Climate modeling EPIC FAIL – Spencer: 'the day of reckoning has arrived'

You would think there’d be joy in the climate science camp that they were wrong about runaway warming from human-produced CO2, and relieved that we can continue bringing cheap, plentiful electricity to the areas of the world that desperately need it.
That this is not the case, and the fact that they write things like “sadly, no warming,” tells me that they are way too invested in their ideas and too much in love with the prospect of being an honored Cassandra. Are they not self-aware enough to see this?

Here is a worthy comment from Dr. Spencers post.

Michael Palmer says:
June 6, 2013 at 11:01 AM
“People like you and John Christy have set an example of high scientific and ethical standards throughout this debacle, and you will still be remembered when those currently still riding high will have long been forgotten.”

I do hope Dr. Death Trains Hansen, Dr. Phil Depressed Jones and John Cook The Books are all paying attention. If sceptics are right your names will for ever be sullied for having participated in the biggest scientific swindle the world has ever know. Billions of Dollars have been diverted away from needy causes to perpetuate this joke.

It seems that the data presented here need better specification.
Please specify
1) the mean altitude of the observations (in meters, or pressure value)
2) the mean altitude of the models outputs (same as before)
And maybe, if at hand, plot the same for different altitudes, and different regions, please.
Regards

The underlying assumption in all climate models is that you can take a chaotic process (weather) and average this over time and the Law of Large Numbers will operate to deliver a non-chaotic results.
This assumption has never been mathematically shown to be true. The Law of Large Numbers requires some very specific statistical properties, which are not present in chaotic systems.
Where is the basic, fundamental math showing that the average of a chaos system is anything but chaos? In all those hundreds of millions spent on models, who has taken the time to show that the underlying math is valid?

ferd berple says:
June 7, 2013 at 6:42 am
The underlying assumption in all climate models is that you can take a chaotic process (weather) and average this over time and the Law of Large Numbers will operate to deliver a non-chaotic results.
——
Not at all. Chaotic or not, every delimited system is bound to the Energy Conservation rule, and this rule is not chaotic.
Chaos acts on the energy distribution into the system, non on the overall energy balance of the system.

Jimbo says:
June 7, 2013 at 4:20 am
Dr. Phil Jones – CRU emails – 7th May, 2009
‘Bottom line: the ‘no upward trend’ has to continue for a total of 15 years before we get worried.’
It has been a further 4 years of the ‘no upward trend’ so what I want to know is this: are you worried Dr. Phil?
Is it a coincidence that around the time when HadCRUT3 reached 15 years of no warming, HadCRUT4 came along? HadCRUT4 shows a different picture than HadCRUT3. Presently, HadCRUT3 has 16 years and 1 month of no warming but HadCRUT4 has 12 years and 6 months of no warming.

Phitio (June 7, 2013 at 7:49 am) responded to ferd berple’s (June 7, 2013 at 6:42 am) comments on LLN:
“Not at all. Chaotic or not, every delimited system is bound to the Energy Conservation rule, and this rule is not chaotic.
Chaos acts on the energy distribution into the system, non on the overall energy balance of the system.”
Someone who might actually understand Tomas Milanovic’s cautions. Remarkable.
However…
Caution: Haphazardly choosing first moment aggregation criteria without due attention to systematic spatiotemporal volatility (second moment) clustering can be fatally misleading.
Topologically the spatiotemporal circulatory loop lattice is balanced by a global multi-axial fluid differential. It needn’t worry readers if that sounds complicated, because it’s just a multi-axial fluid generalization of something profoundly simple — e.g. as occurs in the solid mono-axial case, such as with automobile axles:
http://en.wikipedia.org/wiki/Differential_%28mechanical_device%29
It’s true that LLN alone isn’t enough. An additional constraint is needed. Use the Law of Conservation of Angular Momentum, Earth Orientation Parameter data, and visual hints from the following landmark article that illustrates far more than it addresses in the text:
Dickey, J.O.; & Keppenne, C.L. (1997). Interannual length-of-day variations and the ENSO phenomenon: insights via singular spectral analysis.
http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/22759/1/97-1286.pdf
Tip: See particularly the attractor illustrated in Figure 3a & 3b. Then look for 2300 year modulation of semi-annual and 1500 year modulation of annual — (more details including connections with Steinhilber+ TSI if/when time/resources ever permit…)

This assumption has never been mathematically shown to be true. The Law of Large Numbers requires some very specific statistical properties, which are not present in chaotic systems.
Um, I don’t usually disagree with you but this time I must. A distribution with compact support is a sufficient condition for the central limit theorem, whether or not the underlying dynamics is chaotic. Indeed, chaos would typically give a system a better chance in the long run to ergodically sample the bounds of the space. So I would have to say that this statement is specifically incorrect and/or irrelevant for climate.
What you mean to say(and that I fully agree with and indeed will expound upon below) is that the climate varies with significant dynamics occuring with time scales ranging from as short as days (emergent from the chaos that is “weather”) out to very long indeed, and many of those scales (and the particular dynamics that is important on them) are not well known. A few timescales we can identify that are likely relevant:
a) Days/weeks/months — persistence time of global climate features such as high and low pressure centers, tropical storms, transient variations in the upper atmosphere circulation, humidity and cloud cover.
b) Years. There is obviously a powerful annual signal, one that is driven by interference between orbital eccentricity and axial tilt. Some features of this make little sense — orbital eccentricity alone produces several orders of magnitude greater variation of insolation than CO_2 or the actual variation of the sun, and yet the Earth’s mean temperature variation has the opposite phase. This suggests that the specific distribution of land surface versus sea surface and feedbacks from water vapor and clouds may be far more important than anything else.
c) Decades. There are a number of named decadal oscillations. Some quite obviously have a profound effect on the global climate, notably ENSO. Although they are chaotic, coupled, and not sharply periodic, they do appear to have a very weak characteristic timescale. The Sun strongly varies on a decadal scale.
d) Centuries. The timescales of various oceanic processes are decades to centuries. The timescale of significant alteration of glaciation or major ice pack is usually centuries. The Sun’s solar cycle itself waxes and wanes in strength on a timescale that appears to be centuries (although again, it is not sharply periodic, it is chaotic and irregular).
e) Millenia. Although it is not well-understood, there again appear to be climate variations with timescales of order 1000 years (e.g. RWP to MWP to modern WP). But this is by no means uniformly extensible (as one expects for a chaotic system), and may have no particular cause beyond coincidence in underlying shorter period phenomena in a nonlinear chaotic feedback system.
f) Ten(s) of thousands of years. Orbital and axial tilt variations occur on these timescales, and at least some climate periodicity is weakly and not terribly consistently slaved to this, at least in some parts of the climate record (e.g. the general Pliestocene has an apparent glaciation/interglacial variation that itself varies from ~20ky to ~26 ky to 40 ky to 100 ky). But why the climate doesn’t slave faithfully to this fairly predictable signal is not well understood at all.
g) A hundred thousand years, as noted above, for at least the last 5 glacial/interglacial eras, appears to be important — 90 ky of glaciation, 10 ky of interglacial (roughly) but very erratic. Note well that the Holocene is over 10 ky old already, and this interglacial might well be nearing its nominal/normal end. The factors that drive glaciation are obviously powerful (and not well understood) and personally I don’t think anybody has a clue as to whether or not increased GHG’s will be enough to oppose them if glaciation got a positive feedback foothold. Naturally, there are plenty of people who will tell you that they are certain one way or the other. All I ask of them is that they hindcast the Pliestocene using the arguments that are the basis of their certainty, and then I’ll be certain too.
h) Millions to tens of millions of years. Glacial epochs tend to last anywhere from a few million years through tens of millions of years. Interglacial warm periods tend to last tens of millions of years to as long as (order of a) hundred million of years. However, this is very crude. But I still think that glimpses of 65 million years of climate change:
http://en.wikipedia.org/wiki/File:65_Myr_Climate_Change.png
or 500 million years of climate change:
http://en.wikipedia.org/wiki/File:Phanerozoic_Climate_Change.png
are in order. These figures put the climate nicely in perspective. First of all, the climate has been as cold as it is right now only four times over 500 million years, where we are tied with the coldest temperature visible in the entire 500 million year O_18 record. That’s something you don’t hear much about, but we are in an interglacial epoch of the Pliestocene glaciation. If we plot the last five million years (just to get the rest of the perspective):
http://commons.wikimedia.org/wiki/File:Five_Myr_Climate_Change.png
we see that we are pretty much dead set normal as far as interglacial temperatures are concerned, and still below the peaks of three of the last four interglacials (let alone the stable warm temperatures from before the Pliestocene). As warm as it is now, it is still colder than it has been for 62.5 million years of the last 65 million years.
These figures also reveal something of the chaotic, quasiperiodic, and in many cases utterly unpredicable nature of climate changes over truly geological timescales. Note also that changes on timescales of decades — which is all that we have accurate data on, and entirely where all of the “emergency” is supposedly occurring — are completely invisible on the timescales of any of the graphs above. The temperatures in any given chunk of that record are almost certainly both warmer than the warm peaks and colder than the cold troughs because they are averaged out over decades to centuries to even longer in the inference process. Also, it would be lovely to know the expected error estimates on these temperatures — impossible to tell of course as nobody will ever produce a plot like this with an error estimate in climate science because if they did what kind of story could anybody tell? I like to imagine it as varying from several tenths of a degree in the satellite data in the modern era through 0.5 to 1.5 degrees (K, of course) in the thermometric era and staying somewhere between 1.5 to 5 K (maybe?) over the rest of it. But I’m a cynic…
So can one produce a meaningful “mean temperature” for the globe? Sure. Lots of them. Just bear in mind that the mean temperature for the globe is not constant in time on any meaningful timescale, with or without anthropogenic contributions to that variation. It is not constant on a timescale of days, weeks, months, years, decades, centuries, millenia, ten thousand years, a hundred thousand years, a million years, ten million years, a hundred million years or a billion years. It has never been constant.
Furthermore, it has no meaningful linear trend. This is the really crazy thing — Roy and I completely agree on that one. Indeed, fitting any simple polynomial form to the global temperature data on timescales longer than decades makes no sense at all — the fit doesn’t become more meaningful if one includes “acceleration” (quadratic) or higher order terms. (Note: timescales less than decades one can often average over enough of the longer term stuff that one guesstimate a few years out on either side from a knowledge of the present — there is an autocorrelation time for the climate order of a decade so at least that much is safe enough.) It is apparently equally silly to try fourier analysis of the temperature series — as one expects in a chaotic system, one can find stretches where there is a very lovely signal (see the 5 million year record above) but the minute you try to take that lovely, sharp peak and use it to predict the variation before or after the stretch you get diddly, and nobody knows why the periodicity turns on, works for a while, then turns off, certainly not well enough to predict it in the future (and probably not well enough to honestly hindcast it in the past).
All the climate models are saying is: All things being equal, and subject to the correctness of our assumptions concerning feedback and sensitivity, the climate will do thus and such over the next century. This ignores the stark fact that as far as the climate is concerned, all things are never equal! The climate was steadily, systematically changing from the MWP to the LIA, from the LIA to the Dalton minimum, from the Dalton minimum to 1945, from 1945 to 1975, from 1975 to 1997. If anything, the last 16 years of remarkably stable temperatures are themselves a bit of an anomaly — there are damn few 16 year stretches of nearly uniform temperatures visible in the last 140 years of thermometric data!
This is why the entire field of climate science focuses on the temperature anomaly. The very term is, of course, an lying oxymoron. In order to have an anomaly, one has to have a normal, as an anomaly is quite literally:
From WordNet (r) 3.0 (2006) [wn]:
anomaly
n 1: deviation from the normal or common order or form or rule
[syn: {anomaly}, {anomalousness}]
Note the subtle con, straight out of the book How to Lie with Statistics. By presenting results as an anomaly instead of on the actual relevant scale (degrees absolute) they force you to accept three incorrect beliefs that you don’t even realize are beliefs so that you cannot even argue with them. The argument is over the minute you accept their presentation. The first is that there exists a normal temperature.
There is no such thing as a normal global temperature for the Earth, there never has been
Second, that we know what the normal temperature should be, or would have been if we weren’t monkeying with it.
We have no friggin’ idea of what the temperature outside would have been if it weren’t for anthropogenic CO_2. We cannot hindcast past temperatures or predict future temperatures even if we assume CO_2 is held constant!/i>
Not for as little as three or four years. Certainly not ten years into the future (as Roy’s figures above graphically illustrate).
Third, that the variations presented as anomalies are significant. That is, that these variations are “unusual” and could not have arisen by chance. This, of course, directly implies that in the past, temperatures were normal, predictable, and had little natural variation.
The global temperature (as a measure of “global climate” is always varying. Separating out what fraction of this variation is due to anthropogenic causes is impossible unless there exists a “normal” temperature (there doesn’t) that we can predict (we can’t), that has no possible natural variations of the same order as the anthropogenic contributions.
In thirty to fifty years we might — and I do say might — have enough, good enough, data to start making progress on the causal web of major factors in the climate on timescales up to sixty to eighty years (the length of the reliable instrumental record, see Roy’s graphs above, that are remarkably consistent because they are not subject to the systematic errors and the kind of shenanigans that are routinely pulled with the thermometric surface temperature record). By then we might have a better handle on the Sun. We might have a clue about the upper atmosphere (above the troposphere) and how it affects global climate. Perhaps we will have a handle on the decadal oscillations. And by then the issue of CAGW/CACC will have empirically resolved itself, with the people who claimed to “know” beforehand how it would come out crowing about their “success”.
I’m not one of them. I have only a vague idea of how it will come out, quite independent of the simple physics of the GHE, because the climate isn’t simple and it could (start to) heat all the way up to levels it hasn’t seen in fifty million year or plunge down to temperatures not seen since the LIA and I wouldn’t be surprised either way. After all it could do either one of these things, could it not, in the complete absence of human influence! It has in the past, and we don’t know why it did and cannot predict or explain what it did. So how in the world can we argue that it will or will not do something in the future?
rgb

milodonharlani says:
June 6, 2013 at 9:43 am
Damn those pesky satellites.

I have no idea why he uses mid-troposphere data. Forget these pretty pictures. They do not correspond with lower troposphere where we live and which can be related to station data of thermometer measurements. This graph is actually worthless because neither ENSO oscillations nor the super El Nino nor the temperature trends we are interested in are even visible. Those are the features against which the models should be tested, and this graph does not do it.