A Spurious Correlation between America’s margarine consumption and the divorce rates in the state of Maine. Credit: Tyler Vigen
Are you are a resident of Maine with an interest in dairy-free butter substitutes? Here’s something to keep in mind the next time you are at the grocery store: There is a remarkably strong correlation (99.2%) between the divorce rate in Maine and margarine consumption nation-wide. Are there any demonstrable or theoretical reasons why America’s margarine consumption should be driving Maine divorce rates or vice versa? No, almost certainly not. In fact, this chart was created by a bot whose sole purpose is to scan massive amounts of random data to find spurious correlations for a blog—aptly named Spurious Correlations.
Meaningless correlations exist all over the place if you look at enough data.
The point of the blog, created by Tyler Vigen, is not to find answers to America’s ever-rising divorce rates. Instead, the point is to show that meaningless correlations exist all over the place if you look at enough data. This process is as disparaged as data dredging: it describes scientists who approach a problem by looking for some pattern that they can then attempt to explain after the fact. Though many areas of science have engaged in this practice at different times, one field that is frequently accused of dredging is earthquake prediction.
The basic goal underlying earthquake prediction is to identify localized events, however subtle, that reliably occur before most large earthquakes. Pretty much everything under the sun has been considered—seismic activity, changes to the water table, changes to ground temperature, deformation patterns, and even abnormal animal activity. Often called the holy grail of seismology, earthquake prediction has been a remarkably unsuccessful adventure, thus far. Much like the Holy Grail itself, many scholars and government agencies (including the USGS) are now convinced that it is a fiction.
But there remains a small group of researchers who remain convinced that it is possible, and that the answer lies in a fairly counterintuitive place—space. This assertion, like the connection between Northeastern divorce rates and national margarine consumption, did not stem from a hypothesis or theoretical basis for why a connection should exist, it just came from looking at a ton of satellite data. This data has primarily taken the form of electromagnetic radiation readings—basic measurements performed by many satellites. Indeed, these data have occasionally shown some alluring patterns before major earthquakes.
The first such observation—discovered four years after the fact—was of a March 19th, 1979 magnitude 6.2 earthquake along the Mid-Indian Ridge deep at the base of the southern Indian Ocean. Eight hours before this unremarkable event, a Russian Satellite named Interkosmos-19 studying the electromagnetic waves surrounding Earth—our ionosphere—recorded a weird fluctuation in extremely low-frequency radio waves coming from the location roughly above this area.
There have been myriad other examples published in the literature. Perhaps most notably, a French satellite named DEMETER measured an increase in ultralow frequency radio waves in the month preceding the devastating 2010 Haiti earthquake, though no one recognized this until after the fact. In a 2009 review of numerous earthquake precursor studies, a group of seismologists concluded that “there are some provocative observations that suggest that the earth may well radiate EM energy at perhaps many different frequencies prior to the initiation of a strong earthquake.”
The aftermath of the 2010 Haiti earthquake in Port-au-Prince, that nation’s capital. Credit: United Nations Development Programme
Examples like these are tantalizing, especially given the potential for saving life and property. It is important to note, however, that while there is a general similarity to the observations in that they all come from the ionosphere, and they all involve electromagnetic radiation, there is still wide variability in the actual kind of disturbance recorded. Some studies record increased variability in radio waves (which includes both increases and decreases in certain frequencies), some record gradual rises in amplitude over months, and still others see sharp spikes just hours before.
One of the largest challenges in demonstrating the predictive effectiveness of these disturbances, however, is coming up with a plausible mechanism that links the crust buried deep in the earth with the diffuse collection of ions surrounding our planet in space. Koh-Ichiro Oyama, a senior research fellow at the International Center for Space Weather Study and Education at Japan’s Kyushu University, has published extensively on ionospheric earthquake precursors; he summarized the current thinking on this problem in a paper published this year, broadly pointing to three basic theoretical models.
The first, he told NOW.SPACE is that compressed rock under great strain, through the liberation of electrons from mineral pores, can induce an electric current. In theory, the current could create a closed loop between the ionosphere, the atmosphere, and the lithosphere, and disturb the electrical field in a measurable way, though it is unclear if such a current could be strong enough to be measured in the manner suggested by some studies.
The location of the ionosphere in relation to Earth’s atmosphere as a whole.
“There has been, you can say, quite a bit of quackery.”
The other two models point to gravity waves—essentially physical waves generated below or on the surface of the earth. These waves propagate (like an ocean wave) through the atmosphere to the point where they physically disturb the ions in the ionosphere, creating measurable changes in the electric field. Two positive aspects of these explanations, Oyama said, is that they are consistent with the low-frequency disturbances many papers describe (because higher frequency waves wouldn’t make it through the atmosphere), and that they provide a mechanism that could theoretically amplify the intensity of the gravity wave once it interacts with the electrons in the ionosphere. This would create readings and fluctuations similar to those already documented.
There are two possible (and not mutually exclusive) models of how these low-frequency waves might be generated. The first implicates radioactive radon released from compressed minerals. The radon ions would then kick off a chain of events that is ultimately recorded in the ionosphere. The second model, which Oyama favors, is that there are subtle movements not detected seismically that physically push the atmosphere at nearly undetectably long frequencies (hours to days), and that this signal is amplified in the ionosphere.
These ideas may sound outlandish, but it is undisputed that the waves generated from actual earthquakes can, and have, created detectable low-frequency waves measured from satellites. The disputed element is whether or not there is any reason for similar activity before the quake itself.
This is a question, Oyama argues, that we do not have enough data to answer. He said it’s a challenge to get precursor research published because there is little collaboration and “because this field is still new and no standard way to attack this problem [has been] established.” Funding is also an issue, he said. Without a clear idea of how to approach ionospheric precursors, scientists have no way to evaluate a funding proposal. To make matters worse, many scientists reviewing proposals remain deeply skeptical about this area of research.
Speaking to the online publication, Quartz, this past April, Richard Luckett of the British Geological Survey said that earthquake prediction has become “something of a dirty phrase among seismologists,” adding that “there has been, you can say, quite a bit of quackery.” Critics of funding this work argue that the evidence presented on the topic is driven by hindsight, data-dredging, and questionable statistics. In many cases, critics say, these reports fails to look for—or even mention—the same signals it uses as predictive tools popping up in non-earthquake areas and times.
A scientist (for scale) holding the casing for a CubeSat. Credit: University College London Faculty of Mathematical & Physical Sciences
But numerous recent studies, including two published in the past six months, have argued that small CubeSats built with simple electromagnetic measuring tools would be a relatively inexpensive way to get the data needed to do this kind of science correctly and solve the controversy once and for all. Funded by the Mitsubishi Foundation, Oyama formed the “Ionosphere Precursor Study Task Group” and, in a March 2016 study, argued that a constellation of microsatellites would be instrumental for monitoring activity above Japan. The benefit to a fleet is that they could cover more area and triangulate signals to better remove noise and to pinpoint an epicenter.
Another paper, this one by Valery Korepanov of the Lviv Centre of the Institute for Space Research in Ukraine in the journal Astronautica Astra, recommended a fleet of at least two CubeSats. A similar mission, named TwinSat, was proposed by both British and Russian researchers in 2012 and again in 2014 but was denied funding both times.
One argument is that funding this kind of mission lends undeserved credence to the notion that earthquakes are predictable in the first place, which could serve to detract from precautionary measures and other research efforts aimed at reducing death and property loss from quakes. The argument in favor, of course, is we won’t know for sure until we have solid data and a solid plan to really analyze this program.
Massive amounts of data would certainly be a boon for researchers interested in demonstrating, or even discrediting, a connection between earthquakes and the ionosphere. Such a link could save hundreds of thousands of lives. If and when we get to look at this CubeSat data, however, let’s keep in mind that massive amounts of data also link national margarine consumption with divorce rates in Vacation Land.
Nobody, to my knowledge, believes that maintaining access (or lack of access) to margarine in America is the key to a happy Maine marriage.