Guest Post by Willis Eschenbach
I came across an article with a provocative headline, “Scientists Link Cosmic Radiation to Earthquakes for the First Time“.
So I went to look at the underlying study, Observation of large scale precursor correlations between cosmic rays and earthquakes with a periodicity similar to the solar cycle by P. Homola and 43 others.
Hmmm, sez I. Willis’s Rule of Authors says that the strength of a scientific study is inversely proportional to the number of authors … but I digress. The Abstract says:
The search for correlations between secondary cosmic ray detection rates and seismic effects has long been a subject of investigation motivated by the hope of identifying a new precursor type that could feed a global early warning system against earthquakes.
Here we show for the first time that the average variation of the cosmic ray detection rates correlates with the global seismic activity to be observed with a time lag of approximately two weeks, and that the significance of the effect varies with a periodicity resembling the undecenal solar cycle, with a shift in phase of around three years, exceeding 6 σ at local maxima.
The precursor characteristics of the observed correlations point to a pioneer perspective of an early warning system against earthquakes.
The study says:
Here we report on an observation of the correlations between variation of the average rates of secondary cosmic ray fluxes measured locally and global seismic activity, and we also point to the periodicity of these correlations (or their observability) which corresponds to sunspot number observations back to the 1960s.
They describe their method as:
… an alternative, novel approach on which we report here: comparing the absolute average variabilities of secondary cosmic radiation to the average global sum of earthquake magnitudes.
And their conclusion was:
We have demonstrated for the first time that the variation of the absolute average detection rates of secondary cosmic radiation correlates with the global seismic situation (sum of the magnitudes of earthquakes with magnitudes greater-than or equal to 4, occurring at all locations) that takes place approximately two weeks later than the relevant cosmic ray data. The size of the shift in time between the cosmic and seismic data sets reveals the precursor character of the correlation effect.
So to see if this made sense, I went and got the data for earthquakes, cosmic rays, and sunspots. You can’t do a CEEMD or a Fourier analysis unless the data is regularly spaced in time or space, which earthquake data isn’t. So I used a variation of Fourier analysis that I invented a few years back, only to find out that I wasn’t the first to invent it. Tamino informed me that “The method you describe is very clever. It’s also known (in the astronomical literature at least) as the Date-Compensated Discrete Fourier Transform, or DCDFT (Ferraz-Mello, S. 1981, Astron. J., 86, 619).” Fair enough, it proves that I understand Fourier transforms … the main advantage of the method is that it doesn’t require that the data be regularly spaced.
My logic was that if the magnitudes of earthquakes are related to sunspots and cosmic rays as the 44 authors claim, we should see some kind of a significant ~11-year cycle in the earthquake data. Here’s the result of my analysis.
Figure 1. Fourier Periodograms of monthly sunspots, earthquake average magnitudes, and cosmic rays. The three longest records of cosmic rays are from Thule, Greenland; Jungfraujoch, Switzerland; and Newark, NJ USA. In addition to the earthquakes of magnitude greater than 4 studied by the authors, I also looked at earthquakes of magnitude greater than 7.
From the periodogram, it’s obvious that there are no significant cycles in earthquakes at around 11 years. You can see that the cosmic rays are closely connected to sunspots, for well-understood physical reasons.
But there’s no such connection between sunspots or cosmic rays and earthquakes.
Does this complete lack of any significant ~11-year cycles in earthquakes show that the 44 authors are wrong?
Nope. Maybe they have discovered the secret sauce for connecting cosmic rays and earthquakes.
But it sure makes a conventional explanation of their claim much less likely …
My very best wishes to you all,
w.
PS: Yeah, you’ve heard it before, you’ll hear it again. To cut down the number of misunderstandings, when you comment PLEASE quote the exact words you are discussing.
Sure they do. There are always cosmic rays before earthquakes… and after earthquakes.
Galactic radiation is the most sensitive indicator of the Sun’s magnetic activity and the strength of the solar wind’s magnetic field. As the chart below shows, over the past year only once has the solar wind speed reached 800 km/s.
Cosmic rays have a known effect on ionizing atoms in the atmosphere. All the rest is pure speculation, as unsupported by evidence.
Early during the Holocene, there was a massive star in the Vela constellation, just a mere 800 light-years away that went supernova. What happened here on Earth 800 years later? Nothing. A huge increase in 14C, the biggest in the entire Holocene, and nothing else happened. No noticeable climate change, no noticeable increase in volcanic activity, NOTHING. If there was an increase in cancer frequency among early Holocene populations we would not know, as no statistics were being kept at the time.
It seems to me that a “sanity check” on their process would be to look in BOTH directions, and see if earthquakes predict cosmic rays. That seems to me to be an equally important. Note that “equally important” doesn’t infer important.
Willis is right, it’s junk science.
Y-axis in graph shows positive integers. Shouldn’t the inversion of cosmic ray data be noted somewhere?
In the page 11 in this Journal is article called Links of Terrestrial Volcanic Eruptions to Solar Activity and Solar Magnetic Field
it explains current Knowledge and Studies on this matter.
The article says:
First off, Where are they getting information on “eigenvectors (EVs) of the solar background magnetic field (SBMF)” all the way back to 1762?
And if a purported signal “disappears” for a hundred years, under what interpretation can it be considered a signal?
I analyzed volcanoes and sunspots a while back. See my post on the subject.
Regards,
w.
A sudden geomagnetic storm has occurred. Such a disturbance some time from now, when the solar wind activity drops strongly, can be a harbinger of a strong earthquake.
A strong drop in solar wind speed is predicted for early July.
Willis, I suggest you write this up as a letter to the editor of Journal of Atmospheric and Solar-Terrestrial Physics.
The solar wind causes earthquakes, where a high speed windstream hits the earth is where the earthquake will happen if there is a fault at that location.
It seems logical that periods of low solar allow notable CR events to penetrate into the Earth, resulting in a measurable EQ response, with a time offset. It also seems that the anti-CR/EQ zealots poke fun at recent efforts in an attempt to silence debate and hinder further research. Please note:
Cosmic Ray Extremely Distributed Observatory (CREDO) (Homola et al., 2020) – a recent cosmic ray initiative dedicated mostly to the global search for large scale cosmic ray correlations and associated inter-domain efforts, e.g. those related to the joint research program of the astroparticle physics and geophysics communities (Workshop on Observatory Synergies for, 2019).”
Hello Willis,
Thank you for creating the figure, I like it! I only do not agree with your conclusion:
The claim is that there is a correlation, not causal relation, and this is, I hope, made very clear in the article, the official press release, and even in the Newsweek article you quote. The author of the latter even asked me explicitly whether it would be fair to say that either more or less cosmic radiation might cause earthquakes, and I said “no”, which can be read in the news.
Now the good thing about your figure is that it strengthens the argument about NO causal relation between cosmic radiation and earthquakes, which makes the observed effect even more intriguing! I fully agree that if cosmic radiation, be it solar or galactic emission, triggers earthquakes, there should be an ~11 year cycle in both cosmic and seismic data – and then one could expect a more or less continuous correlation – but it’s not the case! The cosmo-seismic correlation we report on appears only occasionally (see Figs. 2 and 4 in the article), with an average distance between the sharp significance maxima being 10.2 years. OK, so if there is no casual relation between cosmic rays and earthquakes then what is behind the multiply cross-checked correlation? This is the most intriguing question for now as it points to a possible third factor which might modulate both seismic and cosmic effects, making them appear as causally connected. What I’d suggest then is that you consider a small edition of your very nice post: “makes their claim” -> “makes a conventional explanation of their claim” 🙂
Furthermore, if you or any of your readers are interested in discussing details I’ll be available today (6.07.2023, 5 p.m. CEST, 3 p.m. GMT) at an online fireside chat at Photonics Club, (see the invitation link below), and I think I’ll present your figure too. You are welcome to join and bring your friends!
Many thanks once again for your careful reading and for a useful piece of analysis!
Best regards,
Piotr Homola
(the corresponding author of the cosmo-seismic article)
Quantum Photonics FB post iviting to the meeting:
https://www.facebook.com/plugins/post.php?href=https%3A%2F%2Fwww.facebook.com%2Fcecile.tamura%2Fposts%2Fpfbid0AjuWDWuVGMmdwLsp79ZhB6e9KeqwwucF5wKQpa4DaBmf2PuwVrYjQmopdG8neoLDl
Invitation with the link to the Clubhouse app:
https://l.facebook.com/l.php?u=https%3A%2F%2Fwww.clubhouse.com%2Finvite%2FN17IAkDglVOJGbZa96D62nElJwoDInr286j%3AG0ni2qrbmfG18pT1LDPtvCULOhfWsWWYqo3P-An7PAA%3Ffbclid%3DIwAR0HmTQrc28ioWhHN-BZUUjGim7mXqSzbYFaX2X1fw0uHwy-AFj_-iyIy9g&h=AT16d4wDtULA4LXbdKNga_AvbKTChGmA38Q-7Ky3rQ2fprRASg-kpY01c8-rwiT1Fwqs1x4fZlGaVX8gmHwU9T5tImUTixMp9VRPmU_HExhCa-pT_RdNydnJ8nx44bY1GoK2mAvyhgz9&__tn__=R*F&c%5B0%5D=AT2-EjknyOA4NunGds5kwVl9JbFqM2cKlwneFYKNqyo_JiTd-tRStiFLTZogv6hYX2au_8c3m052SZwhcaQ9lXlGbfeIBjzi5FBigOpBpSR0xUF1m4JUeUJl2nC4om6gBbWUNNWqkYl_whh6MXBqp1dKAMqOPizqJg9uUADPKJsKOBxk_vqIs9t-A1zHueTuDNLLWst2CTNioOZ2YRNFyKdIUOi7QmsDkYa8
======================================
Thanks, Pyotr. I fear commenting is closed, so I’m taking this way to answer.
You say:
Done, thanks.
You also say:
First, I am always VERY suspicious of correlations that “appear only occasionally”. This is particularly true when we’re looking at a correlation between a highly cyclical phenomenon (say cosmic rays) and an apparently random phenomenon (earthquakes). If the underlying “random” phenomenon has any inherent cycles, and Joe Fourier says it does, you’ll have a beat frequency against the ~11-year cosmic ray cycle, and you will assuredly have correlations that “appears only occasionally”.
My rules of thumb are that I don’t get interested until I see three cycles in a row, and I won’t claim a real cycle exists until I can see five cycles.
Finally, consider the Bonferroni correction. If you throw 5 coins once, the odds of them coming up all heads is 1 in 128. Looks statistically significant at p<0.05.
But if you throw five coins up ten times in a row, the odds of getting 5 heads is now 0.08, no longer significant at p<0.05.
The same thing is true if you look for a given phenomenon in a bunch of places. Like say looking for earthquakes near a bunch of cosmic ray counting centers …
To adjust for that, the Bonferroni Correction says that if you look for an event in N places, then to assign a true p-value of X, you need to find a result with a p-value of X/N. It’s an approximation of the actual calculation, but a very, very good one.
So if you look at say 10 cosmic ray centers to find a correlation with nearby quakes, and you’re looking for p<0.05, you need to find an even that has a p-value of 0.05/10 = 0.005 …
My very best to you and yours, and thanks for coming to defend your work. It’s sadly rare these days.
w.