Guest Post by Willis Eschenbach
I came across a curious graph and claim today in a peer-reviewed scientific paper. Here’s the graph relating sunspots and the change in sea level:
And here is the claim about the graph:
Sea level change and solar activity
A stronger effect related to solar cycles is seen in Fig. 2, where the yearly averaged sunspot numbers are plotted together with the yearly change in coastal sea level (Holgate, 2007). The sea level rates are calculated from nine distributed tidal gauges with long records, which were compared with a larger set of data from 177 stations available in the last part of the century. In most of the century the sea level varied in phase with the solar activity, with the Sun leading the ocean, but in the beginning of the century they were in opposite phases, and during SC17 and 19 the sea level increased before the solar activity.
Let me see if I have this straight. At the start of the record, sunspots and sea level moved in opposite directions. Then for most of the time they were in phase. In both those cases, sunspots were leading sea level, suggesting the possibility that sunspots might affect sea level … except in opposite directions at different times. And in addition, in about 20% of the data, the sea level moved first, followed by the sunspots, suggesting the possibility that at times, the sea level might affect the number of sunspots …
Now, when I see a claim like that, after I get done laughing, I look around for some numerical measure of how similar the two series actually are. This is usually the “R2” (R squared) value, which varies from zero (no relationship) to 1 (they always move proportionately). Accompanying this R2 measure there is usually a “p-value”. The p-value measures how likely it is that we’re just seeing random variations. In other words, the p-value is the odds that the outcome has occurred by chance. A p-value of 0.05, for example, means that the odds are one in twenty that it’s a random occurrence.
So … what did the author of the paper put forwards as the R2 and p-value for this relationship?
Sad to relate, that part of the analysis seems to have slipped his mind. He doesn’t give us any guess as to how correlated the two series are, or whether we’re just looking at a random relationship.
So I thought, well, I’ll just get his data and measure the relationship myself. However, despite the journal’s policy requiring public archiving of the data necessary for replication, as is too common these days there was no public data, no code, and not even a Supplementary Online Information.
However, years of messing around with recalcitrant climate scientists has shown me that digitizing data is both fast and easy, so I simply digitized the graph of the data so I could analyze it. It’s quite accurate when done carefully.
And what did I find? Well, the R2 between sunspots and sea level is a mere 0.13, very little relationship. And even worse, the p-value of the relationship is 0.08 … sorry, no cigar. There is no statistically significant relationship between the two. In part this is because both datasets are so highly auto-correlated (~0.8 for both), and in part it’s because … well, it’s because as near as we can tell, sunspots [or whatever sunspots are a proxy for] don’t affect the sea level.
My conclusions from this, in no particular order, are:
• If this is the author’s “stronger effect related to solar cycles”, I’m not gonna worry about his weaker effect.
• This is not science in any sense of the word. There is no data. There is no code. There is no mathematical analysis of any kind, just bald assertions of a “stronger” relationship.
• Seems to me the idea that sunspots rule sea level would be pretty much scuttled by sunspot cycles 17 and 19 where the sea level moves first and sunspots follow … as well as by the phase reversal in the early data. At a minimum, you’d have to explain those large anomalies to make the case for a relationship. However, the author makes no effort to do so.
• The reviewers, as is far too often the case these days, were asleep at the switch. This study needs serious revision and buttressing to meet even the most minimal scientific standards.
• The editor bears responsibility as well, because the study is not replicable without the data as used, and the editor has not required the author to archive the data.
So … why am I bothering with a case of pseudo-science that is so easy to refute?
Because it is one of the papers in the Special Issue of the Copernicus journal, Pattern Recognition in Physics … and by no means the worst of the lot. There has been much disturbance in the farce lately regarding the journal being shut down, with many people saying that it was closed for political reasons. And perhaps that is the case.
However, if I ran Copernicus, I would have shut the journal down myself, but not for political reasons. I’d have closed it as soon as possible, for both scientific and business reasons.
I’d have shut it for scientific reasons because as we see in this example, peer-review was absent, the editorial actions were laughable, the authors reviewed each others papers, and the result was lots of handwaving and very little science.
And I’d have shut it for business reasons because Copernicus, as a publisher of scientific journals, cannot afford to become known as a place where reviewers don’t review and editors don’t edit. It would make them the laughing stock of the journal world, and being the butt of that kind of joke is something that no journal publisher can survive.
To me, it’s a huge tragedy, for two reasons. One is that I and other skeptical researchers get tarred with the same brush. The media commentary never says “a bunch of fringe pseudo-scientists” brought the journal down. No, it’s “climate skeptics” who get the blame, with no distinctions made despite the fact that we’ve falsified some of the claims of the Special Issue authors here on WUWT.
The other reason it’s a tragedy is that they were offered an unparalleled opportunity, the control of special issue of a reputable journal. I would give much to have the chance that they had. And they simply threw that away with nepotistic reviewing, inept editorship, wildly overblown claims, and a wholesale lack of science.
It’s a tragedy because you can be sure that if I, or many other skeptical researchers, got the chance to shape such a special issue, we wouldn’t give the publisher any reason to be unhappy with the quality of the peer-review, the strength of the editorship, or the scientific quality of the papers. The Copernicus folks might not like the conclusions, but they would be well researched, cited, and supported, with all data and code made public.
Ah, well … sic transit gloria monday, it’s already tuesday, and the struggle continues …
w.
PS—Based on … well, I’m not exactly sure what he’s basing it on, but the author says in the abstract:
The recent global warming may be interpreted as a rising branch of a millennium cycle, identified in ice cores and sediments and also recorded in history. This cycle peaks in the second half of this century, and then a 500 yr cooling trend will start.
Glad that’s settled. I was concerned about the next half millennium … you see what I mean about the absence of science in the Special Edition.
PPS—The usual request. I can defend my own words. I can’t defend your interpretation of my words. If you disagree with something I or anyone has written, please quote the exact words that you object to, and then tell us your objections. It prevents a host of misunderstandings, and it makes it clear just what you think is wrong, and why.
Willis: “it’s a joke. The same is true of not requiring the “out of sample” testing of the Salvador model,”
I’ve already pointed out that this claim is incorrect. You ignore that I point it out and repeat it.
There is much to criticise in this paper and the whole review question. Repeating inaccurate claims and not reading the papers before trying to dismiss them is not part of it.
… color me unimpressed.
I am not a statistician, but it seems like R-Squared is the wrong measure for determining significance of periodic signals. Imagine you have a signal
y(t) = a Cos(w t) + b Cos(w2 t)
And you have a theory about some phenomenon that affects this signal has an output
o(t) = a(t) Cos(w t)
Now if b>a you do an R-squared on this you are going to get a low R. Low significance. But clearly this downplays the accuracy of o(t) in ‘partly’ explaining y(t).
As an experiment I would challenge you to do an R-squared of Milankovitch cycles versus temperature. I’m pretty sure the result will be terrible. Yet you wouldn’t simple go on to dismiss Milankovitch as irrelevant and coincidental would you? No, because the correlation is too high over too long a period and suggests cause and effect.
Matt: What’s curious, as you noted, is how quickly the curves changed from out of phase to in phase.
Willis: Causal relationships don’t do any of those things.
They do _exactly_ that kind of thing when there is an interference of two or more relatively close periodicities. I explained that in detail above and as usual you ignore what does not suit you.
The paper goes into much detail looking at what other periods are present and highlights the 9.1 year periodicity which would combine to produce something like the effects you noted. I explained that in detail above and as usual you ignore what does not suit you.
whether R^2 of 2 (or whatever) is significant needs to be assessed in relation to the number of data points in question and the kind of data, something which I don’t recall you ever having calculated despite your insistence of people calculating R^2 and publishing it , you seem to have some unstated idea of what is a “good” R2 value.
Calculating the R2 value and publishing it without any criterion to judge whether it is significant does not tell us much except that you think you know more than you do know.
I suspect the offset near 1990 would match up much better if the sea level data were corrected for the effect of the Mt. Pinatubo eruption.
tommoriarty says: “I suspect the offset near 1990 would match up much better if the sea level data were corrected for the effect of the Mt. Pinatubo eruption.”
I suspect we’d do a lot better if we _stopped_ trying to “correct” data . See last graph in this post:
http://climategrog.wordpress.com/2013/03/01/61/
.. and here:
http://judithcurry.com/2012/03/15/on-the-adjustments-to-the-hadsst3-data-set-2/
[i]Causal relationships don’t do any of those things. When you walk out into the sun, you get warm. It doesn’t fade away and get replaced by warming from the stars. Nor does it change phase, so that next time you walk out into the sun it makes you cold. Nor does one sometimes lag and sometimes lead the other—I don’t get warm a few seconds before walking outside in some cases, and a few seconds after walking outside in other cases.[/i]
Please quote the line in the paper where they say sunspots cause changes in sea level. Again, they aren’t saying there is a causal relationship. The sunspots and the sea level are both effects. It is perfectly plausible for sunspots and sea level to lag behind a common root cause at varying times. This is a terrible failure of logic.
There are no error bars in the graph of sea level change. The sea level itself is difficult to measure, and its changes doubly so. I would like to see the uncertainties in that graph.
Greg Goodman says:
January 22, 2014 at 11:44 am
Great, Greg. I’m sure, then, that you can compare for me the R2 and p-value of the out-of-sample forecast with the R2 and p-value of the in-sample forecast.
“Testing”, whether in or out of sample, requires measurement and comparison. Near as I can tell, they have done neither. But you say they have, and perhaps you are right … so where are the results of the out of sample tests?
w.
One option seldom addressed in climate science is the possibility of synchronization. The idea that over time a system can be perturbed into synchronous behavior even with very weak coupling. Many chaotic systems are ‘accepting’ of synchronizing with a periodic driver even if the coupling is very small. These would seem to be a fascinating area to look into and I think should be considered more frequently. Sun activity and climate seems a ripe area for synchronization to occur without the activity having to be at direct forcing levels.
What’s the R^2 of the rotation of the Earth with temperature of Edmonton. Probably not great if you don’t also take into account very important periodicity of Earth’s tilt (seasons) and random noise of ‘weather’ on top of it 😉
Greg Goodman says:
January 22, 2014 at 11:57 am
Thanks, Greg. Please give me a real world example of this from climate data, if you think it exists. You know, something like where the sun warms the earth for thirty years, then cools it for ten years, that kind of thing. Or take evaporation. Show me a dataset where for four decades, evaporation increases with wind speed … and then for three decades evaporation decreases with wind speed.
Next, show me a causal relationship where for several decades, the putative effect lags the claimed cause, and then for several decades the effect precedes the cause. You know, like where for half a century the tides trail in the wake of the movements of the moon and sun, and then for the next half century they precede the movements of the moon and sun. That one should be interesting.
As Fourier showed, and as you correctly state above, you can create anything by combining wave forms. Whether that particular situation exists in causal relationships here in the real world of cause and effect is another question. In particular, it’s rare for the effect to precede the cause …
w.
Greg Goodman said “I suspect we’d do a lot better if we _stopped_ trying to “correct” data.”
When data can be properly corrected it should be corrected.
The first link you provided is about smoothing techniques, but I did not suggest smoothing the data. However there are many occasions when smoothing is appropriate and useful.
The second link you provided concerns a specific data set – totally different from the sea level data that this post considers.
I think there is fairly broad agreement among alarmists and realists that the Mount Pinatubo eruption in 1991 reduced the sea level rise rate for several years. So why not correct for it in this context?
Consider…
Significant decadal-scale impact of volcanic eruptions on sea level and ocean heat content (Church, et. al., Nature, 2005)
http://www.nature.com/nature/journal/v438/n7064/full/nature04237.html
“For the Mt Pinatubo eruption, we estimate a reduction in ocean heat content of about 3 1022 J and a global sea-level fall of about 5 mm…
To better understand the processes involved, we focus on the Mt Pinatubo eruption (June 1991) because of its stronger radiative forcing and the better observations available. The primary driver of the fall in sea-surface temperature, GMSL and GOHC in the PCM is the rapid reduction in net solar flux at the ocean surface (Fig. 3a) of up to 6 W m-2 in late 1991. By early 1994, the net solar flux had virtually recovered to pre-eruption values”
Tom
To understand what is observed it is necessary to understand the mechanisms. Mass balance and temperature changes do not explain the recent ocean level changes. The paleo record shows unexplained large changes in ocean level that cannot be explained. Something not melting ice sheet or change in temperature is causing the ocean level to change.
The explanation as to why the ocean level changes following the periodicity of the solar cycle with a change in lead and lag is the same reason the boost satellites get from the close earth orbit maneuver has changed with time. We know gravity cannot change. There is a missing variable. Working with the laws of physics what variable could affect satellite orbits and ocean level.
(Ocean level is now falling based on raw satellite data. The raw data is adjusted to create the rise in ocean level.)
Mass and volume contributions to twentieth-century
global sea level rise
The rate of twentieth-century global sea level rise and its causes are the subjects of intense controversy1–7. Most direct estimates from tide gauges give 1.5–2.0 mm/yr, whereas indirect estimates based on the two processes responsible for global sea level rise, namely mass and volume change, fall far below this range. Estimates of the volume increase due to ocean warming give a rate of about 0.5mmyr21 (ref. 8) and the rate due to mass increase, primarily from the melting of continental ice, is thought to be even smaller. Therefore, either the tide gauge estimates are too high, as has been suggested recently6, or one (or both) of the mass and volume estimates is too low.
ftp://falcon.grdl.noaa.gov/pub/bob/2004nature.pdf
tommoriarty says: The first link you provided is about smoothing techniques
No, that article is not about “smoothing techniques”. It uses frequency filters to look at inter-annual and decadal scale variation in individual ocean basins and points out exactly the mistake Willis is trying make here in misreading the ‘phase crisis’ problem.
” See last graph in this post:” refers you to the part which is relevant to my comment about doing better if we stop “correcting” the data. It shows how Hadley processing is corrupting the lunar signal that is a strong component in many ocean basins and is the same one Scafetta and BEST (land) data show. It is also found in HadCRUT, presumably due to land portion since the “Had” part has removed , or more likely corrupted it into something else.
With many regarding HadSST as the “gold standard” of climate records, it hardly surprising natural cycle signals are not being detected.
All good clean fun. Of course, the ultimate proof of a hypothesis like this “There is a predictive correlation between sunspot numbers and the rate of SLR” is easily tested, since the graph above inexplicably ends at the year 2000. We have the additional advantage that in the subsequent 14 years, the sun has been at its lowest ebb in 100 years, and we have multiple sources for measuring SLR including the old tidal gauge network and the newer satellite-based measurements.
If the hypothesis is correct, then, we should have very low levels of SLR for the indefinite future, as predictions of sunspot levels are pretty low indeed.
But that raises many questions. One is “which sunspot numbers were used”, since Our Friend Lief suggests that sunspot numbers reported in the early 20th century do not match those reported in the late 20th century. This is an ongoing problem, I understand, but it is apparently an important issue if early numbers are all smaller by an (in)consistent scale factor as has been suggested. Something to clarify.
The bigger issue I have, however, comes from the SLR data. Say what? That’s not the SLR data. This is:
http://en.wikipedia.org/wiki/File:Trends_in_global_average_absolute_sea_level,_1870-2008_%28US_EPA%29.png
Note that both tide gauge data and satellite data (from two sources) are presented. Now, the data presented in the figure above is SUPPOSED to be the first derivative of this curve — its slope. And it’s not.
I count approximately thirty years in the actual record from 1900 to 2000 where SLR was negative. There are at most fifteen such years visible in the plot above. The clustering of negative years is also completely different — they are nearly uniformly distributed in the actual SLR data, and are somewhat antibunched, with a negative year more likely than not to be followed by a (possibly strongly) positive year. The bunching evident in the graph above is difficult for me, at least, to discern.
So I have to question: Not only which sunspot data, corrected or uncorrected, but where the hell did the SLR data come from? Because it certainly didn’t come from the graph I link above. In the graph above, only 4 decades in the entire century had years with negative SLR. In the actual data, there isn’t a single decade without multiple years of negative SLR.
If you plotted sunspot numbers against the actual coarse grained first derivative of SLR, I don’t think there would be any correlation at all.
rgb
tommoriarty says: “Consider…
Significant decadal-scale impact of volcanic eruptions on sea level and ocean heat ”
To support this claimed “decadal-scale impact” you quote a paper saying flux was back to normal by 1994. That’s less than three years.
The big problem is false attribution. There are temporally coincident cooling trends that are already under way _before_ the eruptions of both Mt P and El Chichon. Alarmists confound the two and make exaggerated claims of “decadal-scale impact” of volcanoes that allow them to jack up the CO2 sensitivity. This worked until we had a long period with virtually not volcanism. Then we saw the warming vs CO2 connection broke down.
Willis Eschenbach says:
January 22, 2014 at 3:01 am
Paul Westhaver says:
January 22, 2014 at 2:05 am
Willis,
Notwithstanding your gripes about the overall paper and the journal review process, If someone placed the curves in front of me, I;d be very interested i knowing why they are related. They are related without a doubt. So why?
No, they are not “related without a doubt”. What part of an R^2 of 0.13 and no statistically significant correlation seems unclear?
——————————————————–
Sorry Willis,
an r value is not necessarily an appropriate measure here. Considering its limitations is easy:
Two straight lines have an r value of 1, but the information content of that result would be close to zero.
However, if you have 2 identical curves with multipole ups and downs and thus a high information content, and shift the phase of one curve into different directions for its first and last half, you may end up with an r value of 0, though everyone would “see” that they are “related” or in this example even coherent.
That is why Shaviv can say that some correlations between solar variables and temperature are the best between any two variables in climate science.
And because of curves like this
http://kaltesonne.de/wp-content/uploads/2012/05/bond-et-al-2001.gif
I think, the existence of a solar amplifier is likely (though the r-value is “only” r=0.45)
Good evening, Willis. Would you be willing to publish your data from the digitized graph? I’m afraid my skills in that area do not match up to yours.
I ask because I am curious what the correlation would have looked like if one excluded the single out-of-phase period (which could perhaps be edge-effect in the data?) or what the correlation might be if one hypothesizes that the apparent lagging in cycles 17 and 19 were artifacts of measurement error. Such statistical tinkering after the fact can’t prove anything of course but it might lead to more interesting questions.
Manfred,
Exactly and good example. Correlation tells us if a particular factor is important. R^2 tells us if we how well we have put all the factors together into a predictive model. We can’t look at R^2 of 0.13 and conclude that sunspots are uncorrelated. That’s wrong. We can only conclude that a predictive model using only sunspots is incomplete. We have to do a correlation to determine if it is ‘correlated’ and clearly it is.
Who you gonna believe Willis, or your lying eyes? 😉
RGB, firstly you link to global average SL , that is not what the paper is using,
Secondly I recall a paper from Jevrajeva 2002 (?) that did a careful reconstruction just form tide gauges that had very clear 60 cycles. It may have been plotting 2nd diff IIRC. The official GMSL now includes GAIA [sic] adjustments and inv. barometer and you don’t to look at non-barmon data any more.
Also there were HUGE “corrections” to Jason data. In 2011 the GMSL record roughly followed UAH TLT, then something bad happened to it.
http://climategrog.wordpress.com/?attachment_id=524
Yet another climate parameter gets taken to the cleaners. I regard what Colorado our now presenting as having no analytic value. It has gone through political correction training.
Jevrejeva, S., J. C. Moore, A. Grinsted, and P. L. Woodworth (2008),
Geophys. Res. Lett., 35, L08715, doi:10.1029/2008GL033611.
Can’t find the graph but paper is worth a look.
Ian Schumacher:
At January 22, 2014 at 1:52 pm you ask
I don’t “believe” either. I accept the indication of the r^2 statistic, and that indication is that the paper is junk.
Richard
Greg Goodman said “No, that article is not about “smoothing techniques”. It uses frequency filters to look at inter-annual and decadal scale variation in individual ocean basins”
Yes, you are right. I mistakenly went to the home page of “Climategrog” instead of the sub-page that you specified. Sorry about that.
Greg Goodman also said “See last graph in this post:” refers you to the part which is relevant to my comment about doing better if we stop “correcting” the data. It shows how Hadley processing is corrupting the lunar signal that is a strong component in many ocean basins and is the same one Scafetta and BEST (land) data show. It is also found in HadCRUT, presumably due to land portion since the “Had” part has removed , or more likely corrupted it into something else. With many regarding HadSST as the “gold standard” of climate records, it hardly surprising natural cycle signals are not being detected”
This is only marginally relavent to the Holgate data.
It seems that you are implying a maxim that there are no appropriate corrections when you say “I suspect we’d do a lot better if we _stopped_ trying to “correct” data.” Did you mean the no data should ever be “corrected” (whatever that correction means) or are you referring to a specific reason that the Holgate data should not be corrected for the effect of Pinatubo in the early ’90s?
Willis, if you’d been asked to be one of the reviewers for the Copernicus papers, would you have said yes ?
lsvalgaard says:
January 22, 2014 at 10:34 am
Dr. S.
Curve fitting is my speciality, it’s great fun.
Here is something you might like to amuse yourself when you got some spare time:
http://www.vukcevic.talktalk.net/OldLadysHalo.htm
Any reason why (polarised) Aurora spectrum would have stronger correlation with the Earth’s core field than with the solar magnetic cycle that drives it?
Polarised?
Yes, following the NASA’s guidance on the mater: Solar coronal mass ejections CMEs in the even-numbered solar cycles tend to hit Earth with a leading edge that is magnetized north. Such CMEs open a breach and load the magnetosphere with plasma starting a geomagnetic storm.
Polarisation:
even numbered cycles Auroras = + (positive)
odd numbered cycles Auroras = – (negative)
Notice tiny 5 year ENSO signal in the Aurora’s spectrum, if Mr. Tisdale might like to know about, the explanation is waiting.
.