The Cowtan & Way version of Hadcrut4 uses a kriging technique to extrapolate values into those parts of the world where there are no direct measurements. Originally I had assumed that this fit was guided by the use of UAH satellite tropospheric temperature data and therefore more valid than that say used by GISS. However this is not really true as kriging has been used on all Hadcrut4 gridded data back to 1850, and the difference between the Hybrid (satellite) and kriging only results since 1979 is minimal. The net effect of this is to airbrush away regional variability. This can be seen by comparing the meridional temperature profiles of the original Hadcrut4.5 data with those of Cowtan & Way.
Figure 1. All 117 meridional H4 temperature anomaly profiles from 1990 to 2016. They are coloured blue if the annual global anomalies < -0.2C, Blue,-0.2<grey<0.2, 0.2<yellow<0.4, red > 0.4. Traces are 80% transparent to view them all.
Figure2: Same as above but with Cowtan & Way regional profiles plotted with latitude (meridional)
All profiles have been nicely smoothed out with warming concentrated at the North Pole and the Antarctic. The Antarctic profiles are all sorted so that the warmest also agree with those at the North Pole. What a tidy picture – but is it true or just a by-product of the fit?
Here are the annual averaged anomalies calculated using the above profiles.
Figure 3: Comparison of the Hadcrut4.5 annual temperature anomalies with those calculated using the profiles in Figure 2.
The boost in warming from 2000 onwards is because the high latitude temperature profiles have been untangled by the algorithm. Kriging forces a smoothed ‘spline type’ dependence with latitude all the way to 90N.
“Kriging” AKA Making Stuff Up.
The byproduct of kriging is an error map, is that also available?
What Kriging are you talking about? All the Kriging I have seen are regression/statistical/probability based, NOT spline based. So please show your formulas/references.
Even within Kriging, there are at least 5 major variations (“simpler”, “ordinary”, “co-” etc etc); again, please be specific and show your formulas.
Most Kriging assumes a Gaussian base, and crucially that it is stationary at least in space and sometimes in time … none of which is true for the planet’s climate.
In any case, where are your variograms? As variograms are the essence of Kriging, their absence seems a little odd (it is possible that this is what Hans Erren considers an “error” map).
Some Kriging “de-trends” the data, while others require a knowledge of the mean. So, how can you interpolate something without assuming a mean … but wait a minute, it is the mean that you are actually trying to solve for … hmm sounds a bit circular to me.
Also, all the places where I see Kriging, it is described as an “interpolation” technique, not an “extrapolation” technique. Again, precision in your language is required. If you really do mean extrapolation, then there is a serious problem in how one arrives, say, at the values at the poles.
… once again, to paraphrase PM Disraeli:
“There are three kinds of lies: lies, damned lies, and statistical modelling.”
If you are actually using splines, it may be worse still. How did you arrive at your boundary conditions? What order of spline did you use? If it is > 1, then you may very well have spurious oscillations (which is why “Kriging is visually more plausible” compared to splines, as determined by various authors on the subject … but that of course does not science make).
Much, Kriging was developed for meteorological, or at least “local/regional” fitting, where one can “make up” somewhat more plausible “manual adjustments” … this is decidedly circular for large scale/global application.
The developers and users of Kriging admit that they must make “guesses” and “manipulations” … even within the narrow confines of stationary/Gaussian etc. Making such guessing for weather forecasts can imply a relatively small man introduced error, but not so for large-scale/long-term cases such as climate matters.
However, much worse in the context of climate change in the UN/IPCC et al context, is that once again the methodology is to rely on methods that “break thermodynamics”, guarantee one cannot arrive at a meaningful energy balance, and as previous the highly non-linear nature of radiative flux (at least cubic, but most seem to use/assume quartic) means that this type of massive “smoothing” of the data will cause a substantial error in the heat flux.
Finally, as any satellite image, or wide-range earth based instrumentation shows, there are large scale latitudinal structures in the atmosphere, with very clear cut discontinuities, or near discontinuities with latitude … so, the “continuous in latitude” rhetoric seems to fly directly in the face of well-established and long-standing science.
On shorter time-scales, it is even more complex, and thus completely not amenable to some smooth monotonic latitudinal continuum assumption (or indeed monotonic longitudinally either).
… it is all pretty much meaningless nonsense, since one can guess/fit any values they like, especially when all they need to claim is that the “sky is falling” for +2C/100 years.
… again, please be a proper scientist and do not pretend that the IPCC et al (and Hadley/CRU is certainly climate Gestapo central) and their data have anything to do with proper science.
… btw, why have you again omitted UAH or something similar, which contradicts the manipulated land-based data you are using and is diverging steadily from land-based/manipulated data? It should have been dead easy to overlay the UAH data from 1979 on your Fig 3, and which would have shown that “man made” data/chart to be the non-sense that it is (Kriging or no Kriging) … not sure why you have been for years consistently resisting/excluding unmolested data in favour of molested data.
I would go further than David Walker, it is not merely “guessing”, in the IPCC et al context, it is a special kind of “directed guessing” to create the appearance of “data” to support an ideological agenda, not science.
The short answer is I don’t know what kriging method Cowtan and Way used. I am just comparing their (lat,lon) results with those of Hadcrut4.5
Looking more closely though I see that kriging needs a seed 2-D distribution to calculate covariance for the fit. This can be Gausian as you say, but also Linear, Exponential or even Spherical. So there are some assumptions made about the type of distribution.
This whole exercise followed a tedious discussion at https://andthentheresphysics.wordpress.com/2017/02/12/guest-post-on-baselines-and-buoys/ with some climate zealots .
You’re a jackass! Whine at my site about people not being nice to you and then call them climate zealots. Grow the f**k up!
I wasn’t whining at all. In fact I had some good feedback from for example Nick Stokes, Zeke Hausfather. There are just a few commenters who rubbished what I was saying, simply because they always rubbish anything which may be deemed sceptical. eg. Magma and Willard. Hence the ‘zealot’ label for these guys.
Anyway I thought you were supposed to be having a respectful discussion. ! Try reading what I wrote above.
You wrote this where you complained about what people had said to you. What you failed to acknowledge is that a great deal of what you claimed during the discussion was simply not true. Maybe you should consider that the manner in which people respond to you is driven by the manner in which you chose to engage.
If you want to have a respectful discussion, try not going around calling people climate zealots.
The only thing I said which turned out to be incorrect was that Cowtan & Way did not extrapolate into Antarctica or Africa. Turns out they did despite Hadcrut4 having very few measurements in those regions. Kriging by default will fill any regular grid with values as I am now discovering by doing the same thing as Cowtan & Way.
I would also suggest that you have a slightly annoying habit of always trying to have to have the last word i.e. of scoring the last point. You have the prerogative to do that at your own site – but not here.
Clive,
We all have annoying habits. Do whatever you like on your own site.
Not only is this a rather rose tinted view of what you said (you even suggested that seasonal variations influence anomalies), you probably won’t even recognise the issue with how you’ve framed it (hint: “I’m almost perfect”).
Anyway, have the last word. FWIW, I apologise about the tone of my first comment.
OK – no problem.
P.S. I think seasons do influence warming.
Kriging by default will fill any regular grid with values…
This in general is not the case. Kriging can be pushed to do this but that likely is an error on the part of the practitioner, e.g., extending the search radius beyond the effective range of the semi-variogram.