My Final Word on Klotzbach

There is a sort of a weird convention in science that there is a set of things that are known and that they are all about equally reliable. That set of things is by convention what is published in the literature and hasn’t yet been refuted, and by practice what has made it out of the “literature” and into reputable textbooks.

About 200 years ago, that body of knowledge exceeded the capacity of the human lifetime to absorb or the human brain to access. So it’s always been a pretense. It’s a very useful pretense, don’t get me wrong, it’s essentially a very useful shorthand keeping all the “I thinks” and “maybes” out of the already longwinded and hard-to-read materials, but it’s a formalism, not a reality. Nobody, neither humankind nor Googlekind, actually has a grasp on it all.

So when I am asked for a summary of the dialog regarding Klotzbach, Pielkes, etc. I am left with a choice of telling you 1) what I understood and the sequence of how I understood it, which is surely of greater interest to me than to you or 2) pretending to some omniscient voice, saying what is or isn’t a legitimate part of the “body of knowledge”, which feeds the convenient fiction but isn’t justified based on the state of my own understanding.

What I’ve been trying to do is to elaborate what I am thinking as I go along. This may serve some purposes and not others. At this point it would be too long a story, and too tedious. So I’m forced into omniscience mode. Let me preface all of this with admitting that I have a doubt here and there, but people want to have something succinct to work with.

So with that caveat, usually implicit, made explicit, here is what I think about Klotzbach at this point, expressed in the usual scientific-omniscient style:

1) The mechanism described in Pielke & Matsui is surely real in this respect: as greenhouse gases increase and global warming proceeds, the strength of extreme nighttime near-surface inversions will decline. If it is faster than other effects it will contribute to making the surface temperature trend go up without affecting the middle atmosphere trends.

2) It is implausible that this effect is large enough in the aggregate (common enough as a fraction of space and time) to account for discrepancies in global trends in GCMs. It would take quite a lot of serious revisiting of boundary layer theory and boundary layer implementation in models to quantify the expected effect to demonstrate this one way or the other, work that the Pielke crowd has not undertaken.

3) The effect is most effective in northern latitudes (where there is a lot of cold land) and can’t explain the tropical hot spot, so it’s in the wrong place.

4) The word “bias” is used in a rather atypical way and does not refer to measurement errors, despite being reported as such in the blogosphere.

5) The first of the tests proposed in K et al

1. If there is no warm bias in the surface temperature trends, then there should not be an increasing divergence with time between the lower troposphere and surface temperature anomalies. The difference between lower troposphere and surface temperature anomalies should not be greater over land areas.

is peculiarly stated. Applying basic logic (A implies B if and only if not B implies not A) we can restate the first sentence more clearly for consideration. “If there is an increasing divergence… THEN there is a warm bias”.) This doesn’t seem right. There could be other mechanisms that account for it. In fact, Gavin Schmidt provides a very simple one: the influence of more slowly warming oceans is stronger in the troposphere than at the surface.

6) Applying the same transformation to the second test:

2. If there is no warm bias in the surface temperature trends then the divergence should not be larger for both maximum and minimum temperatures at high latitude land locations in the winter.

yields “If the divergence is larger (in cold conditions) then there is a warm bias” again ignores any alternative mechanism, but this is in fact a much stronger test, since it is relatively far-fetched to suggest an alternative. Remarkably, after listing this test, K et al does not address it .

7) For evidence that the proposed effect is small, one need only go to the remarkable Parker 2004 paper Parker, D. E. (2004), Large scale warming is not urban, Nature, 432, 290 where the proposed effect is strikingly absent. Pielke & Matsui argues that Parker must be revisited in the light of the new results, but this presumes that the new result is quantitatively significant at the large scale. A number of strong assertions are made regarding the large scale amplitude, but the justification of these estimates is obscure.

The fact that the Parker result comes out with this effect pretty clearly pegged to near zero is claimed to “require further analysis”, but any quantity other than practically zero would require a precisely balancing bias (and in this case I really do mean bias) in the opposite direction. Here one may appeal to Occam.

8) The Eastman et al paper makes flawed use of limited area model in a climate study. The reference to it in Pielke and Matsui, as well as its very small sensitivity, doesn’t build confidence. The enhanced warming under low wind appears in the Eastman model for a reason related to the reason its sensitivity is so small.

The sensitivity in the Eastman model is small because the residence time of a particle in the domain where the enhanced greenhouse effect is active is small compared to the time constant of radiative equilibration. The sensitivity in the Eastman model is larger under low wind conditions because under low wind conditions the residence time is longer, exposing the column to a larger cumulative greenhouse forcing. Therefore Eastman provides no support for a boundary layer mechanism.

Or anyway, so I think.

That is my summary of Klotzbach. Pending someone doing better, I have no further interest in it.

If someone wanted to pursue it further, I’d strongly recommend they replicate the Parker study. That’s a straightforward data analysis problem. If Parker’s numbers aren’t drastically mistaken, it’s very difficult to imagine how the Pielke-Matsui effect can have much importance at a global scale.

The Pielke group uses nomenclature (‘bias”) peculiarly, happily references weird (Eastman) and marginal ( results, and in the case of Klotzbach et al, fills a number of pages in a way that is scattershot and does not create a compelling argument. While they are cordial and patient, conversation is time consuming and pedantic, largely because of idiosyncratic use of standard words and ideas. If they really have something to offer, they should polish their presentation technique. Meanwhile, dealing with their work seems an ill-justified expenditure of effort.

This is a big problem with politics mixing with science. If there were no policy issues at stake, if the modest and dubious results of the paper weren’t being egregiously overvalued and misrepresented, the scientific community could proceed in the ordinary dignified fashion of ignoring nonsense and focusing on sense. But with the standard Pielke to Watts to Morano to Inhofe play and its like, we are forced to pay attention.

Regarding getting out of this very unfortunate flavor of time sink, one thing I can imagine is to have gradations of “peer review” more complex than “published” or “unpublished”. Inhofe shouldn’t be waving something like this around in the senate claiming it meets the highest standards.

Even harder, but more urgent, is have a mechanism to dissuade (was “prevent”, see * below) authors from promoting public misinterpretations of their publications. That sort of behavior should have consequences. Making a boundary between PR enthusiasm and untruth is perhaps difficult, but then again, there are instances that clearly cross the line; a reader of Watts’ site could be forgiven for believing that the Pielkes claim to have provided evidence that the surface temperature trend is wrongly overstated, when they have not done so.

Update: (aargh) A few minutes before I composed this, RP Sr published an article on his own site entitled exactly “The Global Average Surface Temperature Warming Really Is Overstated”!

Sorry. No.

The only caveat offered is that the bias is due to “measuring the temperature near the ground”. This is not to say the thermometer is corrupted by being near the ground. Pielke Sr is saying that the temperature itself is corrupted.

Apparently we should inform all plants and animals and glaciers that global warming is an illusion, and that they should simply stop biasing their existence to be so close to the earth’s surface. In fact, if the glaciers and ice sheets would merely move their bases above the boundary layer, the ice melt problem might go away entirely!

Sorry, no. The thing about “surface” temperature is that it’s at the surface. So that’s where the thermometers go.

Meta-Update: Note that on RP Jr’s blog RP Sr is quoted as saying “To emphasize, we are not saying that the surface temperatures trends are not accurately measured at that level. It is just that they are NOT representative of long term trends higher in the troposphere. Their use to characterize a deeper atmospheric layer introduces a bias.” (Comment 12. Emphasis added.) As I’ve said it seems like a dozen times already, that is not the impression being left in places like Watts’ or d’Aleo’s blogs. And that’s a big problem, especially if this ends up as a celebrated publication for reasons completely irrelevant to its contents.

Update: Both Pielke Sr and Pielke Jr point to this article, much to my surprise. They are unwilling to have me disengage, even though at this point I am happy to leave them alone.

Sr is on about me not doing the quantitative analysis to support my point 2. Stipulated.

Given how difficult it is to extract sense from people who think an actual physical quantity is a biased estimator of a nophysical quantity of interest only to themselves, and given the fact that the Parker result strongly supports my quantitative intuition that any such phenomenon, even if unrepresented in models (an implicit claim unexamined as yet) is going to be dramatically too small to carry the vast weight that is expected of it, I leave it to the Pielke team to provide the quantitative analysis in a form that a reasonably informed person could hope to understand.

My refutation is Parker ’04, and anyone hoping to undo my refutation should undo Parker, as I already suggested above. Failing that, I can’t imagine why anyone would bother trying to grind through all this at best eccentric and opaque exposition, much less myself.

Meanwhile Jr is on about my plea for “consequences” constituting a “threat” and a pathological one at that.

It’s not a threat, it’s taking note of a missing feedback on extra-scientific behavior. Sopecifically, I refer to the newfound trend of putting one thing in print and claiming it means something entirely different in public. This is behavior which reflects badly on science and weakens it. It should have consequences for scientific reputation.

Is Pielke Jr suggesting to the contrary that one should be allowed to publicly misrepresent one’s results with impunity?

Presumably not. Presumably he will claim that he is not misrepresenting anything. But if he can show that, my wished-for “consequences” wouldn’t apply to him.

I stand by my plea for consequences in the general case. If science goes extraterritorial, then reputation should follow. I will have more about this in the context of McLean/deFreitas.

(*) However, to reduce the implication of threat, I have changed the word “prevent” (which implies prior restraint, which I think is not a good idea) to “dissuade”.

Update 8/21: James Annan points to evidence that even the conceded point 1 is overestimated by several orders of magnitude in Pielke & Matsui.