People send me stuff.
Engineer and aerospace pioneer Burt Rutan writes to me in an email today:
The chart the Alarmists do not want you to see. Human Carbon emissions vs. The ‘Gold Standard’ global temperature data set (chart from C3).
The alarmists are now fighting hard to protect their reputations and their damaged careers, not fighting to protect a failed theory of Dangerous Human GHG warming.
The grey bars represent CO2 emissions in gigatons (GT).
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Day-By-Day,
Here is where that last chart came from. Lots of good charts and info there, like this, which contradicts Shehan’s nonsensical fabrication. I wonder how many hours he spent re-jiggering the WFT database to get that phony overlay? Here is the true non-correlation between CO2 and T. Note that global warming is not accelerating. In fact, global warming has stalled for the past decade at least.
And here is the only measured correlation between CO2 and T: it clearly shows that ∆CO2 is the result of ∆T — not vice-versa. Because there is no such vice-versa. Empirical evidence proves that CO2 follows temperature. There is no equivalent chart showing that CO2 leads temperature. At current concentrations, CO2 has essentially zero effect on global temperature. Planet Earth agrees.
Warmists used the 97-98 El Nino driven temperatures to proclaim that the warming was irrefutable and that “the science is settled”! I argued (in online discussions back then) that it was inappropriate to use the El Nino as a proof of man-made global warming, and that it would come back to bite them in the backside. They persisted in using it to claim a dramatic warming trend.
Now…it has come back to bite them in the backside. Now it is ‘cherry picking’ if we show the trend starting from that strong El Nino year. Sorry warmists…we are just following your rules.
Tad is absolutely right. This is just the same style presentation as the warmists have always used. Show the graph far and wide. Make it as well known as the bogus hockey stick graph. Make it a modern day icon. Once we stop this stagecoach from going over the regulatory cliff, we will have plenty of time to teach proper graphing in science 101.
Philip Shehan says:
January 24, 2013 at 4:39 am
Doesn’t match the wiggles, so the similarity is superficial. This does match. This shows the CO2 rate of change is affinely related to temperatures, which leads inexorably to the conclusion that temperatures lead CO2, and therefore CO2 is causally related to temperature, and not the reverse. Moreover, it accounts for the whole enchilada, and human inputs are superfluous, i.e., rapidly sequestered and of little consequence.
Here, we see how the relationship unfolds.
Eh, the graph isn’t terrible. The Y-axis thing is very common in business presentation.
Would it better with a zero-bounded Y-axes? Sure. Is it as bad as “hide the decline?” Not hardly.
A much more salient complaint is that it would have made a lot more sense to show atmospheric concentration rather than emissions.
Old wine in new bottles, old arguments in new pictures–nothing really in here but a variant presentation of a well-worn argument, a discussion of whether or not the presentation is disingenuous, and Burt Rutan’s name attached to it. I’d have attributed this to a slow news day, were it not for the other dozen articles that appeared the same day–perhaps a backlog being cleared? Anyway, the scale for the carbon bars is misleading, as is breaking up the curve into two arrows, let alone the choice of the break. I sort of hope Mr. Rutan knows this, and is, perhaps overenthusiastically, trying to emphasize a point, rather than simply being confused into believing this graphic himself. Meanwhile, realeased to the general public, this graphic might score some points, but these would have to be weighed against the backlash that would follow as some, at least, found out that it uses some simple ruses, at least one of which is taken straight from the pages of How to Lie with Statistics.
D.B.Stealey, I expanded the chart to show the two sections, with overlap at 1998. Rising, then flat, far better than Burt’s chart, I agree.
http://www.woodfortrees.org/plot/rss/from:1983/to:1999/plot/rss/from:1983/to:1999/trend/plot/esrl-co2/from:1983/normalise/plot/rss/from:1997/plot/rss/from:1997/trend
JazzyT,
Explain this, which clearly shows the non-correlation between CO2 and T.
The central issue in the debate is whether ∆CO2 causes ∆T. Obviously, it does not.
The alarmist crowd has cause and effect backward. Since their premise is wrong, their conclusion will necessarily be wrong. And in fact, that is what the planet is telling us.
There is no measurable AGW. And there never was.
• • •
Doug Jones,
Excellent chart. Thanks.
And here is the same style of chart, with 1998 omitted altogether, and thus no cherries to be picked. Still no sign of CAGW.
http://www.woodfortrees.org/plot/rss/from:1983/to:1997/plot/rss/from:1983/to:1997/trend/plot/esrl-co2/from:1983/normalise/plot/rss/from:1999/plot/rss/from:1999/trend
cbrtxus says:
January 23, 2013 at 6:01 pm
http://www.mediafire.com/view/?x187ovavw9d9rnh
cbrtxus: That graph is even more interesting in that it shows that the ONLY time that increasing CO2 tracked to increasing temps was during the 1980-2000 time frame — all others, before and after, had NO correlation at all!
Why do people keep looking at correlating CO2 and temperature? CO2 and temperature are not directly correlated, period. The rate of change of CO2 and temperature are.
People will reasonably complain that this graph is misleading because the y axis for CO2 emissions starts at 250, thus making the bars give an exaggerated impression of the difference between emissions between 1983-1997 (331) and 1998-2012 (440).
But, that said, the flat portion of the temperature line ought to nonetheless be compelling to most people.
No, that’s not the misleading part. The misleading part is the missing 287 degrees Kelvin. If we were to draw the CO_2 the same way that the temperature is portrayed, as an “anomaly”, the first bar would be roughly -55 GT, and the second bar would be +55 GT. If we were to portray both CO_2 emissions and temperature on an absolute scale, the CO_2 emitted would go up by roughly 1/3 while the temperature went up — well, you wouldn’t be able to resolve the variation compared to the thickness of the line at this scale!. It is, after all, less than 0.4 C out of almost 300 degrees, roughly one part in 700. A one pixel thick line would be absolutely flat at most reasonable screen resolutions.
I’m just sayin’…
On the other side of the coin, fitting the two segments this way is entirely arbitrary and rather meaningless. Fitting the entire interval gives you a linear trend somewhere in between, say 0.1/decade give or take or 1 degree a century. Or maybe a bit less. It doesn’t really matter, because a linear fit has no more meaning than a quadratic fit or a logistic fit or a skewed gaussian curve with kurtosis fit. Some of these might work better than others, but lacking a meaningful basis for the model selected you might as well just draw a fairly smooth curve through the data by hand and say “Look, I’ve found a good fit!”
If you wanted to do something sorta-meaningful, one could import the curve and the two CO_2 data into R, block average the first half and the second half of the temperature series, do a linear model between the two data points each, and discover that there is a comparatively low p-value, suggesting that the relationship is “significant”. Or import the actual CO_2 concentration as a timeseries with the same number of points, form a linear model, and again find from the p-value that the relationship is “significant”. Or import data from the stock market over the same interval, for a linear model, and find that the relationship is “significant”. Or import data from the annual measured height of my oldest son, for a linear model, and find that the correlation is “significant”.
Sadly, as it is, the graph up above is utterly devoid of statistical or scientific meaning. It doesn’t disprove AGW. If anything it supports the hypothesis, although that support is pretty weak, barely better than an even bet.
If anybody wants to debate the actual statistical significance of this or that, well, I’ve got R loaded and ready to go. Bring it.
rgb
Meanwhile, realeased to the general public, this graphic might score some points, but these would have to be weighed against the backlash that would follow as some, at least, found out that it uses some simple ruses, at least one of which is taken straight from the pages of How to Lie with Statistics.
Yes, exactly. A wonderful book. And of course presenting the global temperature as an anomaly in the first place is right out of one of its chapters, is it not? So that every curve ever presented of Global Warming commits the same sin. Presenting it with the CO_2 “anomaly” on a similar scale is just as bad. Chopping the bottom part of the human fraction of CO_2 — awful. Ignoring the rest of the CO_2 contributed by non-human sources, or the CO_2 sinks — evil. Fitting two linear models at a more or less arbitrary break point — argumentative, not meaningful.
Why bother? The data is what it is, and says what it says without any fits at all unless the fits have some specific basis!
Seriously, the best that can be said for the data is that it weakly supports the AGW hypothesis there is a positive linear trend in the overall data for both CO_2 and temperature. If you think that the AGW hypothesis posits a linear relationship, which it doesn’t.
rgb
I see little reason to declare my graph a “nonsensical fabrication”. It is very simple really. I have plotted the Muana Loa CO2 data since such reords began and the temperature data from the same period and shown that there is a rather good overlay.
A number of other graphs have done precisely the same thing, using shorter time periods and adopting different y-scaling.
Bart: could you explain how the use of the Derivate WFT function clarifies things in the presentation of your graph?
Derivative – Take the first derivative of the data (each sample has the one before subtracted)
Doesn’t match the wiggles, so the similarity is superficial. This does match. This shows the CO2 rate of change is affinely related to temperatures, which leads inexorably to the conclusion that temperatures lead CO2, and therefore CO2 is causally related to temperature, and not the reverse. Moreover, it accounts for the whole enchilada, and human inputs are superfluous, i.e., rapidly sequestered and of little consequence.
I have to admit, Bart, that the curve you present is at least the kind of curve that would be enormously compelling with just a little more underpinning. At least it kinda slaps you in the face and says “There is something here that needs to be explained”, whether or not one accepts the explanation you offer.
rgb
I still see flat, jump flat!
It’s ENSO!
DaveE.
I tend to agree. But the time baseline is absurdly short to make any dramatic assertions in a (probably) highly multivariate function with many factors contributing with many timescales and probably multiple internal nonlinearities. Coincidence is too likely, or it might have been ENSO then, but ENSO might produce this jump only when three other variables (e.g. the phase of the PDO, the phase of the moon, whatever) have just the right values. You can’t read too much into the one-dimensional projection of a curve that really lives in five or ten dimensions.
I don’t think people appreciate this. There may be no simple predictors.
rgb
Philip Shehan says:
January 24, 2013 at 1:27 pm
“Bart: could you explain how the use of the Derivate WFT function clarifies things in the presentation of your graph?”
The graph shows that CO2 concentration is related to the temperature anomaly by
dCO2/dt = k*(T – To)
where “k” is a coupling constant, and “To” is an equilibrium temperature baseline. This relationship holds to a high degree of fidelity over the modern era since precise measurements of CO2 began.
We choose “k” and “To” to match the two series. Then, when we integrate that function of temperature, starting from the initial CO2 measurement, we get a very close match to the absolute level of CO2 as a function of time, as we must, since we are integrating essentially the same area under the rate curve.
It happens that, when we choose “k” such that the variations match, the long term slope also matches. The important part is this: that value of “k” produces an excellent match of the curvature in the integrated result. Since the accumulated human emissions also have a curvature, in order to add them in consistently, we would have to back off “k”. But, if we back off “k” significantly, we will no longer match the variations in the original plot.
The upshot is, there is no room for significant human contributions. The CO2 system on the Earth is very tightly regulated, and human emissions are readily dealt with. But, the equilibrium level changes with temperature, and that is what drives the level of CO2 concentration in our atmosphere.
I should have added, the relationship
dCO2/dt = k*(T – To)
squarely directs the arrow of causality in the direction of temperature to CO2. Why? Well obviously, on the one hand, it makes no sense that temperature would depend on the rate of change of CO2 – if CO2 ever stopped rising, T would go back to To, no matter how much CO2 had accumulated.
Another point: the derivative has a leading phase. For example, if we set T = To + a*cos(w*t), then
dCO2/dt = k*a*cos(w*t)
hence
CO2 = (k*a/w)*sin(w*t) + CO2(0)
Since the sine function lags the cosine function by 1/4 cycle, we see that changes in CO2 occur after the changes in temperature.
As to how the integral relationship comes about, I can only offer my own speculations. It can help to consider the ocean currents. I hasten to say that I am NOT claiming that this IS what is happening, but it is food for thought as to how it MIGHT be happening.
Ocean waters are continually downwelling near the poles and upwelling in the tropics. The waters going down have a CO2 concentration which is dependent on the surface temperature. The upwelling waters, too, give up the CO2 they are carrying dependent on surface temperature. Since this is a continuous transport to and from the deep oceans, the rate of change of CO2 is a function of surface temperature
dCO2/dt = F(T)
A first order Newton expansion of F(T) is then
dCO2/dt = F(T0) + F'(T0)*(T – T0)
setting
k = F'(T0)
To = T0 – F(T0)/k
then leads to the affine relationship
dCO2/dt = k*(T – To)
Other continuous transport processes than the oceans act in the same way.
Great so, now what happens to the human emissions? Assuming there are robust permanent (or, at least, semi-permanent) sinks, we can posit a dual set of differential equations of the form
dCO2/dt = (CO2eq – CO2)/tau + H
dCO2eq/dt = k*(T – To)
where H is the human input rate, and tau is a time constant.
CO2 now tracks the CO2eq level while, if tau is small, the influence of H is rapidly diminished. How could such a two-tiered response to input CO2 come about? Well, I have several ideas.
The simplest has to do with where the input to the atmosphere originates. If we neglect phytoplankton for a moment (assuming perhaps that they are not terribly active in the upwelling and downwelling regions, which certainly seems likely enough for the extremely cold downwelling ones) then the CO2 coming out of the ocean, or inhibited from going into it, has to travel a long way to reach land sinks, becomes well mixed in the atmosphere, and takes a long time to be sequestered due to the slow diffusion process. But, the CO2 generated on land near the surface can be rapidly sequestered by land sinks.
That’s my conjecture and, as such, I am not going to get into any nasty quarrels with people with a chip on their shoulder about it, as so often seems to happen whenever you offer up anything. Others are free to consider other possibilities. But, the result has to conform to the observation. And, the observation says that CO2 is temperature dependent, and humans have, at most, a small effect.
In response to the question: “I wonder how many hours he spent re-jiggering the WFT database to get that phony overlay?”
Not very long at all actually. I used a plot from an earlier post as a template.
Steveta_uk says:
January 24, 2013 at 2:59 am
As you can see there are precisely two series in my plot. The only additional functions added are the mean for a 12 month smoothing of the data and an offset and scaling to bring them into line.
In fact spending just a little more time gives a better fit:
http://www.woodfortrees.org/plot/hadcrut4gl/from:1958/mean:12/plot/esrl-co2/from:1958/normalise/scale:0.75/offset:0.2
It is probably the simplest of all the WFT graphs presented in this thread.
Bart: Thank you for your detailed explanation. It is a bit complex for me to try to understand all in one go but I will give it some consideration.
Going back beyond Muano Loa data and using ice core data gives the following graph for CO2 concentrations.
http://tinyurl.com/atd7efj
This is a reasonably good match for the rise in temperature over the same period.
http://www1.picturepush.com/photo/a/11901124/img/Anonymous/hadsst2-with-3rd-order-polynomial-fit.jpeg
Thanks, Bart. But I doubt you’ll get an answer, or even an acknowledgement.
Regarding the simple-minded overlay of a temperature graph onto a CO2 graph, that means nothing, because it does not show any cause-and-effect relationship. You and I have both shown the CO2/T cause-and-effect: ∆T causes ∆CO2 — not vice-versa. There is no graph like this showing that CO2 leads temperature. The chart proves that changes in CO2 follow temperature.
Finally, Shehan is still posting his thoroughly debunked SkS chart. There is no credible scientific evidence showing accelerating global warming. None. All of the empirical evidence available shows that global warming is decelerating. Just because John Cook can post a fabricated chart does not mean it is legitimate; it isn’t. This is the real deal, straight from HadCrut. Shehan appears to be about as honest as Greg Laden.
Philip Shehan says:
January 24, 2013 at 4:10 pm
I appreciate the consideration.
The problem with your plot showing somewhat of a match is that it is entirely superficial. It’s basically a flip of the coin, in that they have the same sign for relatively shallow curvature. With that, you can generally get a pretty good fit between arbitrary data sets with an affine transformation merely by using a least squares fit to find the best coefficients of that transformation.
When, however, you can match the fine detail as well, as we do comparing the CO2 rate of change with temperatures, then you know you’ve got something beyond mere chance.
D. B. Stealey says:
January 24, 2013 at 5:45 pm
Please note how the temperature lead is generally 1/4 wavelength for the oscillations, consistent with the relationship I have provided.