Monday, January 21, 2013

BEST laugh of the day

I've barely commented on the Muller BEST thing, primarily on the grounds that there is nothing there worth commenting on. There might be something mildly interesting in the details of their approach, but it was clear from the outset that (assuming they didn't mess the whole thing up) they wouldn't reach significantly different conclusions than the several other groups who have been analysing the surface temperature record for years. Certainly, there was nothing to justify the extraordinary press coverage they received before they had even done any analysis.

Now at long last it's published. Previously submitted to JGR, it has ended up...as the first article in a newly-created fake vanity press "journal". That's a few rungs down from just sticking it on the Arxiv, where (1) it is free (2) at least some people will read it (3) no-one will think you are pretending it's undergone any meaningful peer review. No wonder Curry has pulled her name from it. The surprise is that the others have not.

I'm actually half-wondering if the journal has stolen this manuscript as a way of generating content. Have any of the authors commented?

H/T: Stoat

36 comments:

andrewt said...

Its not a good sign when the publisher's mission statement is full of poor/incorrect grammar.

James Annan said...

Yes, it's hard to see how they could have done more to lower their impact and reputation. I wonder how on earth the journal was chosen?

Hank Roberts said...

Wow:

"It promotes awareness to all arenas of public to encourage research and aid scientific community by executing the significance of latest research."

"Would you like to register as a peer reviewer?"
"Would you like to register as a peer reviewer?"

Hank Roberts said...

Oh.

Wow.

http://scholarlyoa.com/2012/05/05/omics-publishing-launches-new-brand-with-53-journal-titles/
__________________________
"India-based OMICS Publishing Group has just launched a new brand of scholarly journals called “SciTechnol.” ...

We learned of this new launch because the company is currently spamming tens of thousands of academics, hoping to recruit some of them for the new journals’ editorial boards....

OMICS Publishing group has exploited many young researchers by inviting them to submit article manuscripts, leading them through the editing and review process, publishing the article and then invoicing the author.

In most cases, the authors have no idea that an author fee applies until they receive the invoice....."
___________________________

http://scholarlyoa.com/2012/04/09/open-access-bait-and-switch/

William M. Connolley said...

> I'm actually half-wondering if the journal has stolen this manuscript as a way of generating content. Have any of the authors commented?

They link to it from the BEST homepage (http://berkeleyearth.org/available-resources/) so this must be legit.

James Annan said...

I suppose they might have just worked their way through the decent options (and been too chicken for Climate of the Past, where the topic would fit very naturally, but the paper could have expected a proper review).

David B. Benson said...

Something Is Not Right.

Alex Harvey said...

Dear James,

This isn't about BEST per se but about the temperature record in general.

It is becoming more accepted that the tropospheric temperatures aloft are not following model predictions (quite) and the discrepancy needs to be resolved one way or another.

I understand that claiming that the models are wrong in their prediction of the "hot spot" is untenable.

Pielke, Klotzbach et al. have of course argued that there may be a warm bias in the surface record, and I recall you got involved in some way in that particular controversy.

Recently, Lindzen has added his voice to this position, and I think put forward an argument of his own. I'd be interested in your thoughts on this passage from Lindzen's most recent paper -

The response is characterized by the so-called hot spot (i.e., the response in the tropical upper troposphere is from
2–3 times larger than the surface response). The models are likely correct in this respect since the hot spot is simply
a consequence of the fact that tropical temperatures approximately follow the moist adiabat. This is essentially a
consequence of the dominant role of moist convection in the tropics.
However, we see in fig. 9 that the temperature trends obtained from observations fail to show the hot spot.
In point of fact, it seems likely that some of the recent temperature data must be wrong!
The resolution of the discrepancy demands that either the upper troposphere measurements are wrong, the surface
measurements are wrong or both. If it is the surface measurements, then the surface trend must be reduced from “a” to
“b”. Although, it is generally ill-advised to estimate climate sensitivity from observed changes in temperature (simply
because of ignorance of all the relevant processes including especially natural internal variability on time scales of
centuries or less), it would be very difficult to simulate the trend at “b” with models having their current sensitivity.
Given how small the trends are, and how large the uncertainties in the analysis, such errors are hardly out of
the question. In fact there are excellent reasons to suppose that the error resides in the surface measurements. To
understand this requires an awareness of the Rossby Radius of Deformation.
3.2 Rossby Radius of Deformation
In dynamic meteorology, there is something called the Rossby Radius. It is the distance over which variables like
temperature are smoothed out. This distance is inversely proportional to the Coriolis Parameter (twice the vertical
component of the earth’s rotation), and this parameter approaches zero as one approaches the tropics so that
temperature is smoothed over thousands of kilometers (detailed formulas can again be found on Wikipedia).
However, this smoothing is only effective where turbulent diffusion is small. Below about 2 km, we have the turbulent
trade wind boundary layer, where such smoothing is much less effective so that there is appreciable local variability of
temperature. In practice, this means that for the sparsely sampled tropics, sampling problems above 2km are much
less important than at the surface (Lindzen and Nigam [5]). Thus, errors are more likely at the surface.
An important philosophical point to this little exercise is that neither ambiguous data nor numerical model outputs
should automatically be assumed to be right or wrong. Both should be judged by basic, relatively fundamental theory
—where such theory is available.


Do you or your readers have any comments?

Paul S said...

Alex,

Regarding the presence of the hot spot on an atmospheric cross-section, this is a function of differential temperature trends in both horizontal and vertical dimensions. The reason long-term model output produces a hotspot over the tropics is that tropospheric amplification (from surface warming) over that zone is at its maximum. Zonal tropospheric amplification over the extratropics is expected to be much smaller, less than 1 in the high latitudes.

As a simple example, if we say tropospheric amplification over the tropics = 3 and over extratropics = 1, and surface temperature trends in extratropics are 3 times greater than in the tropics, you wouldn't expect to see a hotspot: the mid-upper troposphere would have the same temperature in all zones. The point being that there is a dependence here on the spatial distribution of temperature trends at the surface. The simple example I've given is actually not far from what is actually observed.

From what I can see, most of the discrepancy between what models predict for differential zonal atmospheric temperature trends and what is found in observations is a result of a similar spatial discrepancy at the surface. In other words the issue here has little to do with surface vs. satellite trends, and is mostly simply a function of the specific zonal pattern of warming we've seen at the surface over the past thirty years.

Steve Bloom said...

Oh, it's AH with Lindzen's latest. That's way more deja vu than I can handle. Kill me now.

Anonymous said...

The OMICS connection of SciTechnol noted by Hank Roberts above is bad.

On another OMICS publication a fictional character became a member of the editorial board. See here for details:

http://phylogenomics.blogspot.com/2012/01/scary-and-funny-functional-researcher.html

James Annan said...

I'm afraid that Lindzen's signal to noise ratio is sufficiently low that I would need some persuasion to bother looking at it. Where is this "most recent paper" published?

Steve Bloom said...

"(...) executing the significance of latest research." Nice catch, Hank!

Steve Bloom said...

And bizarrely, AH is not so interested in the fresh validation of Dessler (2010). Imagine that.

James Annan said...

What is the Dessler thing? Link?

Steve Bloom said...

Had to wrack my brain for where I saw it referenced since it doesn't seem to have appeared in Google Scholar yet, plus Andy seems not to have updated his site recently enough to show it, but here it is. This may also be relevant.

Steve Bloom said...

There's also this at least somewhat related paper, which was on his site.

FWIW I did an exact phrase search on the Lindzen passage AH quoted and turned up nothing.

Alex Harvey said...

Dear James,

I forgot to provide a citation. The paper is

R.S. Lindzen (2012): 'Climate physics, feedbacks, and reductionism (and when does reductionism go too far?)', European Physical Journal Plus, DOI 10.1140/epjp/i2012-12052-8.

I quoted the relevant section in full. The Lindzen and Nigam paper cited therein is (the much cited) paper -

R.S. Lindzen and S. Nigam (1987). 'On the role of sea surface temperature gradients in forcing low level winds and convergence in the tropics'. J. Atmos. Sci., 44, 2418-2436.

Alex Harvey said...

Paul S.,

Thanks for your comments.

The point being that there is a dependence here on the spatial distribution of temperature trends at the surface. The simple example I've given is actually not far from what is actually observed.

From what I can see, most of the discrepancy between what models predict for differential zonal atmospheric temperature trends and what is found in observations is a result of a similar spatial discrepancy at the surface.


Although I don't fully understand your point, it does seem that you are not actually contradicting Lindzen here. Wouldn't it follow from your point as well that the surface record is less likely to be reliable? In any case when you say this is close to what is observed do you have a paper in mind that says this? Only reason being I'd just like to understand this better.

Steve Bloom said...

Soon-to-vanish spam above: "Troll Images of Girls"

The imagination runs wild. :)

James Annan said...

So does the google :-)

Magnus said...

How novel is this approach to get the sensitivity?
http://onlinelibrary.wiley.com/doi/10.1002/env.2140/abstract;jsessionid=FDF33A1FB2762A4213CD031F01EF2A10.d01t04

2000 figures make the difference
When the researchers at CICERO and the Norwegian Computing Center applied their model and statistics to analyse temperature readings from the air and ocean for the period ending in 2000, they found that climate sensitivity to a doubling of atmospheric CO2 concentration will most likely be 3.7°C, which is somewhat higher than the IPCC prognosis.

But the researchers were surprised when they entered temperatures and other data from the decade 2000-2010 into the model; climate sensitivity was greatly reduced to a “mere” 1.9°C.

Magnus said...

Sorry might be wrong paper...
http://www.forskningsradet.no/en/Newsarticle/Global_warming_less_extreme_than_feared/1253983344535/p1177315753918

James Annan said...

Looks like it might be that paper, though strange timing if so...

Andy Revkin has a post and quotes me in a comment.

Magnus said...

Thnx, I have seen that you had a look at the IPCC part dealing with this... wich seams to me wont change the ECS much... might cut uncertenty in the high end... now that might change... ofc but a bet on 3 c (2-4.5) in ECS in the new IPCC report seams to be a good one?

James Annan said...

Well some of us might have some "inside info" in that we've seen the drafts...though they could change...so betting is probably not a great idea at this point :-)

Magnus said...

True... btw, no new study jet if you missed that.

Paul S said...

Although I don't fully understand your point, it does seem that you are not actually contradicting Lindzen here.

As far as I can see Lindzen is simply referring to a much-raised issue, and then speculating on a possible cause. There isn't much to directly contradict. My point is a more general one: there are really two separate issues here that are often confused on climate blogs.

Wouldn't it follow from your point as well that the surface record is less likely to be reliable?

No, I can't see how that would follow. Essentially what I'm saying is that the mid-upper troposphere temperature change at any particular location can, to some extent, be calculated as a function of surface temperature change (SAT), e.g.

DeltaTropos = c * DeltaSAT

where c is some coefficient relating to the expected Tropos/SAT amplification over a particlar region. The presence of the infamous red spot over the tropics is thus dependent on the distribution of temperature change at the surface - a large bias towards high latitude NH warming and relatively little in the tropics, as we see in SAT+SST observations, is a pattern in which a prominent tropical hotspot would not be expected.

-----------------

The second issue is whether the mid-upper troposphere over the tropics specifically is warming in-line with expectations due to surface warming in the tropics. It seems to me that AMIP model experiments using prescribed SSTs from observations provide a good route for interrogating this question, as described in an Isaac Held blog post. Prescribing SSTs means that surface warming in the models has a similar spatial distribution to that seen in observations. You can see in Held's post that the model output is very similar to RSS-observed TLT data.

This is where I would suggest that Lindzen's speculation concerning the accuracy of SST measurements appears unfounded. I can't see how the model could so faithfully reproduce the variance in the observed TLT time series if surface SST measurements were not sufficiently representative.

This is not to say the issue is at all settled, not even close. Po-Chedley & Fu 2012 provide a wider examination of the AMIP models against observations. Also, comparisons between models and satellite TLT data can't definitively validate since the satellite data doesn't have a great enough fidelity to isolate the particular vertical region of interest: instead it picks up varying trends all through the atmosphere and is a weighted average. Theoretically I think you could get similar TLT time series form very different vertical temperature trend profiles.

Paul S said...

James,

Regarding the Aldrin paper and the report that climate sensitivity reduces from 3.7 to 1.9 when 2000-2010 data is included...

wouldn't that mainly indicate a problem with the method? Making small variations to the scope of input datasets is a common test of robustness, which this method seems to have failed.

EliRabett said...

"No, I can't see how that would follow. Essentially what I'm saying is that the mid-upper troposphere temperature change at any particular location can, to some extent, be calculated as a function of surface temperature change (SAT)"

Except where the constant changes. See 1998.

Paul S said...

Eli,

I'm talking about long term trends over the whole record rather than inter-annual fluctuations. If the theoretical coefficient fluctuates it will still average to a single value over the whole period.

That said, this possibility of fluctuating coefficients is why the situation is better analysed through AMIP GCMs rather than simple statistical models.

James Annan said...

Paul, yes I'm a bit surprised at the 3.7 figure, as it does not seem consistent either with the published Aldrin et al paper or anything else of a similar ilk (which generally find a best estimate of under 3C, even for the older estimates with less data and high tails, eg fig 9.20 in the IPCC AR4).

Paul S said...

James,

Perhaps stupidly, I hadn't considered the press release might simply be wrong. I don't have access to the Aldrin paper but page 25 of this slideshow appears to indicate their posterior mean estimate is remarkably insensitive to a change of scope, whether using data up to 2009, 2000 or 1990.

The way these things go I wouldn't be surprised if this 3.7 figure was a misunderstanding of the paper or someone involved quoting the accepted radiative forcing of doubled CO2. The only other obvious possibility I can see is that 3.7 might have been the posterior climate sensitivity found for data up to 2000 when including a -0.5W/m^2 cloud lifetime term.

Magnus said...

The old Aldrin paper is all there is at this moment. the other comes from unpublished material... What that will say when it is published is hard to know.

Paul S said...

Found a copy of the actual paper.

A couple of things I've noticed: Their fitted posterior, from which they presumably derive the sensitivity estimate, appears to significantly underestimate NH warming.

A related matter is that the posterior fit shows more warming in the SH than NH, despite that being far from the case in observations. This presumably happens because the simple climate model used to inform their method is fed with differential aerosol forcing in each hemisphere, and greater net forcing in the SH is fairly linearly translated into more warming.

However, the more complex CMIP5 models generally produce more NH warming over the historical period, even those featuring large negative aerosol forcing.

Dave said...

Just wanted to say I greatly enjoyed reading this post and comments. Oops, that sounds spammy, I should prove I'm not a spambot... er... looking forward to seeing more tools for estimating the predictive reliability of IPCC future warming estimates? (But then, aren't we all, I suppose.)

Well, I feel human, anyway.