One excellent feature of the Alaskan varvochronologists is that (unlike, say, Bradley and his coterie) some of them show and archive their work. The Kaufman student MSc theses are good at this. So too is Michael Loso’s work on Iceberg Lake. Thus while one can raise an eyebrow at (and criticize) their statistical peregrinations, at least they provide enough material that one can at least analyse the data (Bradley’s 1996 C2 data remaining under security.) In an earlier post, I discussed Loso (2006). Today, I’ve got a few comments on Loso (2009), which I did not discuss in my earlier post. Loso (J Paleolim 2009) revisits Iceberg Lake varves with some interesting comments (without however satisfactorily resolving the fundamental inhomogeneity problems.) Like many paleoclimate articles, though the author is not a statistician, the topic is primarily statistical. And like most paleoclimate articles (also mainly statistical), there is no evidence that any reviewers were statisticians.
One of the fundamental inhomogeneities that plagued Loso (2006) was the changes in lake level, size and shape. Loso (2009) here revisited that problem, developing the scatter plot between the distance of the core from the nearest inlet (using geological information to derive this distance) and the varve thickness.
Loso Original Caption: Fig. 3 Relation between the distance separating the inlet stream and sampling site and average varve thickness. Each point (n = 42) represents the average varve thickness (mm) for a contiguous group of varves at one sampling site. For each point, distance from sampling site to nearest inlet (m) is based upon the position of the shoreline at the time the varves were deposited, taking into account the known history of shoreline variability. Contiguous period represented by each point ranges from 9 to 174 years (mean = 78.7 year; SD = 66.1 year). Solid line is a power-law curve fitted to show trend (y = 622.09 * x-0.66662), r2 = 0.84
Loso discusses this figure as follows:
Because of Iceberg Lake’s history of episodic changes in lake level (Loso et al. 2004), sediment accumulation rate at any given sampling site will vary not only as a function of bulk sediment input (controlled in part by climatic factors of interest), but also as an inverse function of lake size. This is because, when the lake shrinks, sediment inputs are deposited over a smaller surface area, and shoreline regression brings stream inlets closer to sampling sites in the lake bottom. To demonstrate this relationship, I analyzed the relation between distance to nearest stream inlet and average varve thickness during the period of well-dated shoreline occupations (1825–1998 AD, Fig. 3). To do this, all well-dated varve measurements (in this case including those excluded, on the basis of textural anomalies or other criteria, from the master chronology) were separated into contiguous groups so that each group represents deposition at a single sampling site during a single shoreline occupation. The average thickness of all measurements in a group are plotted against the distance from that sampling site to the nearest glacial stream inlet at the time of that particular shoreline occupation. The results (Fig. 3) show clearly that varve thickness, and hence sediment accumulation rate, increases in a nonlinear fashion as proximity to inlets shrinks.
[UPDATE Sep 23 11. 40 pm: No wonder this observation by Loso seemed so sensible. It was first hypothesized at Climate Audit two years ago!
The connection between inlet distance and varve thickness at Iceberg Lake was suggested by Willis Eschenbach at Climate Audit two years ago here:
My own feeling is that the dropping of the lake level in 1957 is the key to the greatly elevated values in recent times. It coincides exactly with the huge jump in the varve thickness in 1957. I think there are two reasons for this. One is that the distance from the inflow to the core sites is greatly reduced, which would make a permanent increase in the varve thickness.
[end update]
This effect is obviously a MUCH stronger effect than the slight effect (if it even exists) between varve thickness and temperature. Indeed, the previous attempt to establish a relationship between varve thickness and temperature is obviously compromised (as discussed in our earlier post on this topic) by inhomogeneities in the lake. The measurement of the distance to the nearest inlet is probably not a measurement that would be available in most varvochronologies, but Loso doesn’t mind doing geology and seems to have done as reasonable job of estimating this value as one could expect under the circumstances.
So far so good (though it raises obvious concerns for the overall varvochronology project.) But having raised this important confounding factor, Loso is then faced with a non-trivial statistical problem of disentangling two factors in order to reconstruct temperature. And here Loso loses his way.
Loso’s remedy for the situation is merely to log-transform all the measurements:
The net effect of these shoreline changes is a heightened sensitivity to summer temperatures during warm periods that, when combined with the known non-linear relationships among stream discharge, suspended sediment concentration, and varve thickness (Gilbert 1975; Meade et al. 1990), contributes to the strong positive skew of the raw varve thickness measurements (Fig. 4a). … To compensate for this heightened sensitivity, which violates the goal of stationarity—a constant relationship over time between climate variable and proxy response—in proxy reconstruction (National Research Council 2006), I log-transform the raw measurements from the master varve chronology.
Yes, non-normality is a major problem with this data set. (Loso notes that a log-transformation still doesn’t get to normality, a point observed in my graphic yesterday showing the best fit of various distributions to Loso’s varve widths.) Loso’s adjustment also comes AFTER the taking of an average in the annual chronology – I’d be inclined to do so BEFORE making the chronology.
But that’s not the real issue – the real issue is that log-transformation does NOTHING to disentangle the fundamental non-stationarity identified by Loso. In order to deal with the confounding factors, Loso would have to use some sort of mixed effects model. And the trouble with that is that it would only work for the limited period in which Loso can reconstruct the nearest inlet distance.
It’s too bad that the senior paleoclimate community (Bradley etc.) are so enormously stubborn about statistical criticism. Instead of conceding and responding to Wegman’s criticism, it’s almost as though they’ve redoubled efforts to keep statistical outsiders away from the field. And thus we see relatively elementary defects repeated time after time, while industry leaders like Kaufman refuse to discuss such matters, in effect putting their fingers in their ears and saying “Nyah, nyah, I can’t hear you”. It’s bad enough for they themselves, but the worse side-effect is that it becomes very perilous for younger scientists to participate in such discussions.
Loso 2009 online here. Willis Eschenbach is acknowledged (h/t bender(:
Climate Audit 
42 Comments
See my recent comment on the comparable problem in dendroclimatology. A much better treatment that is still not ideal, and possibly wildly incorrect.
When there is a tight knit community with a strong consensus they can be blind to what is very obvious to those outside of the group. I’ve noted that you have found some good work in various master’s theses. Maybe there is hope for the next generation.
So lake varve thickness might actually be a much better proxy for inverse inlet (shoreline) distance which might be a reasonable proxy for precipitation which possibly could be a proxy for temperature.
Dakota tribal wisdom (‘no proxy is too dead to beat’) suggests that increasing temperatures increase humidity which increases precipitation which fills lakes which increases the area available for sedimentation thereby decreasing varve thickness.
Thus Loso A is a fallacious proxy and should be replaced by Loso M which should immediately be discarded because of its lack of conclusiveness.
What is the cause of the variations in the lake’s level? Is it possible that the climatic signal might be imposed through lake-level variations so that you wouldn’t want to treat lake-level as a confounding variable?
though the author is not a statistician, the topic is primarily statistical. And like most paleoclimate articles (also mainly statistical), there is no evidence that any reviewers were statisticians
This is arguably one of the most indefensible of all the issues in Climate Science. Reviewers appear to be either ignoring math problems or perhaps they just don’t see them. The peer review process clearly does not function as advertised if the math is not scrutinized by math experts.
Sorry. Please delete the previous post – I left something out.
Steve:
You have done a great job explaining this. It appears to me though that Arctic Lakes that vary significantly in level, especially the smaller ones, all will suffer from this problem making them very difficult to use as proxies.
This appears so obvious now that you have laid it out that perhaps you have also uncovered the reason why the original research plan kind of “enigmatically” morphed?
Distance to inlet?? That’s pretty scientific. Any granulometry??
I think you might have nailed the team lynchpin here. It’s time to stop bashing Mann and start asking which words in these papers and which decisions in these projects are Bradley’s.
Many thanks for this, Steve. I suspect I’m learning more about varvochronology than if I’d continued with my geology course.
The geology of varvology is very interesting, so from my viewpoint there has been merit in examining these papers.
Using varves as a palaeo-thermometer to trash the MWP is several bridges too far, though.
Oh dear. Logging does not alter the degree of “stationarity” (i.e. uniformity) of the response coefficients – and it also does not stabilize variance. These inhomogeneities are preserved under all monotonic functional transforms! Who was this guy’s supervisor?
Re: bender (#11),
Quite so and quite obviously. It’s shocking to see a statement like Loso’s – and here it’s in a specialist journal and not a popular digest like Nature or Science.
Loso is staff in Alaska, not a Kaufman student. The problem is not with any of these folks as individuals or even with their supervisors. Despite clangers like this, Loso’s articles have a number of good points to them – particularly because he tries to give enough information so that other people with different perspectives can consider the site.
I don’t even blame Loso for this particular clanger. The people to blame in this are the senior people like Bradley and Jones who didn’t view Wegman’s comments as an opportunity rather than a reason to circle the wagons. As we’ve discussed endlessly, there is overweening hubris in the field where they think they know things that they don’t – right through to the journals and reviewers.
Combined with the fact these folks are all guys/girls that love the outdoors and care about the earth and all that. Maybe they’ve taken a statistics course or two, but they’re not numbers people. Rob Wilson didn’t become a dendro because he wanted to sit at a desk and think about statistical distributions. The trouble hits the road when they do their articles, which are 99% statistics, but neither the authors nor the reviewers nor the editors really can go much beyond using canned recipes.
Kaufman is a classic example. And rather than engaging with the statistical community here while this topic has our attention, he’d rather stick his fingers in his ears and say that we don’t play nice.
Re: Steve McIntyre (#12),
I’m sure Loso, and even Kaufman, are great guys. I’m equally sure that it’s high time for a “Journal of Statistical Climatology”. And I know which senior leaders WON’T be on the board: Mann, Bradley, North, …
[If they were REALLY smart, they would pre-empt us and take the name for themselves.]
Re: bender (#13),
Right now that journal is called ClimateAudit
#12
Absolutely true in my experience. The people who are drawn to these fields don’t really understand the mathematical analyses. They apply mathematics in “cookbook” fashion and can spout the intimidating lingo but lack the basic insight necessary to match the physics with the models.
Loso 2009 online here.
Re: Steve McIntyre (#15),
typo in the link: .pdf
It’s stated right there in the abstract! The abstract!? For pete’s sake …
You don’t say? Smooth move.
Calibration of Loso’s varves:
However, at a gridcell level, the correlation to the CRUTEM gridcellis -0.006.
Re: Steve McIntyre (#19),
Not bad. -0.06 or -0.6 would be really damning.
Uh oh.
Well, at least the uncertainty wasn’t flat out ignored. We’re making progress!
Hey, Willis is in the acknowledgements!
Steve: my,my….
Re: bender (#23),
The connection between inlet distance and varve thickness at Iceberg Lake was first reported by Willis Eschenbach at Climate Audit two years ago here:
Re: bender (#23),
Now THAT’S progress! If one of these young scientists takes the next logical step and realizes that their paper would be practically bullet proof if they have their paper Climate Audited prior to publication…
Re: Patrick M. (#32),
The mudmometer looks to have as many problems as the treemometer. Seriously bender how many promising young mud scientists want to make their mark on the world by detailing how mudmometers fail. No is not an acceptable answer. They are driven to say YES; yes, we can recover a weak climate signal from this morass.
They have the will to believe.
http://falcon.jmu.edu/~omearawm/ph101willtobelieve.html
You know what? This paper actually tries to be honest. I don’t understand, given this paper and given Tiljander, what Bradley thinks he’s up to.
Re: bender (#24),
Loso has a background in geology rather than climatology. Geologists get a very different training.
I’ve said on several occasions that there was much to like about Loso’s approach (other than his stats). I’m sure that he’d like to do a better job on his stats because he’s actually trying.
Those CI’s on Fig 6 are awesome.
Is it me, or could one fit any curve to that collection of data points?
Re: Andy (#30),
With a correlation of 0.006 you put the Mona Lisa in there…
SMc quote:
“Geologists get a very different training.”
Agreed … I know I can’t make a fault disappear by deleting inconvenient drillholes from my map. And any number of accountants have exhorted me to do just that 🙂
Steve,
I assume you are already aware of this, but just in case:
http://article.nationalreview.com/?q=ZTBiMTRlMDQxNzEyMmRhZjU3ZmYzODI5MGY4ZWI5OWM
Apparently the masters at CRU are claiming they lost the original raw data which was used to “calculate” the +0.6 C increase last century. Oh, my.
A little OT; but kinda not really….
Is there any physical reason for the logging? Or just a way to make some number series correlate better?
Let’s assume logging creates a great correlation (and the other stats are ok too), does that make the reults any more valid (i.e. if there is no physical reason for doing so)?
Steve: The underlying data is hugely non-normal. Correlation statistics implicitly presume a comparison between two normal distributions. I’m OK with them trying to normalize prior to doing a correlation – indeed I’d prefer that they did this even if (and perhaps especially if) the correlation deteriorated after logging. In this case, logging isn’t enough to fully normalize the data and some sort of non-parametric normalization is needed. Plus as noted elsewhere, logging doesn’t do deal with inhomogeneity and Loso (and his editor, Kaufman) have simply gone astray on this point.
Opps, Sorry Patrick I meant to say your name and not bender’s.. arrg
Re: steven mosher (#36),
I knew by context you were addressing Patrick.
.
I think varvometry has WAY more problems than dendrometry. I think these stats show it. Would love to hear from a real dendro on this topic.
.
Read Lamoureux and Bradley (1996). It is an honest appraisal. But there is this one line, which I can only imagine came from Bradley, where he argues, bizarrely, that the burden of proof is on the other guys to show sediments do NOT correlate with temperature; i.e. until shown otherwise the varves should be considered proxies for temperature. Really, that’s what he says. So Tiljander does her thing, and Loso his … and … nothing changes! It’s still a valid proxy?!? Talk about “denial”. Can’t wait to see the IPCC chapter on paleovarvoclimatology.
Steve, thanks for the feedback re: logging.
But, the question remains, if there is no physical reason for a logarithmic relationship between X and Y, what possible scientific value would come out of carrying out such an operation?
I appreciate the acknowledgement of Mike Loso, it was a lovely and totally unexpected gesture. By way of history, I had emailed him that I thought I had found several errors in his work. He provided his original data, acknowledged that I was correct, and fixed his errors. He was a gentleman throughout, would that all climate scientists were so forthright and forthcoming.
.
I did not discuss with him the idea that varve thickness would be affected by the varying distance to the nearest rivermouth. Either he came up with it independently, found it elsewhere, or he read it here. In any case he ran with it, did the hard yards, and found an association … but that just compounds the problem. That just makes it harder to read the varvometer, it’s another confounding variable.
.
The problem is that the mouth of the river changes and wanders about over time. In silt-laden rivers, a delta rapidly develops. A single storm can divert the main channel into the lake from one side of the delta to the other. This capricious wandering of the rivermouth can change the silt load at a given point by a huge amount in a very short time. My own feeling is that this cannot be adjusted for, and makes varve thickness a very difficult proxy to use for anything.
Perhaps I don’t get it, but the points on the top graph in the leader approach 5 km apart. That is the separation of the cored point from the “nearest inlet” point of the sediments, which is sometimes a few tenths of a km. Noting that sampling points H and L are not shown on the fig 3 map, the sampling points are mostly within 1 km of each other. Moving 5 km from the centre of this cluster to an unshown “nearest inlet” takes one well away from the lake as it is at present. The main inlets that are shown are mostly within 1 km of the sampling cluster. The graph is not capable of interpretation without the provision of more data and hence cannot be verified.
Re: Geoff Sherrington (#40), while the current shoreline is within about a km of the core sites, the historical shoreline is not. Here’s the historical situation:
As you can see, the distance from the inlet to the cores has changed markedly over that time.
w.
Two items I do not see in this string.
1. The lake level variation for this lake is related to this being a glcier-ice dammed lake. It periodically partially released (drained) in sudden outbursts. The lake level changes were not seasonal precipitation related per se. I have documented with satellite imagery that the final complete drainage of the full lake, enabling Dr Loso’s varve work, occured in late 1999 in less than 48 hours.
2. Dr. Loso works in close association with two statisticians in his department at Alaska Pacific University. One is specifically trained in mathematics, with a PhD in the subject from Stanford. How much they were consulted I can not say, but casual discussions with co-workers often contribute to scientific theory development without specific acknowledgement.
One Trackback
[…] log values in his temperature reconstruction, a device that was irrelevant to the inhomogeneity (as pointed out at CA at the time). In Loso’s most recent publication (Dietrich and Loso 2012), Loso […]