Episode 50: Lunatic Earthquakes: Do Tides Cause Quakes?
Recap: It would seem that if lunar tides cause or trigger earthquakes that everyone would know about it and beware of doom during the supermoon, but according to a few people, a vast conspiracy by Some has kept this free and easy-to-use earthquake predictor out of the general knowledge. Find out what the statistics actually show.
There was no puzzler for episode #49.
Puzzler: There were two puzzlers this episode. First, while the Moon causes tides on Earth, there will also be tides on the Moon. When are tides largest on the Moon? Same time their largest on Earth, or at a different time -- why? Second, we don't think that Mars has any plate tectonics. Would it experience quakes, and if so, what could cause them? Keep in mind for both of these that there may be more than one answer.
Q&A: This episode's question comes from a guy on the blog who I'm just going to say the first initial is J, 'cause that's easier for my lousy American pronunciation. J asked a follow-up from my episodes on image processing and the Q&A from Belgarath where I talked about white balancing photos taken from the Martian surface. J asked how we actually KNOW what the white balance is, what the colors "should" be, because stuff should get deposited on any color card included on the craft almost instantly.
The answer is two-fold. One way is that yes, they do have effectively color calibration cards on the craft that they can photograph to figure out what the colors should be. The deposition rate is fairly slow on Mars, in fact it's been measured based on how quickly the solar-powered landers lose power, and so early on in the missions they can use the cards to approximate what the colors would be on Earth.
The other method is that there is extensive calibration that goes on on Earth before the craft is ever launched. How the cameras respond, sample photos, the detailed response of EVERY filter, etc. All of those are known well before the spacecraft is ever launched, and so at a bare minimum, those can be used in order to estimate what the colors would be based on balancing the filter responses from the tests on Earth.
- Audio Source: Coast to Coast AM from October 19, 2005, Hour 2
- Audio Source: Coast to Coast AM from November 14, 2011, Hour 1
- Additional Resources
- Wikipedia on Earthquake Magnitude, the İzmit (Turkey) Earthquake of 1999
- Jim Berkland's "Syzygy Job" Website
- Southern California Earthquake Data Center
- My Statistical Data Writeup
- Dialog with David Nabhan (lunatic earthquakes) on the CoastGab Forum
- Logical Fallacies / Critical Thinking Terms addressed in this episode: Non sequiturs, argument from persecution, argument from authority, argument against authority, argument from personal credulity, misuse of statistics, correlation = causation, remembering the hits while forgetting the misses, and data mining.
- Relevant Posts on my "Exposing PseudoAstronomy" Blog
- Meetup: Sometime Between September 18 and 23 in Flagstaff, AZ -- Send E-Mail if Interested
Claim: The idea or claim in this case is that there are a handful of folks who state very emphatically that earthquakes can be predicted based on where Earth's Moon is in its orbit. On its face, this seems like it could make sense, after all, the Moon causes tides and stabilizes our obliquity, so it has a noticeable effect on many things, so why not earthquakes. Similarly, if it were true, it would seem like it should be obvious and so along with the claim there is oftentimes a vast conspiracy proposed by its proponents.
There are a lot of ways that I can talk about this topic, and I've written it in a way that I think flows best, but it may seem a bit disjoint. I'm also going to talk about statistics to some extent - yes, more math - and I've prepared a document that I've linked to in the show notes with the data, sources, and statistics and graphs that I'll be referring to.
Background Information: Prediction Windows, Earthquake Frequency
First, I'm going to talk about some background information so we're all hopefully on the same page with this. I'll get into some very specific predictions and prediction windows later on, but throughout this background information, it's important to keep in mind that proponents of this idea will often use prediction windows for earthquakes, saying that one is going to happen, within about a one-week timespan. Sometimes their window is a bit longer, around 10 days.
The next, brief bit of background information is how frequently earthquakes occur, and what a "magnitude" means. Earthquake magnitude is a measurement of how much energy is released during the quake. As with many scales in science, it's a logarithmic scale, meaning that each increase in the linear scale is an exponential increase in energy.
So a magnitude 1 earthquake is equivalent to about 500 grams of TNT and is considered a microquake. Magnitude 1 quakes happen continually.
When you reach magnitude 2 quakes, it's a minor earthquake and usually people can't feel them. Over 1 MILLION of these happen per year.
Magnitude 3 quakes are still minor. You can usually feel them if you're close to the epicenter, but they VERY rarely cause damage. An estimated 130,000 happen every year, or about 15 per hour, one every 4 minutes. It's equivalent to about 15 kg of TNT.
Magnitude 4 quakes are equivalent to around 15 metric tons of TNT and are considered light quakes. You'll hear shaking of indoor items, but rarely is there any significant damage. Around 13,000 of these happen per year, or about 1 every 45 minutes. This is also the kind of quake that some earthquake predictors will claim to start to predict. One that happens on average 36 times per day.
When you reach magnitude 5 quakes, instead of 15 tons of TNT at a 4.0, a 5.0 is around 480 tons of TNT. As I said, this is a logarithmic scale of energy. A 5.0 is considered a moderate quake and can cause major damage to weak buildings, slight damage to buildings that are well designed. An average of 1319 of these happen per year, or about 25 per week, one every 7ish hours.
Magnitude 6 quakes are considered strong and can be destructive across populated areas within 100 miles (150 km) from the epicenter. It's equivalent to 15 kilotons of TNT and around 135 happen per year, or about 1 every 65 hours -- 2.6 per week. So if I say that I predict that a 6.0 quake is going to strike somewhere in the world next week, I would actually be UNDER-predicting.
By 7.0, we're talking about major earthquakes that will cause serious damage. Only about 15 of these happen per year, or about once every 24 days.
8.0 magnitude quakes will cause damage over several hundred miles or kilometers, releasing the energy of 15 megatons of TNT. On average, one of these happens PER YEAR. The largest nuclear bomb ever exploded on Earth was the Tsar Bomba, 50 megatons of TNT that would be an 8.35 magnitude quake.
A 9.0 is stronger than the Krakatoa volcanic eruption in 1883 and releases the energy of 500 megatons of TNT. These quakes are much more stochastic and only a few have happened in recorded history. It's estimated these happen every 1 to 10 years on Earth.
A 10.0 quake has never been recorded and would release the energy of 15 gigatons of TNT. It's around the theoretical maximum that you can get on Earth from an earthquake, though something like the asteroid impact that created the Chixulub impact crater thought to have caused the death of the dinosaurs would be around 100 teratons of TNT, equivalent to a 12.55 magnitude quake.
So what's the point of going through this? The point is to emphasize that even though earthquakes release an enormous amount of energy, and they can cause a lot of damage even from something seemingly "minor" like a magnitude 6 quake, these still happen frequently. If you ever hear someone claim to predict, such as Cal Orey, that, "An earthquake will hit Italy in 2011," that prediction is meaningless because magnitude 1 quakes are happening all over all the time. Similarly, anyone who claims to predict a California earthquake of 4.0 magnitude or larger should know that around 40 of those happen per year, about once every 9 days just in California.
Background Information: Lunar Orbit and Tides
Getting into the lunar data we need for this episode, the Moon orbits Earth roughly once every 28 days, or 4 weeks. In reality, the time is 29.531 days, but 4 weeks is a good enough approximation for this discussion. It has a new phase once during that time, and it has a full phase once during that time. That's because TWICE during this four-week period, it is aligned with Earth and the Sun, the new moon being when it's Sun-Moon-Earth, and the full moon being when it's Sun-Earth-Moon. This alignment of three celestial objects in a row is called "syzygy."
It's during those times that we get particularly strong tides, called Spring Tides, because of the combined tidal forces of the Sun and Moon.
There are also two times during the Moon's orbit that the Moon appears to be half-full, and we call those first quarter and third quarter moons. That's because they happen 1/4 and 3/4 of the way through the four-week lunar period. During those times, the Sun-Earth-Moon alignment forms a right angle, and the tides from the sun, which are about 45% of the strength of the lunar ones, act to suppress the lunar tides and so the overall tides are particularly weak. We call those neap tides.
Remember, these phases happen one week apart -- one week you'll have a new moon, then just over 7 days later a first-quarter, then just over 7 days later a full moon, then just over 7 days later a 3/4 moon, then just over 7 days later a new moon again. Keep that in mind when you hear people claiming one-week windows for their earthquake predictions, and I'll get into it more later in the episode.
An added complication to this model is that the Moon is on an elliptical orbit around Earth. When it's closest, it's called perigee, and when it's farthest, it's called apogee. The difference in distance is around 43,000 km, or roughly 11.8%.
I said that you get one apogee and one perigee each during a lunar month. They're actually not aligned quite that well. The average time between apogee is about 27.6 days, 2 days less than the time between new moons. The average time between perigee is around 27 days, but it can vary between about 24.5 and 29 days, complicating any models.
Tides, however, don't scale directly with distance. Gravitational force goes as the inverse-square of distance while tidal force, because it's the DIFFERENTIAL of gravity experienced on one side of an object versus another, goes as the inverse-cube of distance.
This means that when the Moon is closest to Earth, the tidal force is about 40% more than when it's at apogee. For the Sun, the difference is about 11% when we're at perihelion - our closest approach to the sun - versus aphelion.
Basic Model Proposed
And so, the very basic model proposed by people is that earthquakes happen on the planet. They're not necessarily CAUSED by tidal forces, but if a fault line was "about to go," then a strong tide will trigger it.
Hence, they claim that when we are near syzygy - when it's either a new or full moon - AND the moon is at perigee, then the combined effects of the closer moon and the added solar tides will trigger more earthquakes.
Since it's a tidal thing, other planets' positions generally don't factor into this at all.
This seems like it's a testable hypothesis, and it is. And I downloaded data for over 43,000 earthquakes spanning the past 80 years to test this hypothesis, but before I talk about that, I'm going to talk about a few examples claimed, and some of the conspiracies claimed.
[Coast to Coast AM clip, October 19, 2005, Hour 2, starting at 0:48]: "Geologist Jim Berkland was suspended from a California government geology job when he made a prediction that a major earthquake would occur during the 1989 World Series during in the Oakland Bay area. It hit. The government told him not to make anymore predictions. Well, now that he's retired, he publicly states 'quake windows.' Mr. Berkland uses tidal flooding tables based on lunar perigee - that's the time when the moon is close to the Earth to effect [sic] more gravitational pull on the Earth."
Jim Berkland is probably one of the biggest names out there in non-psychic earthquake predictions. He runs a website called "Syzygy Job" which hasn't been updated in two years, and is a very frequent news guest on Coast to Coast AM. I listened to 10 hours - just a small fraction of what I have of him in preparation for this episode.
Jim's method is exactly what I explained in the basic model, so I thought that it would be informative to take a look at some of his past predictions.
From his July 1997 newsletter, he wrote the following predictions would take place during his one-week window of July 19-26, 1997:
(1) An earthquake of 3.5-5.5M within 140 miles of San Jose, CA; (2) A similar event within 140 miles of Los Angeles; (3) A similar event within 140 miles of Seattle; and (4) A major quake of 7+ magnitude somewhere, probably within the Pacific Ring of Fire.
So how did he do? First, we need to look at the background levels. I don't have data for earthquakes that small for Washington State, so I'm only going to look at his San José, Los Angeles, and fourth prediction. And remember that a 3.5-magnitude quake is barely detectable to a normal person.
A 140-mile radius is fairly large, in fact Los Angeles and San José are about twice that distance apart, so these circles just overlap each other. San José is in the north part of California, and it, on average over the past 80 years, experiences 8-9 of those kinds of quakes per year, giving him a 16% chance just based on dumb luck. Los Angeles, near the middle of the state, gets a lot more quakes and experiences 30 3.5-5.5 quakes per year, on average, or 1 every 12 days, so he has a 60% chance just of randomly getting this right. Meanwhile, as I said in the background info, a 7-magnitude quake goes off about once every 24 days, so if he's quoting a one-week window, he has a 30% chance of being right.
As it turned out during that week-long window in July 1997, there was a 3.73 magnitude quake 131.2 miles south of San José on the 24th, and a 3.73 quake about 53 miles east of Lost Angeles on the 26th. There was NOT a 7+ quake anywhere in the world, the most intense being a 6.3 in the middle of Mexico, not near the ring of fire.
So he got an easy hit, a bit of a harder hit, and missed the intermediate probability one.
In Coast to Coast episodes, most of the major quakes that he predicts happening don't, but he makes light of significant ones in the past that he predicted. One example is that bio that I started out with. George reads that bio every time, an obvious indication that Jim is using that as a major example of something he got right.
But, in a 2005 episode, Jim predicted that the March 29, 2006 total solar eclipse visible across Africa and the middle East would cause an earthquake over the land that was in the eclipse path. He said this happened in 1999 and that only 6 days after a total solar eclipse in Turkey, it experienced an earthquake.
So there are two claims there, the first being a prediction, and the second being an example of his claims from previous events. The prediction ended up not being true, with three 6-magnitude quakes happening within a week of that eclipse, and none within the eclipse path. Remember that a 6-magnitude quake happens on average about once every 2.5 days.
The second claim is that the solar eclipse triggered the Turkey quake. The Turkey quake of 1999 was a 7.4-magnitude earthquake, but it happened, as Jim said 6 days after the eclipse. In other words, on August 17, 1999, during a first quarter moon just two days before the moon was at its FARTHEST point from Earth ... the exact OPPOSITE of when it should have hit based on Jim's model.
Correlation ≠ Causation, and Remembering the Hits, Forgetting the Misses
There are many more examples of Jim's failed or overly broad predictions, but at this point I think some side discussions are in order. First, is correlation does not imply causation. All because one example or even a few examples you have of a prediction that came true based on your model, that does not mean your model is correct. The two events could be correlated, but you need to examine a lot of data and do a lot more studies to determine if the correlation actually IS due to causation.
Which brings us to something that skeptics encounter all the time when looking at predictions - whether claimed to come from psychics or some other method: Remembering the hits and forgetting the misses. Jim had a good hit in 1989. He's been riding on that pretty much ever since and has made high-probability predictions since that time and gone out on a limb on a few others.
The low-probability ones are often missed, because they're low probability. He never talks about those after the fact. But, if he ever gets a low-probability hit again, you'll never hear the end of it.
If one were to point this out to these folks, a likely reaction would be to accuse you of being in on the conspiracy: [Coast to Coast AM, November 14, 2011, Hour 1, starting 12:11]
That was David Nabhan and while he gets a bit more emotional than Jim Berkland, it's a good example of the conspiracy that many of these folks revert to: Because their method is so obvious and foolproof, there must be a vast conspiracy of silence that prevents it from getting out but they are the Lone Candle in the Dark to get the Word of Truth out there. Like what happened with Galileo.
As opposed to a lot of the stuff I've talked about on this podcast before that deals with conspiracies, this is not a NASA one, but it's blamed on the United States Geologic Survey. Which in itself is interesting and shows how fallacious the conspiracy claim is. The USGS didn't get started until 1879, but the alleged conspiracy by it goes farther back by decades. It's a lot like people who claim that Thomas Jefferson didn't believe in evolution therefore evolution is false ... well, evolutionary theory wasn't formulated until over 30 years after Jefferson died.
But beyond that, it also gives enormous power not only for the USGS to censure scientists before USGS was even founded, but also those all over the world, over whom USGS has no apparent power. For example, to believe this conspiracy, one would have to believe that an enterprising geologist in, say, Japan where they get some of the worst earthquakes in the world, is somehow controlled by the USGS, and the ancient Japanese, as well, were somehow also prevented from figuring this out before the United States was even a country.
Argument from NOT Antiquity
That brings us to what I will term an Argument from NOT Antiquity: The fact that this very very basic observation of higher tides triggering earthquakes was NOT figured out centuries ago despite other very basic things related to the Moon-Earth interaction, is another argument against an obvious link between tides and quakes.
True, it's not iron-clad evidence, but if it's so obvious today, and ancient people easily could watch tides and measure the lunar phases and estimate when it was bigger or smaller, they should have been able to figure this out.
Argument from Persecution and Non Sequiturs
But, David Nabhan is quite kind and provides a host of other logical fallacies that I'd like to mention. After all, one of my favorite segments on SGU is the rare Name that Logical Fallacy, so when I get a chance here, I try to point them out: [Coast to Coast AM, November 14, 2011, Hour 1, starting 37:39]: "The whole idea that earthquake prediction is impossible, that's gotta be ditched right now. It's absurd on the face of it. We've measured the background radiation of the Big Bang from 14 years -- 14 billion years ago, that's possible but lonely little earthquake prediction isn't?! We've cataloged the whole genome of the human race, 3.5 billion combinations. That wasn't impossible. Well, who said earthquake prediction was impossible by the way? Albert Einstein? Isaac Newton? I've looked, I can't find it. A few goofy people at USGS have said it, and then they've latched onto it from 100 years ago and that's been the orthodoxy, the Byzantine orthodoxy that we're sluggishly dragging along into the 21st century."
How many can you find in that 48-second clip? I found three main ones.
First, argument from personal credulity. Effectively, "I believe it, therefore it's absurd not to." Okay, I'm not sure if that's a formal fallacy, but I'm sticking it in there, anyway, as the opposite of Argument from Personal Incredulity.
Second is the Non Sequitur, meaning "doesn't follow." As in, we can measure the cosmic microwave background, we can map the human genome, therefore, somehow, these have something to do with predicting earthquakes? One could just as well put anything else there because that whole argument is meaningless. Finding the CMB has nothing to do with predicting earthquakes, and mapping the genome also has nothing to do with predicting earthquakes.
Third is argument from authority. He's trying to say that because Einstein and Newton didn't say that predicting earthquakes was impossible, then it IS possible and USGS is blocking it all from coming to light.
What he apparently fails to realize is that anyone who figures out how to predict earthquakes has their funding set for life. If scientists who work with the USGS figured it out, the'd get some huge government funding for the next few centuries to implement whatever system they come up with and put out bulletins of their predictions.
Testing the Hypothesis
So let's actually look at the data. Independent of any one person's prediction windows or one particular prediction or set of predictions or retrodictions or predictions, let's look at the data.
I downloaded, from the Southern California Earthquake Center's website, the times, locations, and magnitudes of over 43,000 earthquakes since 1932. I then modified a Java script to output the dates and times of new and full, apogee and perigee moons during that time. I then spent many hours, including some time getting help on a statistics forum, going through earthquake, and now's when I share the results. And you can find a writeup on this within a day or so in the shownotes for the episode.
Remember that the basic idea is that stronger tides trigger earthquakes. You get stronger tides when we're close to the sun called perihelion, when the moon is closest to us called perigee, and when we have syzygy - when it's either a new or full moon.
So one thing we can look for is whether there are increased earthquakes when we're closest to the sun, and decreased ones when we're farther away. We're closest to the Sun in January and farthest in July. The data do not show ANY trend with month, so we can rule that out, despite the solar tides being about 10% stronger in January.
But, perhaps the sun is too subtle, so we can just look at the moon's effect, if any.
First, I wanted to make sure that my code was working properly, and to get baseline statistics if everything really were random. So I created 1 MILLION earthquakes that happened at random times over the past 80 years and 8 months.
The analysis I then performed was to determine when earthquakes occur relative to a perigee moon, apogee moon, new moon, and full moon. I binned the data in different intervals just to make sure that didn't affect anything, but I settled on intervals of half-days for the random trial which is called a Monte Carlo simulation. The result was a completely even distribution of times of earthquakes relative to these four kinds of moons out to about 14 days on either side. As one would expect.
I then determined the fraction of earthquakes relative to the whole that occurred within BOTH perigee and syzygy, so perigee and new or perigee and full moons. For this, I also looked in multiples of half-days. From basic combinatoric probability, you would expect that when you require that an earthquake happen at exactly the time of a new or full moon AND the time of a perigee moon, you'll get effectively zero, even from a large sample, because that represents 1 minute over the course of several months. You'd also expect that when you require it happen within one week of both perigee and a new moon, you'd get roughly a 25% chance of that happening, and perigee and a full moon would also have a 25% chance of happening.
That's because you have about a 50% chance of an earthquake happening within 1 week of a new moon, of 50% chance of an earthquake happening within 1 week of a full moon. And, the chance that perigee happens within 1 week of a new moon is also about 50%. The chance that it happens within 1 week of a full moon is also about 50%. So, multiply 0.5 by 0.5 and you get 0.25, or a 25% chance that an earthquake will happen within one week of perigee and a new moon, and 25% chance that an earthquake will happen within one week of perigee and a full moon.
Based on the Monte Carlo simulation with 1 million earthquakes, the probability was 24.16±0.17%. The reason that it's not 25% exactly is because the interval between new and full moons changes, over the years with a range of about half a day, but the time between perigee moons changes by up to 4 days depending on everything else going on in the solar system, and it's not normally distributed. So you actually can't use basic algebraic statistics to figure this out, you have to run simulations.
Doug C., you got your wish for more math. For the rest of you, even though 38% of you know that you can prove anything with statistics, just trust I did them right. And that my code was working right and I got a baseline for what the distribution of earthquakes SHOULD be if they were purely random.
You can do more complicated tests, I'm sure, but something my first advisor in grad school told me was that if your complicated statistical test shows something that looks significant, but you don't see it in the actual data, you should be skeptical. So it's really these basic tests that I did.
So I then went to the data. I separated them out into California-only data, for which half of the earthquake events I had were and those went down to magnitude 3, and I did the test for the whole world data, which I had complete to about 6th magnitude and comprised 3073 earthquakes.
What I found was purely random. I seriously found nothing that was statistically significant in any of the data. In the California data, there was a slight increase in earthquakes during an apogee moon -- the OPPOSITE of these peoples' model -- but it didn't show at all in the world-wide data. In terms of the most telling analysis, the fraction of events within 0.5-day intervals of both syzygy and perigee, the world-wide earthquake results are completely within 1-sigma of pure, random chance, for between 0.5 days and out to a full week. For the California-only data, there is a bit of above-chance for within 3.5 and 4.0 days of both syzygy and perigee, but it drops below-chance for 6.5 days. But, for the IMPORTANT non-retrodiction impressive prediction of the earthquakes happening within, say, 24 hours of BOTH syzygy and perigee, the data show this happens only about 1% of the time, which is pure chance.
I'd love to see the statistical analysis used by some of these folks who claim earthquakes are influenced by tides, because I found absolutely nothing. And as I said, a document detailing these will be in the shownotes within a day or so.
The final topic I want to talk about this episode is something that I haven't really gone into before, and that's a non-formal but still real fallacy termed the "Statistics Fallacy." Otherwise known as a Misuse of Statistics. A good example is this clip of David Nabhan -- and yes, the 39 minutes he was on the air was a host of Name That Logical Fallacies: [Coast to Coast AM clip, November 14, 2011, Hour 1, starting at 16:02]:
"I looked at the history of the great killer quakes that struck in southern California that struck between Long Beach in '33 and North Ridge in '94 and during those 60 year, uh, during that 60-year duration of the 20th century, 23 large magnitude 6.0 earthquakes or greater struck in Southern California. Now George, one-third, ONE-THIRD of them struck during the astoundingly thin target window represented by the hours between dawn and dusk during new and full moon near syzygy and or perigee. That is to say, a third of the killer quakes in southern California struck within a time frame that represents about one-half of one percent of the total time during those 60 years, with the other two-thirds spread all over the clock all over the calendar. Now, that resounding empirical evidence that the earth has laid squarely in front of anybody with a pair of eyes and an open mind, the math says would require random probabilities of one out of 50 billion for that to have happened randomly. Uh, I don't know about you, but I'm not willing to get on board with that -- with those kinds of numbers. Nor do I believe any thinking person would be."
There are a couple things in there, and this is a common kind of claim in this particular branch of pseudoscience and related ones. One statement he made is is that there were 23 magnitude 6 earthquakes in California between 1933 and 1994. Based on the source I have, the number is 38. So either he's using a much narrow range for southern California, or he's right away cherry-picked his data, meaning that he's selecting just points he wants to use to support his claim and rejecting other ones.
He then says that one-third of his 23 quakes struck between dawn and dusk during a new or full moon that was also at perigee. So I looked at the times of the day that these 38 quakes struck, and they were evenly split across all hours with nothing specific about daylight hours. Then I looked at ones that struck within a day of a perigee moon -- of my 38, only 5 did, or 13%, which was about chance if slightly higher. Same thing about within a day of a new or full moon -- only 5 of the 38 struck within a day of a new or full moon, also chance. Only 1 of the 38 earthquakes struck within 24 hours of BOTH a new or full moon AND when the Moon was at perigee.
So, the fallacy there first off is just not understanding the data.
But the next one is the outrageous statistics that he gives. He states that a third of the quakes struck within a time frame that's only 0.5% of the total time available, which based on my Monte Carlo simulation means that he's talking about a time period not 24 hours on each side, but only 12 hours on either side of both a syzygy and perigee moon.
He then says that this is only a 1 in 50 BILLION chance of happening randomly. If he were correct, that 1/3 of 23 events happened within that 24-hour window, he's actually talking about a 33-sigma event, which has much, much worse odds than 1 in 50 billion. I tried a few different common statistical mistakes and couldn't figure out what he did to get that number.
But based on the 1 million random simulations I did with the actual times for perigee, apogee, full, and new moons, there is a 0.27% chance of an earthquake happening within 12 hours on either side of syzygy. For a sample size of 23, the counting statistics give you an uncertainty of ±0.76%, meaning that you can realistically expect 0 or 1 earthquakes to occur. Which is what I found for southern California for ≥6.0 magnitude earthquakes between 1933 and 1994.
When the sample size is much larger, say, for a 43,000-earthquake database worldwide from 1932 to the present time, you would still have the same expectation of only 0.27%, but because of the much larger sample size, the expected range is only ±0.02% from that. A very tight range, but it's what I found was that out of those 43,000 quakes, 110 were within 12 hours of syzygy, which is 0.25%. Given its own counting errors, it has an uncertainty of ±10.5, and so it overlaps very well with the theoretical value for there being NO significance.
And that brings us to the end of this main segment. To recap, 'cause I took you over a large range of material, the idea is that high tides will trigger earthquakes to happen, and people claim vast conspiracies to cover up this obvious correlation.
Instead, they commit a host of fallacies that everyone should be aware of, including: non sequiturs, argument from persecution, argument from authority, argument against authority, argument from personal credulity, misuse of statistics, correlation = causation, remembering the hits while forgetting the misses, and data mining.
I've gotten some flack lately from some people on my blog about exploring the minutia of a claim, or not taking people at their word. The problem there is that isn't how things are done. Everything is up to a testing of the evidence, everything should be subjected to objective analysis, and if someone is scared of their claim being investigated, or they want you to look at something else that they've said instead, you have good reason to be skeptical and do your own, thorough, investigation.
Provide Your Comments:
Comments to date: 10. Page 1 of 1. Average Rating:
Chew Location unknown
6:24pm on Monday, September 17th, 2012
David Nabhan, the article could easily have been directly linked to from your website, instead of providing instructions on how to find it.
Stuart Robbins Boulder, CO, USA
5:52pm on Monday, September 17th, 2012
David, I'm preparing for a conference now so can't really respond. Have you looked at the statistics document I wrote? Have you done statistics on what the null hypothesis would be? You also need to define your terms. What is "Degree Syzygy," as Chew asked. And, as I pointed out in the podcast, and pulled from your interview on C2C, you must be aware that it does not help your case when you use so many logical fallacies.
David Nabhan Pittsburgh, PA
5:29pm on Monday, September 17th, 2012
Dear Chew: Those are good questions, the kind of questions that ANY normal investigative reporter would have asked ANY subject…BEFORE the “story” above was published. It’s odd that Mr. Robbins not only didn’t pose those…nor any other…nor ask for a comment or even inform me that a piece was in the offing, which is something that even Pravda would have done. Look, this seems an interesting site. I’m sure that plenty of pieces are done to great effect. But honestly I have no interest in monopolizing the conversation here, so I’ll simply now offer my thanks for having been heard at all. I answer all my emails, from all over the world, from friend or foe. I’m at email@example.com and the website is earthquakepredictors.com. And, Chew, see how easy it was to find those quakes? Pretty interesting list, no? Do me a big favor and forward them to Mr. Robbins—he can’t seem to find them.
Chew Location unknown
3:41pm on Monday, September 17th, 2012
David Nabhan, why do you define dawn and dusk as set times and not use the astronomical definition?
David Nabhan Pittsburgh, PA
12:18pm on Monday, September 17th, 2012
Dear Chew: Thanks for your interest; it is much appreciated—as are simple, straight-forward questions like that. “Southern California” insofar as the book “Forecasting the Catastrophe” is concerned, is that area south of just shy of the 36th parallel, the northern demarcation line of San Luis Obispo, Kern and San Bernardino Counties…along with the seven other counties below them. “Dawn” and “Dusk” are between 4:45 and 7:55 AM and/or PM. Again…I’d request that any person curious about the real facts involved in this matter go to earthquakepredictors.com and weigh the evidence published in the journal EdgeScience for themselves: “Courting Disaster.” Then…email Gov. Brown and make your opinion known. The link to his office is there also. Thanks!—David Nabhan
Chew Location unknown
10:29am on Monday, September 17th, 2012
David Nabhan, what are the coordinates you use to define "southern California"? Which specific earthquakes did you use in your analysis? What is your definition of "dawn" and "dusk"?
David Nabhan Pittsburgh, PA
7:03am on Monday, September 17th, 2012
Important PS: I hadn’t actually listened to the podcast that accompanies your article above when I jotted those quick comments to you below. Thankfully, the same person who emailed the article to me gave me to realize that the odd mix of errors and half-truths you publish above is just the tip of the iceberg—and that the real stunners are to be heard listening to you…interview yourself…about me. I’ll leave off commenting about all the logical fallacies you put in my mouth as you spoke for me, adroitly guessing for you audience how my mind works—instead of simply phoning me up. I will though address the most inexplicable part of your “interview,” having to do with your failure to find those SEVEN earthquakes (the one third part of all the great quakes that struck in Southern California between ’33 and ’94).
David Nabhan Pittsburgh, PA
3:34pm on Sunday, September 16th, 2012
It's always nice to have somebody repeat your words; much better, though if it's done properly? You got a few things wrong in your references to me in your article above. I wonder if you'd be kind enough to allow me to point them out?
Autymn D. C. CA
9:57pm on Monday, September 3rd, 2012
greater or smaller
Rick K. St. Louis, MO, USA
3:08pm on Sunday, September 2nd, 2012
We had some fun with Iben Browning back in 1989, when we had a minor quake in October of that year. Browning then predicted a "major earthquake" for December 2 or 3, 1990. Nothing happened, of course.