Jo Nova writes:
Murry Salby was sacked from Macquarie University, and Macquarie struggled to explain why, among other things, it was necessary to abandon, and strand him in Paris and hold a “misconduct” meeting in his absence. Since then he has been subject to attacks related to his previous employment. I’ve asked him to respond, which he has at length in a PDF (see below). The figures listed below refer to that PDF, which encompasses 15 years of events.
I don’t have the resources (unlike the National Science Foundation, the NSF) to investigate it all, but wanted to give Murry the right of reply.
On closer inspection the NSF report used by people to attack Salby does not appear to be the balanced, impartial analysis I would have expected. Indeed the hyperbolic language based on insubstantial evidence is disturbing to say the least. Because of the long detailed nature of this I cannot draw conclusions, except to say that any scientist who responds to a question about Murry Salby’s work with a reference to his employment is no scientist.
Remember the NSF report was supposedly an inhouse private document. It was marked “Confidential”, subject to the Privacy Act, with disclosure outside the NSF prohibited except through FOI. Desmog vaguely suggest there “must have been an FOI”, but there are no links to support that. In the end, a confidential, low standard, internal document with legalistic sounding words, may have been “leaked” to those in search of a character attack.
My summary of his reply:
See: http://joannenova.com.au/2013/08/murry-salby-responds-to-the-attacks-on-his-record/
The PDF:
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Nick Stokes says:
August 14, 2013 at 2:27 am
I do want to say, Nick, in case my frustration hides it, that I do have respect for your capabilities in general. You have shown in the past that you clearly have some formal training and reasonable understanding of issues involving signals and systems. If my comments to you appear caustic, it is because I am frustrated by what I see as your dogmatic refusal to think outside the box, and these needling potshots you take when you undoubtedly could reason these clearly evident considerations out for yourself.
Nyq Only says:
August 14, 2013 at 12:25 pm
You bring nothing to the table in this discussion. Please go away.
Bart says:
August 14, 2013 at 11:42 am
Show me your plot your plot like this one.
Here is the plot for the influence of temperature on the trend. Add to that 53% of the human CO2 emissions/year over the same period (which unfortunately is not possible in WFT) as a first approximation.
The derivative of the temperature stands for the short term reaction of CO2 on temperature changes, while these in reality react over 1-3 years. The longer term reaction of CO2 levels on temperature is ~8 ppmv/K, the ~0.4 K increase in temperature would give an increase of ~3.2 ppmv over a period of 50 years, hardly detectable in the 70 ppmv increase.
“You have commented in various ways that CO2 is the cause of global warming.”
Well I certainly believe that CO2 is a major driver of global warming the main issue I have been discussing in this thread has been whether the long term trend in atmospheric CO2 has an anthropogenic cause.
“I’ve been asking for empirical evidence in the form of a chart showing that claimed cause and effect.”
And that is a nonsensical request. A chart can only show correlation between two variables. How can a chart show cause and effect? Even if the variables match or there is a lag between one and the other you have no way just from a chart of knowing what other underlying causes might be in play (for example a third uncharted variable which is a common cause of the changes in two charted variables). That isn’t some wacky warmist position either but high-school interpreting statistics.
“I have posted charts from different sources that clearly show this cause and effect: ∆T causes ∆CO2.”
The best you can say about your chart is that it shows that the seasonal and year to year wobbles in both temperature and rate of change of CO2 are very similar. That takes us to the stunning conclusion that climactic variables are related – hey wow! it is almost like we are measuring stuff on the same planet! You want to get from that to a strong claim that temperature cannot be affected by carbon dioxide despite the fact that we know from the physical chemistry of carbon dioxide that it certainly can affect temperature.
Sorry but I don’t do faith.
Bart says: August 14, 2013 at 10:41 am
“No, not on your planet, where a rate of change is somehow the same as absolute concentration. Clearly, you are not following the conversation. The rate of change of CO2 and the temperature anomaly are affinely similar. Let us have no more denial of this fact.”
The graph I drew your comment is relating to is one showing the DERIVATIVE. You do know what a derivative is? No? oops
Allan MacRae says: August 14, 2013 at 8:03 am
“I cannot agree with Nyq at all – his arguments seems to be religiously-based rather than scientific – he says “we KNOW CO2 is a greenhouse gas”. This is apparently a specious statement, either false or insignificant.”
Well me and Murry Salby think CO2 is a greenhouse gas – I know the Slayer crowd have some issues with that but I think it should be OK to refer to basic physics and chemistry as stuff that we know without it being regarded as a religious statement. Of course if you think I’ve gone too far and would like to challenge the claim that CO2 is a greenhouse gas (one of the most significant greenhouse gases according to Prof Salby) then you should explain your argument.
Bart says: August 14, 2013 at 12:28 pm
“You bring nothing to the table in this discussion. Please go away.”
Well not the most gracious admissions of defeat that I’ve ever read but I guess it will do. Regarding your request for me to “go away” I believe that is a request solely within the domain of the managers of this blog rather than yourself.
Bart says:
August 14, 2013 at 12:14 pm
What is important here is that it the result of the approximation clearly indicates that what we are dealing with is mostly a temperature dependent pumping of CO2 into the atmosphere. This relationship, whatever it is in its full glory, accounts for ALL of the observed behavior, both in the long term and the short.
Even if we don’t take into account the proxies of earlier times, the alternative fits the data as good as your approach. The temperature changes are responsible for the short (1-3 years) variability of the CO2 rate of change. The temperature increase 1960-current is responsible for a small increase of CO2, in line with Henry’s Law of ocean releases. And human emissions are responsible for the bulk of the increase.
Further, as repeadetly said, your temperature dependent ocean process runs counter the observed 13C/12C ratio change, it should increase the turnover of CO2 in the atmosphere (for which there is no proof) and increase the decline of the bomb spike 14CO2 (no sign of that either). And last but not least, it violates Henry’s Law, as an increase of 3.2 ppmv in the atmosphere would stop any extra release of CO2 from the oceans for the 0.4 K increase in temperature.
Thus if you have two competing theories, both fitting the same CO2 increase in the atmosphere for short and medium term, one which fits all available current accurate observations, the other violating about every observation, which of them is the right one?
Ferdinand Engelbeen says:
August 14, 2013 at 12:37 pm
“Here is the plot for the influence of temperature on the trend. “
It is 90 degrees out of phase. This is very clearly not a match.
Ferdinand Engelbeen says:
August 14, 2013 at 1:03 pm
“Even if we don’t take into account the proxies of earlier times, the alternative fits the data as good as your approach.”
Not even close. You have a phase error of 90 degrees. Not even close.
Nyq Only says:
“…that is a nonsensical request.”
Not really. Your comment is simply a copout, because in fact my charts do clearly show cause and effect. Your problem is that you cannot find a chart that shows that the rise in CO2 is the cause of the rise in temperature. CO2 may have some tiny effect, but that effect is too small to measure.
I posted 2 charts that clearly show that ∆T is the cause of the change in CO2. One chart covers a few decades, and the other one covers the past 400,000 years. Everyone else can see that T leads CO2 on those time scales. If you look closely, you will see it too.
Nyq Only:
At August 14, 2013 at 12:39 pm
http://wattsupwiththat.com/2013/08/11/murry-salby-responds-to-critics/#comment-1390029
you ask
The reason you ask is that you mistakenly think,
“A chart can only show correlation between two variables.”
But a chart can also show coherence, and both correlation and coherence can each and both provide information pertaining to causality.
Correlation is a mathematical relationship between two parameters. If the correlation is known over the length of the data sets, then their correlation indicates the magnitude of a change in one parameter that is expected when the other parameter changes by a known magnitude.
Correlation does NOT indicate a causal relation between two parameters.
But
Absence of correlation indicates absence of a direct a causal relation between two parameters.
Coherence of two parameters indicates that when one parameter changes then the other parameter changes later.
Coherence can disprove that change of one parameter causes change in the other; i.e. if change in parameter A follows change in parameter B then the change of A cannot be the cause of the change of B (because a cause cannot occur after its effect).
So,
1.
absence of correlation indicates absence of a direct causal relationship
and
2.
when there is a direct causal relationship then coherence indicates which of the two parameters is causal.
Furthermore, coherence in the absence of correlation is strongly suggestive that both parameters are affected by another parameter (or other parameters).
For example, leaves fall off trees soon after children return to school following their summer break.
The coherence is great; i.e. both effects occur each year.
But the effects do not correlate; i.e. the number of returning children is not indicative of the number of falling leaves.
In this example, the time of year is the additional parameter which causes children to return to school and the leaves to fall off trees.
So, in the context of your question, yes, a chart show cause and effect if it is known that there is a causal relationship between two parameters. The coherence between the parameters indicates which is causal.
And, as has been repeatedly explained in this thread, global temperature and atmospheric CO2 concentration cohere such that changes in the CO2 follow changes in the temperature at all time scales.
Thus, if it is assumed there is a direct causal relationship between global temperature and atmospheric CO2 concentration then their coherence indicates the temperature changes cause the CO2 changes.
Unless, of course, you can provide a chart (as requested from you by dbstealey) showing the changes in the temperature following changes in the CO2).
Richard
Bart says: August 14, 2013 at 11:01 am
“Of course you can get a better fit of the slightly quadratic curve with a least squares fit. That’s because… wait for it… it’s a zarking LEAST SQUARES FIT, by definition the best quadratic fit there is!!!”
Yes, and I did a least squares regression fit with GISS temp too. It’s the green curve. It’s actually very similar to the quadratic fit. The integration means that GISStemp behaves very like its linear trend apporoximation would. The integration attenuates the residuals.
I think what you did in that integrated test is the right thing to do. Your diffentiated plots emphasise the fluctuations, and show that you have something that correlates with those. But in fact the Keeling curve doesn’t vary much – it’s a steady increase, and it’s the reason for that that people want to know. And matching fluctuations doesn’t help, while if your model could explain the bulk rise, it would help.
But I’ve put two datasets into the same optimising process – a straight line, and GISStemp. They do equally well. The extra information in GISS isn’t helping at all.
Bart says:
August 14, 2013 at 1:04 pm
It is 90 degrees out of phase. This is very clearly not a match.
Of course there is an out of phase: the CO2 rate of change lags the temperature changes with 6-9 months, which is visible if you take 12 month averages for both CO2 and temperature changes. If you only take a 12 month average for one of them, as you do, there is no out of phase, while in reality there is one…
dbstealey says: August 14, 2013 at 12:25 pm
“I have posted charts from different sources that clearly show this cause and effect: ΔT causes ΔCO2.”
Yes, it does, in those circumstances. It’s not in dispute. CO2 is less soluble in warm water, and charts like yours show about 10 ppmv increase in CO2 for each °C rise in temperature.
But we’ve had a 120 ppmv rise in CO2, and haven’t had a 12°C rise in temperature. Instead, we’ve been digging up carbon and putting CO2 directly in the atmosphere. This plot from the AR3 does the same differencing as BART and your plots, but clearly shows the cause – our emissions.
Ferdinand Engelbeen says:
August 14, 2013 at 1:31 pm
If you only take a 12 month average for one of them, as you do, there is no out of phase, while in reality there is one…
Here I was wrong. WFT does it as it should do: the 12 sample average is plotted in the middle of the samples…
Nick Stokes says:
August 14, 2013 at 1:30 pm
“…while if your model could explain the bulk rise, it would help.”
But, don’t you see? The temperature relationship accounts for the curvature, i.e., the quadratic term. It accounts for that, as well as all the variation. It is hardly a far stretch to say it accounts for the linear term as well, given that the temperature anomaly must have some baseline – the one it already has is arbitrary.
But, here is the nub: you cannot claim that human inputs account for it, because they would also induce curvature which is already accounted for by the temperature relationship.
Ferdinand Engelbeen says:
August 14, 2013 at 1:31 pm
“If you only take a 12 month average for one of them, as you do, there is no out of phase, while in reality there is one…”
No, the averaging is not why. The WoodForTrees site automatically advances its averages to account for filter lag. You can see this clearly if you choose, e.g., 12 month averages and 24 month averages. They both line up, despite the fact that the lag for a 12 month average is 6 months, and the lag for a 24 month average is 12 months (the time lag is half the length of the average). The averaging is zero phase (because the filtering operation is non-causal, due to the WFT advancement).
The reality is that CO2 lags temperature by 90 degrees, which is consistent with the derivative of CO2 being in phase with the temperature. You can see this in your plot, where you plot the derivatives of both. For instance, focus in the range from 1970 to 1980. There are about 3.5 cycles, corresponding to a frequency of 0.35 cycles/year. The time delay is the phase delay divided by the frequency, all transformed to radians and radians per year. So, that is pi/2/(2*pi*0.35) = 0.71 years, or about 8.6 months.
A 90 degree phase lag tells you that there is an integral relationship between the variables (CO2 is related to the integral of temperature, or temperature is to the derivative of CO2). It really cannot be simpler.
Nick Stokes:
In your post at August 14, 2013 at 1:59 pm
http://wattsupwiththat.com/2013/08/11/murry-salby-responds-to-critics/#comment-1390109
you wrote
I assume you were joking, but just to be sure I write to ask you to either confirm you were joking or – in the unlikely event that your strange assertion is serious – to please explain what makes you think it “clearly shows the cause” to be “our emissions”.
Richard
Bart says:
August 14, 2013 at 1:04 pm
It is 90 degrees out of phase. This is very clearly not a match.
Revision:
CO2 changes follow temperature changes on (near) all time scales, Thus the derivative of CO2 changes follows the derivative of temperature changes on (near) all time scales. Thus the out of phase of the derivative matches the out of phase of the the CO2 change in the atmosphere with the temperature change.
The “perfect match” of your plot is simply the result of not taking the derivative of the temperature change.
Ferdinand Engelbeen says:
August 14, 2013 at 2:04 pm
I see our missives passed one another.
Nick Stokes says:
August 14, 2013 at 1:59 pm
“This plot from the AR3 does the same differencing as BART and your plots, but clearly shows the cause – our emissions.”
But, what they don’t show is the affinely mapped temperature, which matches the CO2 rate of change. Surely, you can see that there is very little correlation between the bumps and wiggles of the emissions curve and the atmospheric concentration rate of change. This plot may help you see it more plainly. The two have been diverging since 1990, with accelerating divergence coinciding with the temperature lull of the last decade. You may reasonably expect that, if temperatures continue on the same trajectory or start decreasing, and it is a given that fossil fuel combustion is going to continue increasing, the divergence will soon become very pronounced.
“But we’ve had a 120 ppmv rise in CO2, and haven’t had a 12°C rise in temperature.”
See my recent post here @ur momisugly Aug 14, 2013 at 5:58 PM for my hypothesis on that score. I will repeat the pertinent parts here:
Ferdinand Engelbeen says:
August 14, 2013 at 2:19 pm
We keep dancing circles around each other.
“The “perfect match” of your plot is simply the result of not taking the derivative of the temperature change.”
Yes! Because the rate of change of CO2 is proportional to temperature anomaly with respect to a particular baseline.
That is the whole point! Your model does not match, because you are not acknowledging this relationship.
Bart says: August 14, 2013 at 2:12 pm
“The temperature relationship accounts for the curvature, i.e., the quadratic term.”
No it doesn’t. The curvature is multiplied by a number selected by the optimisation, or by your estimate attempting to optimise. Any straight line will give the same result. Curvature comes from the integrated trend. But it could have been the trend of anything. Gistemp is not contributing knowledge there.
Nick Stokes says:
August 14, 2013 at 2:32 pm
“The curvature is multiplied by a number selected by the optimisation, or by your estimate attempting to optimise.”
But, that number is the number which matches up both the trend and the variation. This is not happenstance.
Bart,
No, the trend is arbitrarily variable, via your T0 (-0.4). And the curvature is arbitrarily variable by the multiplier of Gi (your 0.2). Both are varied independently.
Bart says:
August 14, 2013 at 2:25 pm
We keep dancing circles around each other.
Agreed.
That is the whole point! Your model does not match, because you are not acknowledging this relationship.
Because there is no such relationship. On all time scales, a change in CO2 follows a change in T and then it ends: no further increase or decrease of CO2, if the temperature doesn’t change.
That is the case for seasonal changes, where CO2 lags temperature with ~3 months, as good as the derivatives of both do.
That is the case for short term changes, where CO2 lags temperature with 6-9 months, as good as the derivatives do.
That is the case for (very) long term changes, where CO2 lags temperature with 50 to several thousand years as good as the derivatives do.
The time delay is the phase delay divided by the frequency, all transformed to radians and radians per year. So, that is pi/2/(2*pi*0.35) = 0.71 years, or about 8.6 months.
Indeed, that is what is found in the observations: the change in CO2 increase rate lags the temperature increase after e.g. an El NIño with 6-9 months. Thus my “model” reflects reality…