Spot the science error

Guest post by Dr. Leif Svalgaard

The following abstract of a poster to be presented next month at the Fall Meeting of the American Geophysical Union caught my eye:

Session Title: GC11A. Diverse Views From Galileo’s Window: Solar Forcing of Climate Change Posters Chair: Willie Soon, Nicola Scafetta, Richard C Willson

ID# GC11A-0685: Dec 14 8:00 AM – 12:20 PM

Revised Assumptions and a Multidiscipline Approach to a Solar/Climate Connection

C. A. Perry (US Geological Survey, Lawrence, KS, USA).

Click to enlarge

Abstract:

The effect of solar variability on regional climate is examined using a sequence of physical connections between solar variability , Earth albedo, ocean temperatures, ocean currents (Ocean Conveyor Belt), and atmospheric patterns that affect precipitation and streamflow. The amount of solar energy reaching the Earth’s surface and its oceans is thought to be controlled through an interaction between Galactic

Cosmic Rays (GCRs), which are theorized to ionize the atmosphere and increase cloud formation. High (low) GCR flux may promote cloudiness (clear skies) and higher (lower) albedo at the same time that Total Solar Irradiance (TSI) is lowest (highest) in the solar cycle which in combination creates cooler (warmer) ocean temperature anomalies. These anomalies have been shown to affect atmospheric flow patterns and ultimately precipitation over the Midwestern United States. A study has identified a relation between geomagnetic index aa (GI-AA), and streamflow in the Mississippi River Basin for the period 1878-2004. The GI-AA was used as a proxy for GCRs. There appears to be a solar “fingerprint” that can be seen in hydroclimatic time series in other regions of the world, with each series having a unique lag time between the solar signal and the hydroclimatic response. A progression of increasing lag times can be spatially linked to the ocean conveyor belt, which could transport the solar signal over a time span of several decades. The lag times for any one region vary slightly and may be linked to the fluctuations in the velocity of the ocean conveyor belt.

A graph is attached to the abstract (as seen above):

http://www.leif.org/research/MissGeomagGraphBW.jpg

The poster seems to report on earlier work presented here:

http://ks.water.usgs.gov/waterdata/climate/

Where the same figure appears.

Now, what is wrong about this graph [and the conclusion, of course] ?

I’ll let you all find out what.

It is an example of three things:

  1. The desperate need for establishing a Sun-Climate [or is it weather, when on a decadal basis?] causing this kind of sloppy work (the graph contradicts the mechanism given for it)
  2. The lack of internal quality control by USGS
  3. The lack of quality control by the conveners of the AGU session.

UPDATE:

Thanks to all the readers who so generously [some gleefully] have pointed out my misinterpretation of the figure. This, of course, makes my initial assessment of the quality control moot and void, with an apology to those involved. Perhaps this shows how important a graph can be [cf. the impact of the Hockey Stick] and how important is clear labeling of what is shown.

UPDATE2:

click to enlarge

Since GCRs follow the the sunspot numbers and not the aa-index, the proper parameter to compare with would be the sunspot number. This also allows use of the streamflow data back to the beginning of the series in 1861. The following Figure shows the correlation with this parameter, providing a prediction of the flow to beyond 2040, should the flow indeed be correlated with the sunspot number 34 years earlier.

Advertisements

  Subscribe  
newest oldest most voted
Notify of
Alvin

It appears that his hand slipped as he was drawing the graph 🙂

R. Craigen

Easy. Except for the 1930 and 1998 peaks (in which no directional cause may be inferred) it would appear (assuming a causal relationship) that Mississippi flow forces Geomagnetic flux with a 1-3 year lag.
That’s why they call it “The Mighty Mississippi”, I guess.

crosspatch

Something seems to be, in the vernacular, bassackwards.
Why would a higher flux result in greater flow? Wouldn’t that result in fewer cosmic rays, clearer skies, and potentially less flow?

crosspatch

And yeah, why does the graph of flux extend into the future?

Looking at my guest comment, it struck me that an interpretation of the graph could be that the streamflow could be controlled by solar activity three solar cycles earlier. This would make my complaints moot, but that conclusion was so far from my mind that it did not register as a possible one. What say all of you?

R. Craigen

Well, Leif, I’ve heard it said that the Mississippi is an old man, thus he has a LOONNGG memory. You never know …
‘course, one might ask if the paper itself posits a 33-year time lag. I’d bet against it.

R. Craigen

While the numbers for Geomagnetic flux for 2011, and perhaps 2012, appear to be in already, it appears that the Mississippi stopped flowing somewhere around 2004 or 2005…

Mike McMillan

I’d say it’s not connected because the Mississippi river flow is controlled by a continuous set of locks and dams along its entire length, not so much by nature.

Kurt

“Leif Svalgaard (21:33:34) :
Looking at my guest comment, it struck me that an interpretation of the graph could be that the streamflow could be controlled by solar activity three solar cycles earlier. This would make my complaints moot, but that conclusion was so far from my mind that it did not register as a possible one. What say all of you?”
I think this was what was intended. Following the second link you provided, I found the following quote below the graph:
“Comparison of the 36-month moving average of the Mississippi River streamflow and the Geomagnetic Index AA. Streamflow has been lagged 34 years after the Geomagnetic data.”

rbateman

#1 – the Mississippi River flow shows little overall trend
#2 – the aa index mirrors, falls below and rises above in agreement and contrarian to the river flow. There is no consistency to the relationship, linear or non-linear.
What it says is that climactic zones of the Mississippi River basin are too varied to utilize flow rates as an indicators of continental climate.
There is no way to distinguish competing zones having disparate precipitation from a truly basin-wide average year. GCR levels would be too far ahead in the chain of command even if they were unquestionably first cause.

Kurt (21:45:09) :
>i>“Comparison of the 36-month moving average of the Mississippi River streamflow and the Geomagnetic Index AA. Streamflow has been lagged 34 years after the Geomagnetic data.”
The graph makes it look like the aa is lagging. OK, so I was wrong [given that interpretation – goes to show how careful a graph should be made in order not to make a wrong conclusion easily]. If indeed a 34-yr lag was meant, then the last 34 years of aa would provide a prediction of the streamflow for the next 34 years. THAT would have been an interesting plot.

rbateman

If you have several thousand years worth of data, like the Nile or the Yangtzee Rivers, and a good proxy for aa, you might be able to find some sort of repeatable sequences. Failing that, compare as many varying river basins for the 1860-present timeframe and see what emerges.

Mark Twain commented on the Mississippi flow years ago. Here is his quote
“In the space of one hundred and seventy-six years the Lower Mississippi has shortened itself two hundred and forty-two miles. That is an average of a trifle over one mile and a third per year. Therefore, any calm person, who is not blind or idiotic, can see that in the Old Oolitic Silurian Period, just a million years ago next November, the Lower Mississippi River was upwards of one million three hundred thousand miles long, and stuck out over the Gulf of Mexico like a fishing-rod. And by the same token any person can see that seven hundred and forty-two years from now the Lower Mississippi will be only a mile and three-quarters long, and Cairo and New Orleans will have joined their streets together, and be plodding comfortably along under a single mayor and a mutual board of aldermen. There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.”

I was wrong in calling the work ‘sloppy’. Tricky to interpret, perhaps, and a bit far out: the 3 cycle delay is hard to swallow. Perhaps, that is what makes the paper a breakthrough, of sorts…

a jones

I showed it to the cat. As far as I can tell he is laughing.
Of course I cannot be sure of that since I have not done a statistical analysis as to whether he is laughing.
Must get government grant to investigate this further.
Kindest Regards

R. Craigen

While the hypothesized causal relationship sounds plausible, I question a 33 or 34 year time scale. Geomagnetic phenomena affect incoming cosmic rays, which affect atmospheric ionization (instantly — there is no time lag). The ionization affects cloud formation, presumably on a time scale of hours to .. perhaps … weeks (that’s probably too long). Cloud formation affects precipitation on a scale of weeks or days (here I’m completely ignoring the chaotic dynamical system, just looking at the direct path of water from evaporate to precipitate). From the event of precipitation one might imagine perhaps a 2 year residency within the Mississippi river system (not including deep lake water that may be considered, for purposes here, fixed and outside consideration). There is very little multi-year snow/ice in the Mississipi basin, so there is no further storage beyond the system of locks and dams mentioned by Mike McMillan above.
At the outside I can imagine sources for up to 3 years of lag between geomagnetic flux changes and affect on Mississippi flow.
I think the hypothesis remains plausible, but that Mississipi flow is too problematic an indicator. Is there not a record of worldwide precipitation that can be obtained or reliably proxied? That would be the right place to look to support or falsify the hypothesis. Too many things can go wrong in a geographically localized proxy such as this.

It takes a big man to admit when he’s wrong.

The author has it backwards: The flow of water in the Mississippi controls the sun.

Roger Knights

This, like much here, is way over my head. But here’s a nit I can pick, in Leif’s intro: “at the Fall Meeting of …”
The seasons aren’t capitalized.

Mike McMillan

I agree with Mark Twain.
Measured at St Louis, you’re just downstream of where the Missouri and Illinois rivers join in, and lest we forget the #%@& up in Chicago reversed the Chicago river in 1900 so their sewage dumps into the Mississippi, rather than polluting their sacred Lake Michigan. Did I leave anything out?
The connection with the oceanic conveyor is laughable. Increasing ‘lag’ times just mean there’s no correlation, and therefore no causation. That doesn’t mean I intend to take off my tinfoil hat, though.

David Ermer

The attempt to correlate a single factor to another in a complex system where any number of factors might dominate at different times?

Jeremy

The graph seems to roughly correlate but it AIN’T LAGGED! OOOPs it kind of undermines their thesis of a multi-decadal delay.

Paul Vaughan

I’ve been working on these interannual-timescale aa index phase-relations with precipitation and other climate indices on-the-side. I am 100% certain there is something to them [that is possibly going to look dead-simple once someone works out some key detail(s)]. I can devise dozens of aa-based variables that show clearly-nonrandom phase-relations and I can get the eras of phase-match to shift around (by working with different timescale-contrasts), but I’ve put the investigations aside for awhile because:
a) it is complex (…so I have to wait for revelation — complex discovery is not easily forced onto a schedule).
b) the research community is not terribly supportive of pioneering that does not [yet] have all the phase-reversals worked-out.
c) funding & opportunity direct focus.
Btw: I’m not convinced that the signals of interest are (entirely) of solar origin. (See the works of Yu.V. Barkin and run comparisons with SOI.) Also, I don’t assume this is about GCRs.
Very interesting. I plan to look into this again in future (hopefully following a revelation).

Roger Knights (22:18:50) :
This, like much here, is way over my head. But here’s a nit I can pick, in Leif’s intro: “at the Fall Meeting of …”
The seasons aren’t capitalized.

Nit pickers are my specialty 🙂
From AGU’s website:
The Fall Meeting is expected to draw a crowd of over 16,000 geophysicists from around the world. …

Steve Schaper

I do believe that this is a case for Dr. Boli.

tokyoboy

Could this become a peer-reviewed paper and an additional source for their funding? I’m dying of curiosity.

GGM

This might just be 2 data sets that superficially appear to have a correlation.

rokshox

Isn’t it remarkable, though, that Mississippi flow rates appear to have an 11 year periodicity?

Steve Schaper (22:41:52) :
I do believe that this is a case for Dr. Boli.
1st question in Dr. Boli’s IQ test:
1. What is the next number in this series?
827, 827, 827, 827, 827, ___

Paul Vaughan

Mike McMillan (22:22:48) “Increasing ‘lag’ times just mean there’s no correlation, and therefore no causation.”
Acoustic info can manifest itself in interesting ways in time-integrated cross-correlation best-lag matrices. It’s not meaningless, but rather challenging to decipher (particularly when there are nearly-overlapping harmonic scales).

Bulldust

I don’t see anything wrong with this paper that could not be fixed with judicious use of one of these:
http://www.thinkgeek.com/geektoys/collectibles/9fc6/

Kurt

“There appears to be a solar “fingerprint” that can be seen in hydroclimatic time series in other regions of the world, with each series having a unique lag time between the solar signal and the hydroclimatic response. A progression of increasing lag times can be spatially linked to the ocean conveyor belt, which could transport the solar signal over a time span of several decades.”
If I understand this correctly, the abstract is claiming that they have matched streamflow to solar activity in many regions besides the Mississipi basin, albeit by using different lag times. I’d ordinarily not be impressed by this because if you get to pick the lag time for each region, it could just be coincidence that variations in streamflow have ups and downs at some point that match an arbitrary patten. But wouldn’t you expect the lag time to vary randomly if there is no correlation here? The abstract also seems to be claiming a pattern in the lag times, i.e. the lag time runs in one direction (increasing) along the ocean conveyor belt. That would seem a little too much to be coincidence.

Roger Knights

Leif wrote:
Nit pickers are my specialty 🙂
From AGU’s website:
The Fall Meeting is expected to draw a crowd of over 16,000 geophysicists from around the world. …

You beat me to the punch! I just realized that the phrase “Fall Meeting” was a title (indicated by the capitalization of “Meeting”), not just a descriptive phrase, and was rushing here to post a correction first.
(Well, anyway, at least readers here now know, if they hadn’t before, that in plain text the seasons aren’t capitalized.)

rbateman

It’s not a one-on-one correlation at all. It’s a mixture of a whole bunch of signals and events against a single one. It would be much more interesting to see separate signals compared against aa to see what goes along with it and what doesn’t.
What’s the significance of 34 years? A PDO tenure perhaps?

ztev

Does the Great Mississippi Flood of 1929 have anything to do with it.?

peat

A number of commenters see the decrease in river flow with increase in cosmic rays and presumably cloud formation (i.e. stemming from lower geomagnetic flux) as problematic. Why not an anti correlation with cosmic rays? Might cosmic rays primarily cause extra moisture depletion before air masses move from ocean regions to the interior of a continent? It seems that cosmic rays would best influence precipitation nearest the source of evaporation where the air is closest to saturation with water vapor. I would wager C. A. Perry has thought of this.
As an aside, I take issue with Leif’s charge of poor quality control by conference organizers by allowing such a poster. It is not uncommon to find factually incorrect ideas displayed in poster sessions. Posters do not receive the same peer review as journal articles. Their public display does not imply sanctioning by the conference committee. Posters can be admitted for the sake of generating interesting discussion. A presenter may use the forum as a way of testing the waters on a novel idea. Look what the poster is doing on this thread.

yonason

I don’t know, but my first impression is that this is strange.
From this link, that was given above, we read in the graph legend that “Streamflow has been lagged 34 years after the Geomagnetic data.” That and some of the other numbers given there indicate there may have been a lot of not-so-transparent parameter adjustment going on?
Also, shouldn’t it say “correlation” and not “connection?” in the graph title we see ‘above the fold’?
I do believe that cosmic rays have a lot to do with earth’s climate, but this looks like it’s pushing the envelop, to say the least. If this were the only evidence, I wouldn’t rush to buy.

Paul Vaughan

Note in response to comments about lags:
Say there is some underlying shared-(quasi)-periodicity, but there is one major event that seriously affects one-but-not-the-other of a pair of somewhat-related variables — this will introduce an (apparent) lag. The investigator has to discover the conditioning variable (the major event) to realize that the lag does not indicate a lack of relationship, but rather a conditional relationship.
I don’t have time for elaborate explanations, but here’s an image that was, for some time, a mystery:
http://www.sfu.ca/~plv/CCLR1LPPT1.png
It was part of a pattern of best-lags for precipitation at a site I was studying:
http://www.sfu.ca/~plv/BestLagMatrix.PNG
A clean relationship looks more like this:
http://www.sfu.ca/~plv/CCPxXTR.png
This^ clean image led me to discover the importance of this event:
http://www.sfu.ca/~plv/PolarMotionPeriodMorlet2piPower.PNG
I’m sure people around here have heard of the Dirty 30s. An event like that can “mess up” precipitation-linear-correlations with other variables.
Having worked on this sort of stuff for 2 years now (after changing fields once again), I’m not surprised to see how “stuck” climate science appears to be (in general). I’ve taught stats – and there weren’t many of my students that would figure this stuff out – and I don’t mean because of lack of intelligence, training, &/or resilience to plow through a series of obstacles – more importantly it’s a matter of having the intellectual, social, & economic freedom to pursue things that require prolonged, undivided attention.

tokyoboy

Marginally on topic, a question to Dr. Svalgaard:
The sunspot number is again down to zero these days.
What is its prospect for coming weeks or months?

Edouard

@Leif Svalgaard
1) Shouldn’t your error be corrected in your guest-post?
2) They say: “There appears to be a solar “fingerprint” that can be seen in hydroclimatic time series in other regions of the world, with each series having a unique lag time between the solar signal and the hydroclimatic response.”
If there is a similar correlation for many rivers, and the patterns of the variations of the streamflows are rather complex, is this not a proof for a link between the geomagnetic index aa (GI-AA), and the streamflow?
I have the same question for the exact correlation between the MWP, little ice age minima etc.. and the glacier advances in Europe and in the tropical Andes.
http://ff.org/centers/csspp/library/co2weekly/20060725/20060725_08.html
“Another important aspect of the Polissar et al. study is that it clearly implicates solar variability as the cause of the climatic variations they observed. They note, for example, that “four glacial advances occurred between AD 1250 and 1810, coincident with solar-activity minima,” and they state that the data they present “suggest that solar activity has exerted a strong influence on century-scale tropical climate variability during the late Holocene, modulating both precipitation and temperature” and demonstrating the “considerable sensitivity of tropical climate to small changes in radiative forcing from solar irradiance variability.””
or here http://www.co2science.org/subject/s/summaries/solartempsamer.php
My question is:
If all events occur at the same time, and if these changes have very complex pattern (LIA-minima) and last over a long time period (1000 ad until now), isn’t this a 100% proof that the solar fluxes did influence the climate?

ann riley

OT, I suppose, but the geomagnetic data appears to correlate just as well with war. There is a deep dip around the start of WWI, WWII, the Korean war and the Vietnam war. Surely a coincidence.

Neil Jones

The Word “connection” on the graph should be correlation.

yonason

Paul Vaughan (23:29:09) :
Just one last thing. Am I supposed to believe that from solar activity in (or for some years around) 1958, the great Mississippi flood of 1993 could have been predicted?
http://www.noaanews.noaa.gov/stories/s1125.htm
Sorry, but I’m not buying it. It’s way too Rube Goldberg.
Also, I’m not sure if this is what L.S is referring to, but increased solar activity results in a decrease in cosmic ray flux, which should result in less cloud cover, and hence less rain, not more. As this link clearly shows, there is an inverse correlation between solar flux and neutron detection.
http://www.leif.org/research/CosmicRayFlux.png

Nigel Calder

A terrible lack of quality control here, worthy of RealClimate. For goodness sake, why couldn’t Svalgaard find out what Charles Perry is really saying before wading in with comments on a graph he didn’t understand, and inviting all these other snide comments.
A place to start reading Perry’s story is http://ks.water.usgs.gov/waterdata/climate/

son of mulder

Early on the troughs of the geomagnetic index are lower than the troughs of river flow. More recently they are higher. So I’d posit no discernable connection.

Glenn

ann riley (23:56:15) :
“OT, I suppose, but the geomagnetic data appears to correlate just as well with war. There is a deep dip around the start of WWI, WWII, the Korean war and the Vietnam war. Surely a coincidence.”
Perhaps not entirely.
“War has historic links to global climate change”
http://www.newscientist.com/article/dn12936

jorgekafkazar

Leif Svalgaard (21:33:34) : “…This would make my complaints moot, but that conclusion was so far from my mind that it did not register as a possible one. What say all of you?”
Peak matching is always problematic, in my opinion. Over a short term, someone could probably correlate Mississippi flow with, say, the positions of Jupiter and Saturn…
Just kidding, Leif. Nobody would do that.

Rabe

“The Word “connection” on the graph should be correlation.”
No, no. They have connected two points of the curves and show that they aren’t correlated.
BTW, thanks for the reference to Dr. Boli.

Mike McMillan

Paul Vaughan (22:52:09) :
Acoustic info can manifest itself in interesting ways in time-integrated cross-correlation best-lag matrices. It’s not meaningless, but rather challenging to decipher (particularly when there are nearly-overlapping harmonic scales).

Acoustic from cosmic rays? You must mean that ping I hear when they bounce off my tinfoil hat.
.
I compared the Illinois corn crop failure years with the shifted river curve, and I get some bad years lining up, and other bad years totally opposite. Nothing that would be worth pasting together a chart. The good years, closer to 9 out of 10, seem impervious to the river or the cosmic rays. Yields sure do track CO2 nicely, though.
I think the GCR/river link is pretty thin, thinner than the Mississippi water, anyway, which is too thin to plow.

Stephen Wilde

In broad terms I see no problem with a periodicity in river flows worldwide from the effect of the 30 year or so phase shifts of the oceans and in particular PDO.
Those phase shifts move all the air circulation systems latitudinally and change regional rainfall distributions globally.
The next step, in linking that with solar variability is the real issue. I am torn between Leif’s certainty that the solar variations are too small and the historical correlations (not perfect admittedly) which I feel I have to give some weight to (but no idea how much).
Given that Leif himself has spotted the three cycle lag and acknowledged it’s potential significance and given that the narrative specifically refers to it I am prepared to attribute some weight to this study and others that I have seen, including a similar finding relating to South African waterflows.
I think tallbloke here has pointed out that the oceanic phase shifts seem to occur at every third solar cycle.
The isue is worthy of further investigation.