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).
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:
- 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)
- The lack of internal quality control by USGS
- 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:
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.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Could this become a peer-reviewed paper and an additional source for their funding? I’m dying of curiosity.
This might just be 2 data sets that superficially appear to have a correlation.
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, ___
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).
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/
“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.
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.)
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?
Does the Great Mississippi Flood of 1929 have anything to do with it.?
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.
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.
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.
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?
@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?
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.
The Word “connection” on the graph should be correlation.
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
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/
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.
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
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.
“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.
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.
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.