Anthony Watts' protest about Cowtan, Way and the Arctic

Anthony Watts has found something new to protest (archived here).  It's an interesting paper by Kevin Cowtan and Robert G Way published in the Quarterly Journal of the Royal Meteorological Society.

The Arctic is warming at about eight times the pace of the rest of the planet

The paper has been described and discussed in various places.  What the researchers did was take a novel approach to work out recent temperature changes in parts of the world where there are gaps in the data.  Here is a report from ScienceDaily.com:

An interdisciplinary team of researchers say they have found 'missing heat' in the climate system, casting doubt on suggestions that global warming has slowed or stopped over the past decade.

Observational data on which climate records are based cover only 84 per cent of the planet -- with Polar regions and parts of Africa largely excluded.

Now Dr Kevin Cowtan, a computational scientist at the University of York, and Robert Way, a cryosphere specialist and PhD student at the University of Ottawa, have reconstructed the 'missing' global temperatures using a combination of observations from satellites and surface data from weather stations and ships on the peripheries of the unsampled regions.

The new research published in the Quarterly Journal of the Royal Meteorological Society shows that the Arctic is warming at about eight times the pace of the rest of the planet. Previous studies by the UK Met Office based on the HadCRUT4 dataset, which only covers about five-sixths of the globe, suggest that global warming has slowed substantially since 1997. The new research suggests, however, that the addition of the 'missing' data indicates that the rate of warming since 1997 has been two and a half times greater than shown in the Met Office studies. Evidence for the rapid warming of the Arctic includes observations from high latitude weather stations, radiosonde and satellite observations of temperatures in the lower atmosphere and reanalysis of historical data.

A member of the Department of Chemistry at York, Dr Cowtan, whose speciality is crystallography, carried out the research in his spare time. This is his first climate paper.

He says: "There's a perception that global warming has stopped but, in fact, our data suggests otherwise. But the reality is that 16 years is too short a period to draw a reliable conclusion. We find only weak evidence of any change in the rate of global warming."

Robert Way adds: "Changes in Arctic sea ice and glaciers over the past decade clearly support the results of our study. By producing a truly global temperature record, we aim to better understand the drivers of recent climate change."

More about the research

There are some very readable discussions about the paper at:

• realclimate.org by Stefan Rahmstorf
• skepticalscience.com by the authors, Kevin Cowtan and Robert Way, with Dana Nuccitelli
• Variable Variability by Victor Venema - with lots more links
• Arctic Sea Ice Blog by Neven

The authors used a statistical technique known as kriging to interpolate data from neighbouring sites that have temperature observations to determine the temperature in between.  Here is how Wikipedia describes kriging:

The basic idea of kriging is to predict the value of a function at a given point by computing a weighted average of the known values of the function in the neighborhood of the point. The method is mathematically closely related to regression analysis. Both theories derive a best linear unbiased estimator, based on assumptions on covariances, make use of Gauss-Markov theorem to prove independence of the estimate and error, and make use of very similar formulae. They are nevertheless, useful in different frameworks: kriging is made for estimation of a single realization of a random field, while regression models are based on multiple observations of a multivariate dataset.

Kriging was also used by the Berkeley Earth Surface Temperature team. What was novel about this research was that the authors also used a hybrid method, combining satellite data and surface data.  They found by testing that the hybrid method gave the most accurate results over land and sea ice, whereas kriging was best over open oceans.

It's all a bit too much for Anthony Watts and his band of fake skeptics

Anthony would discount any reasonable discussion of the paper as being too sciency.  He managed to find a denier slant proffered by Judith Curry (archived here). And he did find some words of his own to help get his crew fired up (archived here):

Breathless interpreters of Cowtan & Way claim that by doing the same with satellite data instead of tortured surface data, Voilà “the pause” disappears.

It looks as if Anthony didn't read the paper or any discussion of it.  If he had, he would have seen that the authors used both satellite and surface data.  I guess he didn't watch the video he posted, either.  Here it is:

Judith Curry shows her ignorance - and she has researched the Arctic

Over at Judith Curry's blog (archived here), Kevin Cowtan, one of the authors, explains, rather nicely and politely, that (my hyperlinks):

Dear Dr Curry
Thank you for you comments. We indeed hope that one of the results of our paper will be to stimulate a vigorous discussion in this area.

With respect to kriging across land ocean boundaries, we note that this is a problem in the paper. Can I draw your attention to our update memo [Sou: I think Kevin is referring to this] in which we test separate reconstruction of the land and ocean data before blending, which is in our view a better approach. To do this properly would require access to the HadCRUT4 land ensemble which is not currently distributed, but with the CRUTEM4 data (which lacks some corrections) the results of blending pre- or post-reconstruction is almost indistinguishable, even under different ice-coverage assumptions. (There is no reason why this must be the case, it is a result of the distribution of the unobserved regions). Dynamically changing ice is more difficult, and you can’t do it with anomalies as you don’t know what kind of bias you introduce when changing a cell from land to ocean, so we’ll have to leave that problem to the BEST team.

Most interesting is the issue of the UAH data over Antarctica. We’ve recently been looking at this with respect to both Vostok, and the Bromwich 2012 Byrd reconstruction. Byrd particularly interesting – it sits on a cell boundary and is remarkably well modelled by the cell to the north in the hybrid reconstruction. The cell to the south models the year-to-year variations, but not the long term trend. We’ve made some preliminary analysis of what is going on based on differencing North-South transects in the UAH data. Some regions show no significant changes, whereas others show large changes in either direction around 2000. I hope to write this up as another update, and maybe Dr Christie will be able to shed more light on the issue, although I’m afraid everything takes a long time when you’re doing it in your spare time.

So it may be that kriging is a better approach for Antartica, especially with remediated data from some of the isolated stations – Byrd is critical here, and I want to do some detailed comparisons with BEST too. Against that, the holdout tests actually favour the hybrid approach for most of the existing station locations, including the SP.

Having said all of that, the difference between the hybrid and kriging reconstructions of Antarctica is only really significant around 1998, so it doesn’t greatly affect our conclusions. And the Arctic is sufficiently small that the two reconstructions are very similar. Most of the Arctic coverage bias also arises in the NH winter, when the Rigor result is most relevant.

If I may appeal to your own expertise, there would seem to be a parallel between our results and those of Cohen et al 2012 (doi:10.1088/1748-9326/7/1/014007). Do you think there is a plausible connection? November 13, 2013 at 4:23 pm 

Robert Way added some comments of his own in response to Judith Curry's complaints (my bold, with Judith Curry's comments in italics ).

First, Kriging. Kriging across land/ocean/sea ice boundaries makes no physical sense. While the paper cites Rigor et al. (2000) that shows ‘some’ correlation in winter between land and sea ice temps at up to 1000 km, I would expect no correlation in other seasons.

Response [1] Actually in the paper we show through rigorous cross-validation tests (see Table 1; Table 2; Figure 3) that kriging is an effective approach for estimating temperatures, even across boundaries. However the hybrid approach performs better than any other method at reconstructing high latitude temperatures (see Figure 3 – cross validation) even at distances of 1650 km). In the case of sea ice this hypothesis has been tested (see Figure 4) where it is shown that kriging from land regions outperforms kriging from ocean cells.

Second, UAH satellite analyses. Not useful at high latitudes in the presence of temperature inversions and not useful over sea ice (which has a very complex spatially varying microwave emission signature). Hopefully John Christy will chime in on this.

Response [2] As indicated in the response to the 1st comment – we have tested the methodology adopted in this study against both held-out observations and against grounded/floating buoys in the Arctic ocean, often located on sea ice. The results of our study indicate that the performance of the hybrid method is reasonable over ice (Figure 4; Figure S5).

We also provide an attempt at showing the impacts of changing sea ice conditions on the reconstruction. Although not available in the supplemental information we have also tested the method in Antarctic against the reconciled Byrd station located in one of the most icebound, isolated places on the planet. The results of this test show very reasonable performance with the hybrid method.

Third, re reanalyses in the Arctic. See Fig 1 from this paper [Sou: the paper Judith cites is one of hers published in 2002], which gives you a sense of the magnitude of grid point errors for one point over an annual cycle. Some potential utility here, but reanalyses are not useful for trends owing to temporal inhomogeneities in the datasets that are assimilated.

Response [3] Since the paper in question was published there have been significant advances in reanalysis methods. In particular, 4-D methods such as those employed by ERA-Interim have shown to be much more reliable in the Arctic and Antarctic. There are a series of papers by James Screen at Exeter which delves into many of these issues and examines the performance of reanalysis products in both the Arctic and Antarctic. I would suggest that Dr. Curry take a bit of time to have a look at the results of some of these studies. That being said the paper does not use reanalysis to infill temperatures, nor do we use it with the kriging, reanalysis is simply presented as an additional source of evidence in additional to satellites, radiosondes and isolated weather stations which show that the Arctic is rapidly warming. Physical evidence is also available in the form of sea ice reduction and glacier changes as well as melt records from high Arctic ice caps. There is a wealth of literature supporting the conclusions that the Arctic is warming rapidly and this relationship (Arctic Amplification) is clear in the paleorecords.  November 13, 2013 at 4:31 pm

Judith, having mispelt Robert's family name throughout (Wray instead of Way) posts a reply immediately under Robert's but for some reason decides to call him James and then, for no reason at all, says she doesn't believe him.  (Yeah, I went 'huh?' too).  Judith also said she doesn't know of any reanalysis studies done by James Screen despite her having done some recent research on the general topic.

James, thanks for stopping by and engaging here. I agree that there is evidence of warming in the Arctic, however, I remain unconvinced that your methods are verified in any meaningful way for surface temperatures of open water and sea ice in the Arctic Ocean. I see no reference to papers by James Screen in your paper, I don’t know what papers you are referring to. I have recently done a comprehensive literature survey regarding in situ surface temperature and surface flux measurements in the Arctic Ocean (for a grant proposal). I have not seen any recent studies evaluating reanalyses using these data sets. November 13, 2013 at 6:56 pm

So much for Judith's research skills! I would have said that a ten second Google Scholar search would have saved Judith Curry a certain amount of embarrassment, except that she has never shown any sign of being embarrassed by her bloopers.

From the WUWT comments

I'll leave interested readers to do any further reading with the links I've provided above.  Let's just see what the various members of the WUWT brigade have to say (archived here).

Judge starts the ball rolling and maybe got up Anthony's nose a bit by pointing out that he mimicked Judith Curry's mis-spelling of Robert Way's name (which Anthony later corrected) and says:

November 14, 2013 at 9:32 am
Way not Wray – “Robert G. Way”

omnologos points out that it can't be warming everywhere.  Someone is probably wrong.  Thing is, I don't think the Cowtan and Way necessarily claims to account for all the difference between the models and the other global temperature data sets, but I could be wrong:

November 14, 2013 at 9:35 am
Just like Kloor has shown it’s ridiculous now for alarmists to do their regular ambulance chasing at every hurricane or drought, likewise scientists should stop claiming to have found the missing heat in the most convenient of places, namely where nobody can get much reasonable data from (it was the depths of the ocean, not it’s the most remote of the North Pole).

There are also many other problems with this pausebuster. Have the scientists involved deliberately misled the IPCC by telling nobody about what was incoming two months later? If they are right, isn’t Dear Kev wrong about the oceans?

If they are right, then people claiming that there was a pause (based on non-infilled data)were right, and Dana and SkS wrong in dismissing the pause.

Also if the North Pole puts the trend back to expected values, this means the recent warming is becoming more and more northern-polar than global.

Furthermore this would be yet another AGW miracle, with values magically going back to be exactly as expected.
Etc etc.

Kev-in-Uk says he can't stomach science:

November 14, 2013 at 9:36 am
Here we go again – more data from the ‘middle of nowhere’, literally! I dunno whether I can stomach to watch the method video. Someone tell me if it makes sense, and if I can use the same methodology to magic money into my bank account, as there is none there at present!

Paul Homewood, recently of super-typhoon infamy, hasn't bothered reading anything about the paper but agrees that it's been getting very warm in the Arctic lately.  He says:

November 14, 2013 at 9:42 am
Even GISS figures with their 1200km smoothing (right or wrong) show the same pause as the other sets, so they cannot argue the poles are being ignored.
Indeed, take out the poles from the GISS dataset, and I would imagine you would end up with a cooling trend.

tWISTERdATA says:

November 14, 2013 at 9:44 am
Lots of sciency talk. Lets just simplify: “We made stuff up.”

Then tWISTERdATA posting again, this time as tornadomark, decided he could improve on his previous comment and says:

November 14, 2013 at 9:45 am
Lots of Sciencey talk. Here’s the bottom line: “We made stuff up to support the cause.”

Salvatore Del Prete makes believe he knows something about science and says:

November 14, 2013 at 9:48 am
Filling in data is worse then doing nothing.

Eliza says she doesn't know why WUWT bothers with sciency stuff (maybe it should stick to paranoid conspiracy theories) and says:

November 14, 2013 at 10:15 am
I dont’ know why such drivel (The paper) even gets mentioned here.

Ian W just knows all those scientists are wrong and so is the Royal Society of Meteorology and all the reviewers.  He lets everyone know why all the scientists are wrong.  A change in temperature tells you nothing, he cries, and proceeds to build a heated strawman (excerpt):

November 14, 2013 at 1:05 pm
Water does not scatter the infrared it absorbs it. Thus it raises the heat content of the air without raising its temperature. That is the enthalpy of the air (its heat capacity) increases as the humidity increases.

Mosher and all the other non-engineers are averaging temperature are thus averaging the wrong metric. They should be measuring heat content in kilojoules per kilogram. After all isn’t the ‘global warming hypothesis’ about trapping heat – you can’t trap temperature Steven can you?
So as davidmhoffer says it requires almost no heat at all to raise the temperature of a volume of dry arctic air at minus 30C by one degree, but a large amount of energy is required to raise the temperature of a similar volume of 90% humidity equatorial Pacific air by one degree because the humid air enthalpy is so much higher. Yet you ‘highly trained climate scientists’ average the temperature of these volumes of air to measure heat???

For that reason alone the Cowtan and Way should have been thrown out at peer review.

Gunga Din lightens the mood and says:

November 14, 2013 at 1:13 pm
So………….the missing heat is hiding in the cold?

 Steven Mosher and Nick Stokes chime in with some science stuff, much to the disgust of the WUWT-ers.  For example, Nick Stokes writes:

November 14, 2013 at 1:32 pm

“At first glance, this seems an admirable and reasonable goal, but one should always be wary of trying to create data where there is none, something we learned about in Steig et al’s discredited paper on the supposed Antarctic warming.”

Steig’s paper was not discredited, and particularly not his method of fitting satellite-derived EOFs to station data to infill. His critics Ryan et al used the same method. They used more EOFs, which gave an improvement.

Whenever you calculate a space average from sampled data, there is an implied assumption that the samples are representative of data in between. That’s not just climate science, it’s for any continuum analysis. The average is justified by interpolation. Any form of rational interpolation is better than leaving areas out.

Even a couple of fake sceptics concede the point Nick Stokes made. Mark Bofill says:

November 14, 2013 at 1:45 pm
I’m often critical of Nick but I’ve got to say I think he’s got a point here. I haven’t read the paper (with my limited grasp of the science it wouldn’t make all that much difference if I did) but from what I’ve gathered so far from the discussions about it I’m thinking this paper is probably pretty solid. Perfect? No, but it’s probably a step in the right direction.

Louis doesn't understand how scientific knowledge grows and says:

November 14, 2013 at 2:02 pm
If climate scientists accept this paper, won’t they be admitting that they have been calculating the average global temperature wrong all this time? If they could make such a big mistake in how they estimated polar temperatures, doesn’t that open the door to the possibility of other major mistakes in their methodology? Do they really want to admit that the science isn’t so settled after all?

The paranoid arrive a bit late to the WUWT playground.  Dolphinhead says:

November 14, 2013 at 2:05 pm
I’m guessing this is another of those papers that will be used by politicians as if it was written on tablets of stone and will be quoted as the reason why we must continue to de-industrialise the West.

Philip Peake is sad that the scientists don't take heed of all the fake experts at WUWT and says:

November 14, 2013 at 2:12 pm
The sad thing here appears to be using data which is not understood.
The even sadder thing, is that a (supposedly) respected journal and its expert reviewers didn’t catch that.

Janice Moore gets up a head of steam and as one of seven posts in a row, all of them meaningless, shouts (excerpt):

November 14, 2013 at 7:23 pm:
...This stupid paper (given that it reveals ANY truth) does at least indicate that the IPCC models which projected Arctic cooling are even WORSE than we already knew that they were ……….which isn’t saying much …………………. which makes this paper just a piece of junk.  As someone above aptly said: THEY ARE JUST MAKING IT UP AS THEY GO. Pitiful.

That about covers it, unless you're a glutton for punishment and want to read more WUWT comments - archived here.

---

Kevin Cowtan and Robert G. Way (2013), Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends, Quarterly Journal of the Royal Meteorological Society, http://dx.doi.org/10.1002/qj.2297