We’ve already discussed Cowtan and Way’s infilling of HADCRUT4 data in the post On Cowtan and Way (2013) “Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends”. The paper is available here. In that earlier post, I presented the following graph and noted:
If we compare the HADCRUT4 data to the CMIP5 models (historic and RCP6.0) for the period of 1997 to 2012, Figure 1, we can see that the models over-estimate the warming from 65S to 65N (the vast majority of the planet) and underestimate the warming at the poles. Therefore, if the Cowtan and Way (2013) data are increasing the warming in the Arctic, they are creating a greater divergence from the models there, but failing to reduce the differences between the models and data where the models overestimate the warming.
Figure 1
(I changed the above Figure number for this post. It was Figure 9 in the earlier post.)
The Cowtan and Way (2013) data do increase the warming at the poles and exaggerate the failings in the models there, while doing little to explain the hiatus in the non-polar regions, which make up about 90% of the planet.
Note: For those not familiar with the type of graph shown in Figure 1: It illustrates the warming and cooling rates of the HADCRUT4 data, and the average of the CMIP5 (IPCC AR5) climate runs for the period of January 1997 to December 2012…the hiatus period. The vertical axis (y-axis) is scaled in deg C/decade, so we’re showing the warming and cooling rates (that is, the trends). The horizontal axis (x-axis) is scaled in latitude, so the South Pole is to the left at -90 (90S) and North Pole is to the right at 90 (90N). From 1997 to 2012, the HADCRUT4 data show the very slow warming rates (and cooling at some latitudes) extending from the mid-latitudes of the Southern Hemisphere to the mid-latitudes of the Northern Hemisphere. Both poles continue to show warming, however. On the other hand, the models do not show the lack of warming in the non-polar regions during this period. That is, they do not capture the hiatus, the pause, the halt, the cessation of global warming in the non-polar regions. And the models underestimate the warming at the poles, especially in the Arctic, and that means the models do not properly simulate polar amplification. But we already knew the models cannot simulate polar amplification—we discussed and illustrated that failing in the posts here and here.
Back to the Cowtan and Way (2013) data:
In the earlier post, I had not presented warming rates (or lack thereof) for the Cowtan and Way data on a zonal-mean (latitude-average) basis (like Figure 1)because their data is not available on a gridded basis in an easy-to-use format. However, blogger Nick Stokes made the effort to determine those trends for the Cowtan and Way “hybrid” version, for the period 1997 to 2012. (Just what we’re looking for.) See Nick’s post Cowtan and Way trends. Nick was also very kind and he listed the trends in a table. (Thanks, Nick.)
Figure 2 presents the trends of the Cowtan and Way “hybrid” data (courtesy of Nick Stokes) versus the trends of the multi-model mean of the climate models used by the IPCC for their 5th Assessment Report (AR5), for the period of 1997-2012. I’ve also included the HADCRUT4 trends as a reference (dashed lines), because they’re the basis for the Cowtan and Way data. The Cowtan and Way infilling make the models perform worse at the poles, and they had performed very badly with the HADCRUT4 data without the “help” of Cowtan and Way. And the Cowtan and Way infilling did little to eliminate the hiatus in the non-polar regions. Most notably, Cowtan and Way reduced, but did not eliminate, the cooling taking place in the Southern Ocean surrounding Antarctica, a place where sea ice has been expanding in recent decades…and where sea surface temperatures have been cooling.
Figure 2
CLOSING
The Cowtan and Way (2013) revisions to the HADCRUT4 data do nothing to explain the absence of warming that is occurring in the non-polar regions during the hiatus period. Those non-polar regions cover about 90% of the planet and it’s there that climate models cannot explain the slowdown and absence of warming. The Cowtan and Way revisions also exaggerate the warming at the poles which further undermines the current generation of climate models, because the models are unable to explain the observed warming at the poles. That is, the models are still not capable of properly simulating polar amplification.
Those who promote the Cowtan and Way (2013) revisions to the HADCRUT4 data don’t understand where the hiatus is taking place and they don’t understand the model failings at simulating polar amplification—or—they are intentionally being misleading.
SOURCES
The HADCRUT4 data and the climate model outputs are available through the KNMI
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
A global cooling spell involves more meridional / equatorward jets which allows more warm air inflows to the polar regions whilst cold air flowing out from the poles cools the middle latitudes.
A global warming spell involves more zonal / poleward jets which tends to isolate the poles and allows them to become colder whilst the middle latitudes warm up.
The additional warmth at the poles during the cooling scenario allows faster energy loss to space from the poles and the more meridional jets gives more global cloudiness to reduce solar input to the oceans.
Open water in the Arctic allows energy out to space faster and more than offsets the effect of reduced albedo in the Arctic Ocean.
The chart above makes the point very well by showing the slight cooling beneath the mid latitude jet stream tracks in each hemisphere and beneath the ITCZ (which is slightly to the north of the equator) along with the polar warming.
If we were in a global warming spell I would expect to see the opposite, namely, colder poles and slightly warmer areas beneath the jet stream tracks and the ITCZ.
The cynical would say just 2 Sceptical Science insiders/contributors trying their damnedest to put a ‘pause’ rebuttal out there… the cynical amongst us, might say using/misusing peer review to do it?
Ie nobody will look at the paper, just wave around, a rebutall – look this paper says no pause,(ignore Met Office) aren’t those sceptics dumb (oh and its going in the oceans anyway)
After all what do the Met Office know vs 2 guys from Skeptical Science, who don’t like the ‘pause’
Kevin C [Cowtan}
https://www.skepticalscience.com/posts.php?u=3863
Robert Way
https://www.skepticalscience.com/posts.php?u=2360
My conclusion, based on observing a lot of AGW promoters in action is that when a paper is published to explain away why yet another prediction failed, it is actually a demonstration of after the fact arm waving designed to give comfort to the believers.
Stephen Wilde says:
February 22, 2014 at 4:22 am
Stephen, I think your description of jet direction could do with some rework.
Bob, is there an agreed figure for what percentage of the global surface is polar? Your estimate of 10% sounds high.
It strikes me as being the same as rearranging the deck chairs on the Titanic but in this case it is HadCRUT4 temperatures that are being rearranged. The ship is still going to sink.
The problem with Cowtan & Way is that they manage that mathemagical trick of getting a larger trend in their result data set than is present in either of their two source data sets. That alone should cause ‘pause for thought’.
Barry Woods is absolutely right.
It’s time these activist-inspired “peer reviewed” papers from the Cook/Nuccitelli “et al” Skeptical Science crowd were called out for what they are – manufactured pseudo-science, conceived and designed only to plug gaps in the climate alarmism narrative.
We saw from the leaked SkS private forum how these people connived behind the scenes to distort proper scientific climate debate by faking up their 97% “paper” which had no statistical or scientific merit and was simply a elaborate exercise in dressing up their own prejudices.
It’s worth noting that both Kevin Cowtan and Robert Way were active & vocal plotters on the “secret” SkS forum well before they decided to re-write the HADCRUT historical temperature record and attempt to get rid of the embarrassing “pause”.
Cowtan is a crystallographer who describes his interest in climate science as a “hobby” and Way is still (AFAIK) a PhD student. Whether you accept Bob’s interpretation of their results here or not – is it not bizarre that senior climate scientists like Schmidt & Rahmstorf at Real Climate, and various others, are now quoting “Cowtan & Way” as if it were part of the accepted historical temperature record – rather than just an interesting piece of mathematical extrapolation by a couple of amateurs.
The fact is – every piece of pseudo-science that has been dreamt up by the SkS crowd has been conceived to produce a pre-determined outcome – which could be exploited to advance a cause.
That is the exact opposite of the spirit of scientific enquiry.
Hadcrut4 added 628 new stations compared to Hadcrut3. These were mostly in Arctic regions and this was the main reason that global average temperatures increased slightly after 2000. Meanwhile. 176 stations in the Americas were dropped and Africa still has hardly any coverage at all. The new stations are shown in red while the dropped stations are shown in blue in this map.
Now Cowan and Way want to add even more stations in arctic regions to get the average temperature anomaly up even higher ! To be consistent they should request that the missing grid points in Africa get sampled.
One other point to remember in all this is that CRU use a 5×5 degree grid. The area of each grid cell increases dramatically the further north you go so if you walk in a circle around the north or south pole you will pass through 72 grid cells ! It pays to add stations near the poles especially when you know that the temperature rise is greater.
Bloke down the pub says:
February 22, 2014 at 4:36 am
“Stephen, I think your description of jet direction could do with some rework.”
Such as ?
Clive Best says:
February 22, 2014 at 5:29 am
Clive: shouldn’t that read: “The area of each grid cell decreases dramatically the further north you go …”
Bloke down the pub says: “Bob, is there an agreed figure for what percentage of the global surface is polar? Your estimate of 10% sounds high.”
The latitudes of 65N-90N (or 90S-65S) represent about 5% of the surface of the globe. And as a reference, the Arctic Ocean covers a surface area of about 4.3% of the surface of the global oceans, according to NOAA webpage here…
http://www.ngdc.noaa.gov/mgg/global/etopo1_ocean_volumes.html
Regards
Stephen Wilde says:
February 22, 2014 at 4:22 am
Meridional flow (or amplified flow) has big ripples in the track of the jet stream. What goes north has to come south. Like this year – the southern lobe over the north central US is balanced by the northern lobe that has brought such warmth to western Alaska and kept rain away from California. There’s equatorward and poleward air flow.
Zonal flow is more east-west and brings relatively boring weather. I suppose it lets the arctic get cold since there’s less exchange of air mass between that and the mid latitudes.
Bloke down the pub says:
February 22, 2014 at 4:36 am
Bob, is there an agreed figure for what percentage of the global surface is polar? Your estimate of 10% sounds high.
Working out the area using the formula for a spherical cap……. Take the ACOS of 0.9. (100% -10%). Gives. 25.8 degrees. 90-25.8 =64.2deg lat as the 10% of hemisphere (5% of globe). Pretty close to the 65 degrees being the quoted figure.
@Gilbert K. Arnold
Oops – Indeed it should ! The area of each grid cell decreases dramatically the further north/south you go.
More on sampling biases here:
CRU’s Arctic Fix and
A Study of Hadley-CRU weather station data
IMHO, Hadcrut3 temps should have been used. Just because most models have been parametrized with HadCrut3.
I know I am alone in this battle, but I still think I am right. If you move the termometers, the analysis is not valid.Lets compare apples to apples.
Ok, I read Clive Best’s post and I am happy to see that I am not alone. 🙂
huh cowtan and way dont warm the arctic. its warming fine by itself.
1. hadcrut is biased cold in arctic.
2. Cowtan and way estimate the arctic.
3. That estimate can be tested.
4. it passed
From the whistle blown SkS author/contributor forum:
extract: General Chat – 12/02/04 – Can Someone help me…
Cowtan:
“What are you after? If you want to do a HADCRUT3 takedown, that’s what I’m working on at the moment (as a paper, but got some nice analogies for popular usage). It’s arguably better than the BEST argument because it’s global, although I’m not ready for publication yet! But I can give a trivial argument by analogy which shows that HADCRUT3 must be wrong, and give a corrected version in which 2010 beats 1998.”
now the paper may be good;
but I have a problem with the attitude here – ‘takedown’ , for popular ‘usage’ which makes me question everything they do, something to be waved around, for the cause, not to advance science?
Except that the MultiModel Ensemble Mean of the CMIP5 models is a Meaningless quantity, indefensible by any of the laws of statistics.
One could fairly compare the models in CMIP5 one at a time to the temperature distribution of HADCRUT4 and fail them one at a time, but discussing the collective “mean behavior” of a bunch of not-even-independent models, all of which contribute completely different numbers of perturbed parameter ensemble runs to the single model mean, which is then averaged into the MME mean with an equal weight, when many if not most of those models would independently completely fail simple hypothesis tests such as this one against the data makes no sense, and using the MME mean in comparisons simply continues to give credence to a complex statistical fraud.
If one generated (say) 36 different versions of the graphs above, one per model, one would, I’m sure, find that some models are so far away from the data as to be completely absurd. Yet they are still in that good old MME mean. Who knows, one or two models in CMIP5 could be doing pretty well against the data, but are they given greater credence or weight? Of course not. We don’t even know which ones they are — if we did, we’d use them and throw the rest away, and most of the predicted warming would disappear alone with the discarded part.
rgb
To Clive:
Yes, one of the greatest sources of idiocy in climate science is the use of a lat/long grid both in CRU and in GCM’s. Why both with icosahedral tesselations that don’t have a singular Jacobean? It’s so much easier to pretend that the world can be covered with a rectangular grid, even at the poles.
I often wonder how much the computations would change if anyone built climate models on top of an automagically rescalable icosahedral tesselation and did proper adaptive quadrature and interpolation on an unbiased adaptive grid. From my experience (which is fairly considerable) trying to integrate in spherical polar coordinates the other way, it is virtually impossible to get decent accuracy in any adaptive algorithm on a pseudorectangular grid — one simply cannot get the weighting right — one oversamples the hell out of the poles to get adequate sampling at the equator.
rgb
foxgoose says:
February 22, 2014 at 5:24 am
“is it not bizarre that senior climate scientists like Schmidt & Rahmstorf at Real Climate, and various others, are now quoting “Cowtan & Way” as if it were part of the accepted historical temperature record – rather than just an interesting piece of mathematical extrapolation by a couple of amateurs.”
Rahmstorff et al shurely noticed that Cowtan&Way makes the problems for the models worse. But they bank on the media never telling the populace. And the media never will.
Barry
Good find – that nails it precisely in Kevin Cowtan’s own words.
If an individual, scientist or not, embarks on a piece of work with the main objective being to “takedown” some other work and achieve “popular usage” – it ceases to be science.
Feynman understood this perfectly when he said :-
If you’ve made up your mind
to test a theory, or you want to explain some idea, you should
always decide to publish it whichever way it comes out. If we only
publish results of a certain kind, we can make the argument look
good. We must publish both kinds of results.
Why can’t the SkS “Crusher Crew” understand what science is?
Clive Best says:
February 22, 2014 at 5:29 am
The use of a 5×5 degree gridding may also have a material effect on basic calculations in the Cowtan and Way paper that the authors may not be aware of. Kriging involves the use of a covariance function which measures the degree of relationship of the temperatures of grid cells a given distance apart. The covariance function is estimated from the data as described in the supplementary materials for the paper:
Assuming that great circle distances between the centers of cells were used, the lower distance bins are dominated by differences from high latitude cells while the relative influence of low latitude cells is manifest in the bins for greater distances. What effect this might have on the estimation of the covariance function parameters is not immediately clear, but it does reasonably indicate the need for further analysis by the authors.
Steven Mosher says:
February 22, 2014 at 7:16 am
“2. Cowtan and way estimate the arctic.”
C&W managed to get a result set for the Arctic to have warmed faster than the warming patterns in either of their two input data sets.
How do you support that conclusion?
Why do you believe that a short geographic and time period alignment is possible whereas a longer time and/or geographical between Satellite and Thermometer has so far been resisted by everybody?
Here’s another interesting re-analysis.
http://judithcurry.com/2013/11/28/is-earth-in-energy-deficit/