downblip was somehow caused by the behavior of the data. Because of his rolling average, 1975 could be affected by data as early as 1960. Mostly likely a relatively warm proxy dropped out circa 1975, leaving only other, relatively cool ones, whence the suddenness of the drop.

As I noted above somewhere, if his series is continued after his 1980 endpoint to 2000, this blip is just as suddenly reversed c. 1985. Evidently a relatively cool series drops out at that point. This uptick is just as inconsequential as the preceding downtick. Weighted Least Squares softens both movements, but does not remove them.

Susann

Stickhandled. 🙂

Hu McCulloch

Continuing #181, it turns out that the specific “perfect storm” that leads to the inconsequential downblip at the end of Craig’s series and subsequent upblip if it is continued past 1980 is that in 1976 #18 Kim, which was running cool, drops in, pulling the series suddenly down. Then in 1983 #2 Gajewski, which was running cool, drops out, while in 1986 #17 Farmer drops in, running warm. The two 80s events roughly offset the 1976 event.

As Craig pointed out, such anomalous years or even decades don’t mean much of anything. Interpolation would iron such wiggles out, giving a perhaps illusory sense of continuity.

John Norris

With respect to RC not referencing CA, some here may remember that RC went the extra mile to provide a modified Google capability that searches the internet but will filter out CA, and others. They elected to not disclose what websites they filtered out, but if you search climate audit on regular Google, and then through their modified Google search function, you can see what they did.

http://www.realclimate.org/index.php/archives/2006/10/new-google-search-function/

John Norris

Oh one more thing to add to my #187.

If getting to the truth were more important to me than winning the argument, I would do as Steve did, and reference RC. If I put winning the argument ahead of getting all information on the table, I would do as RC did; not reference CA, and filter it from a customized Google search function.

Phil.

Re 186
” in 1976 #18 Kim, which was running cool, drops in, pulling the series suddenly down.”
That dataset ends in 1940: http://doi.pangaea.de/10.1594/PANGAEA.438838

The downkick starts in 1967 and with a vengeance in 1971.

With the erroneous dating the data point at ‘1990’ is assumed to start at 1976, so the value measured in 1940 is used in 1976.

“while in 1986 #17 Farmer drops in, running warm.”
That dataset ends in 1950: http://www1.ncdc.noaa.gov/pub/data/paleo/paleocean/by_contributor/farmer2005/farmer2005.txt

In this case the data point at ‘2000’ is assumed to start in 1986, actually the 1950 measurement!

This isn’t an anomalous decade it’s an anomalous century but since we’re not too worried about the odd 50 years here or there I guess it doesn’t matter?

Jay Currie

190, that is one very interesting paper. And to my statistically naive eye, reasonably well tested.

Armagh Geddon

Over at RC, Gavin said this:

Finally, I agree, real analyses will win out in the end. I’m hopeful that is too far off. – gavin]

Never was a truer word said!

steve mosher

re 176 Hu..

I should have been clearer. When I refered to MWP I mean to imply a MWP that is warmer than
the post 1990s. So, I agrre with you that a MWP is ruled in as your CIs show, but the contention
that MWP is as warm as or warmer than CWP will have to await updating proxies.. Is that about right
for your position?

Hu McCulloch

Phil (191) wrote,

186
” in 1976 #18 Kim, which was running cool, drops in, pulling the series suddenly down.”
That dataset ends in 1940: http://doi.pangaea.de/10.1594/PANGAEA.438838

The downkick starts in 1967 and with a vengeance in 1971.

With the erroneous dating the data point at ‘1990′ is assumed to start at 1976, so the value measured in 1940 is used in 1976.

“while in 1986 #17 Farmer drops in, running warm.”
That dataset ends in 1950: http://www1.ncdc.noaa.gov/pub/data/paleo/paleocean/by_contributor/farmer2005/farmer2005.txt

In this case the data point at ‘2000′ is assumed to start in 1986, actually the 1950 measurement!

The Farmer (2005) link given says it is quoting age in “calendar age”. The column is headed “cal age (yr)”, with no mention of BP. “Cal BP” would mean dendrocalibrated C-14 dates before 1950 ordinarily, but since there was an explicit reference to “calendar age” in the text and nothing about BP, it would appear that this is indeed “calendar age”. The youngest age given is precisely 0 yr, not -55 yr. I would infer from this that the youngest date is whenever they collected their data, sometime after 2000. What gives you the idea these “calendar ages” are BP1950?

The Kim (2006) reference gives its first (youngest) age in “kyr BP” as .01, at .02m deep. The second is .08 kyr BP, at .13m. The third is .16kyr. at .24m. The surface rate of accumulation seems to be pretty constant, at about 1/6 cm per year. If 0 BP is 1950, this would place 2006 at roughly 9 cm above the surface. It’s easier to think that their 0BP is about when they did their study, as in one of the 2 official versions of the Calvo table cited by Steve McI here. I can’t get ahold of the Kim paper right now. Does it clarify the table?

It’s all the same to me either way, as long as we get it straight. The downblip at the end of Craig’s series is simply what the data said was happening, as he understood it, and given his rolling mean approach.

theduke

I’m gonna edit Joe Black in #192 and improve slightly:

CrossCheck

Good, pithy double-entendre.

Full credit to Joe.

Phil.

Re199

The Farmer (2005) link given says it is quoting age in “calendar age”. The column is headed “cal age (yr)”, with no mention of BP. “Cal BP” would mean dendrocalibrated C-14 dates before 1950 ordinarily, but since there was an explicit reference to “calendar age” in the text and nothing about BP, it would appear that this is indeed “calendar age”. The youngest age given is precisely 0 yr, not -55 yr. I would infer from this that the youngest date is whenever they collected their data, sometime after 2000. What gives you the idea these “calendar ages” are BP1950?

Because the authors said: “PERIOD OF RECORD: 21 KYrBP – present”

The Kim (2006) reference gives its first (youngest) age in “kyr BP” as .01, at .02m deep. The second is .08 kyr BP, at .13m. The third is .16kyr. at .24m. The surface rate of accumulation seems to be pretty constant, at about 1/6 cm per year. If 0 BP is 1950, this would place 2006 at roughly 9 cm above the surface. It’s easier to think that their 0BP is about when they did their study, as in one of the 2 official versions of the Calvo table cited by Steve McI here. I can’t get ahold of the Kim paper right now. Does it clarify the table?

So despite what the authors say in their official archiving of the data you’d rather use another date?

From the archive ref : AGE Age kyr BP

It’s all the same to me either way, as long as we get it straight. The downblip at the end of Craig’s series is simply what the data said was happening, as he understood it, and given his rolling mean approach.

Agreed I have no axe to grind but get the data right, it’s not the first time I’ve seen this mistake made, notably with Carbon dating.
I agree that it’s what his data says is happening and I assume based on your comments that it’s the result of his sparse data kicking in and out, however given the mess it makes of the 20th century it doesn’t inspire me with confidence about the rest of the reconstruction!

Sam Urbinto

For those curious about absorbtion bands. The band at about 4 micrometers is the only wavelength where CO2 has no competition. Interesting, isn’t it? Move along, there’s nothing to see here.

yorick

Jimmy

while we’re at the discussion of the 20th century and paleoclimate dataseries – what would the Loehle reconstruction look like if he used the Dye 3 borehole instead of GRIP (Dahl-Jensen et al., 1998)? Why did he pick that one of the two boreholes from Greenland??? He should have at least used an average of the two to be fair…

You can go here to read a discussion of the GRIP and DYE-3 boreholes. It might be that one of them, GRIP, is more recent and uses updated techniques. But if you insist on picking that cherry, go ahead and use the outdated one.

Click to access icecores_palaeoclimate.pdf

Pat Keating

205 Sam
I’m not sure of the point you are making. (Perhaps it’s in response to a post I’m not aware of). Could you elaborate a little?

Sam Urbinto

Pat, generic comment about keeping other things in mind than a gas that absorbs IR.

A quick look at the atmospheric raditation transfer on that chartshows that almost all absorbtion and scattering is due to the things we have no real direct control over — water vapor, oxygen, ozone and rayleigh scattering. Only at ~3.5 um does CO2 show a signal on its own, and a chunk of that is where nitrous oxide also plays a large part (a little wv also) and doesn’t even take particulates into account.

Look at these margin of error, and know why I’ll die happy if nobody ever mentions CO2 again and starts focusing on the system, of which CO2 is a minor part (human created or not)
http://en.wikipedia.org/wiki/Image:Radiative-forcings.svg

Sam Urbinto

I’m just sick of everyone who is making some kind of simple x=y comparison moan about how they’re not doing something when they are. 🙂

This relates to Loehle in general; he took existing proxy papers and combined them, and just like the “But I’m not saying CO2=TEMP+ of course we know there are other variables.” nonsense, we are getting complaints on certain aspects of his paper that apply equally to others; that whole simple answer to a complex question that smacks of hypocracy thing. You know, opinions stated as fact. Par for the course.

Just sayin’.

yorick

Jimmy,
Whatever. Pick your cherry. I don’t care. In 1983 I was listening to vinyl records. The Dye core shows amplitude differences from the GRIP core. You want to jump on them and use a core that ended in 1983 in the place of one that ended in 1995 becuase you are fishing for that uptick at the end. An awful lot of scientists say that the warming signal from CO2 didn’t even show up until 1980.

yorick

If you use ALL data you are bound to get a lot of garbage. This will give you a nice flat climate as all of the random data averages out to an anomoly of zero. Then you can graft on the instrument data and get a nice hockey stick. I think a far better approach would be to select data for quality, not for desired outcomes, such as extending the 1983 peak in the DYE 3 core to the present. I am certainly not angry, nor resentful. It kind of reminds me of how my sister used to come up with one “fair” solution after another to Bush Gore 2000 Florida. The common thread? Gore would be president under every one of them. “Fairness” is no standard.

Hu McCulloch

Jimmy #217 wrote,

My WHOLE point, that goes for ALL reconstructions, is that ALL data available should be used. I don’t think that anyone would disagree when I say there is not enough paleoclimate data. Period.

I can see why one would not use a series if it questionable (ie. BCP’s)… but why leave out good data???

Hopefully in Loehle’s next try (and I’m it seems likely that he will update the 2007 paper), he will keep that in mind.

I’m just trying to get to the truth. That is the point, right???

I agree with Jimmy that tutti frutti is the best way for Craig to go — include the cherries, the lemons, the apples, the kiwis, the uglis, and anything else that comes to hand that has been T-calibrated in a peer-reviewed pubication (or in conjunction with one?). Craig’s signature approach is that he is excluding sticks and twigs (treerings), but then he is making fruit salad, not granola.

As for Yorick’s objection in #218, “If you use ALL data you are bound to get a lot of garbage. This will give you a nice flat climate as all of the random data averages out to an anomoly of zero.”, Jimmy is right in 219 that this will only happen if there is no global signal.

To be sure, the unweighted average will come out with a big standard error. However, this can be objectively improved upon in a second round of weighted least squares, which looks at the variance Vj of each series j about the overall time-specific means (not about its own mean over time as Gavin advocated in his RC note), and efficiently weights each inversely to Vj. The variances I computed using the original, uncorrected Loehle (2007) data are in the second column below. The third column has the corresponding standard deviations (in dC), and the last column has the WLS weights 1/Vj:

1.0000 0.1588 0.3986 6.2953
2.0000 0.2198 0.4689 4.5489
3.0000 1.7430 1.3202 0.5737
4.0000 0.2102 0.4585 4.7571
5.0000 0.2100 0.4583 4.7612
6.0000 0.0417 0.2042 23.9714
7.0000 0.3294 0.5740 3.0356
8.0000 0.1886 0.4343 5.3019
9.0000 0.1480 0.3848 6.7550
10.0000 1.7116 1.3083 0.5843
11.0000 0.3818 0.6179 2.6195
12.0000 0.1028 0.3206 9.7304
13.0000 0.0400 0.1999 25.0284
14.0000 0.3199 0.5656 3.1258
15.0000 0.2669 0.5166 3.7463
16.0000 0.1978 0.4447 5.0561
17.0000 0.0877 0.2962 11.4003
18.0000 0.1078 0.3283 9.2782
It may be seen that #3(Cronin) and #10 (deMenocal) are truly rotten apples that may as well be pitched — and effectively are by WLS. #17 (Farmer) on the other hand, ugli though it may be on the exterior, turns out to add a valuable, if exotic, flavor to the mix.

I just got a note from Christa Farmer (now at Hofstra) that the ages in #17 are indeed measured in BP1950 years. However, I got another from Peter deMenocal, who was a co-author with Farmer on #17, that the dates are in any event so approximate in both #17 and #10 that neither should have been used by Craig. It seems they took a few C-14 dates along the core, the first around 1800 BP, and then extrapolated to where “year 0” (1950) would be. The depths are then measured from this point, wherever it may lie on the core.

While it would be foolish to use such a series by itself to make sweeping statements about climate change, noise is not in itself a problem if one is going to average together several noisy series. So my advice for Loehle II at this point would be to go ahead and throw both of these in, along with Dye 3, Law Dome and any other peer-reviewed T-calibrated series, at least for a first cut unweighted average. But then use WLS in a second round to effectively get rid of the rotten apples while retaining the flavorful uglis.

Note that the criterion being used by WLS is not the conformity of the series to one’s preconception of whether there was or wasn’t a MWP or whatever, but rather its conformity to the signal in the other series. Gavin’s proposed normalization would instead weight each series by how flat it is — a good way to fabricate a nice HS shaft!

yorick

Jimmy,
I see your point. The garbage series should cancel each other, roughly, and the good series should influence the final product.

By the way, the Law Dome shows a nice MWP and LIA in the SH, not that that is a criteria for including it. It just does.

Mark T

The climate series would only average out to zero/flat if the MWP was not global. If it was global, wouldn’t it still be there?

Um, this is true only if the proxies are truly proxies for temperature, the effect is fairly consistent through all time and all confounding factors are accounted for. The series could easily average out to zero if they simply aren’t measuring temperature, or they do so non-linearly, or their statistics exhibit non-stationarity, or the “inputs” are correlated. This is not unlike the whole “you can fit an elephant” idea…

I think yorick unwittingly noticed this with:

and the good series should influence the final product.

Ya gotta have some “good” series to find anything (actually, you need at least enough to describe the space of your inputs lest you wind up with an under determined system).

Mark

steven mosher

There is a benefit to narrowing the estimates of the LIA as much as possible.
There is a benefit to extending the global coverage of this time period.

Eye of the needle.

Larry

When you get that page full of incoherent error messages…..don’t submit again. Hit ‘back’ on the browser, and then refresh. You won’t lose the comment in the box, and 99% of the time you’ll find that the comment took.

Sam Urbinto

“Launching the puck with a trebuchet.”

Larry

“Get the puck out of there”?

Andrey Levin

“Weather Change”