Friday, March 11, 2011

Fifth greatest earthquake since 1900

Japan has experienced a magnitude 8.9 earthquake several hours ago. Later, the magnitude was upgraded to 9.0.

According to USGS, it was the fifth greatest earthquake since 1900. Only O(2000) people have died after the largest Japanese earthquake in 140 years (temporary figure); the number of casualties would be counted at least in tens of thousands if the country were not as developed as Japan. For example, the 2010 Haiti earthquake killed 100,000-300,000 people although its magnitude was just 7.0 - one hundred times smaller.

The Japanese managed to extinguish a fire in a nuclear power plant but this struggle is clear not yet over.

Tsunami in the Pacific isn't over yet.

Japanese researchers study a black hole

This blog already existed and discussed the Indonesian tsunami on December 26th, 2004 (third strongest since 1900), as well the Chilean earthquake on February 27th, 2010 (sixth strongest since 1900). So three out of six largest earthquakes since 1900 took place in the recent 6 years.

Frankly speaking, I am surprised that the media haven't begun to inform the people about the looming doomsday. Well, it's probably because no one has yet invented a good story that has a well-defined culprit and that would allow him or her to get richer.

What is the probability that 3 or more out of 6 largest earthquakes in a 111-year period (1900-2010+) takes place during the last six years of the interval? Well, for one earthquake, it is 6/111. For three earthquakes, it is (6/111)^3. However, any triplet of top 6 entries is equally good, so we should multiply the odds by (6 choose 3) = 20 (the probability of 4 or more is double-counted etc. but I neglect those small terms). The result is
20 x (6/111)3 = 0.003
One gets a similar figure for the probability that 2 out of 6 greatest earthquakes occur in the last 1.1 years.

That means that we have exactly a three-sigma or 99.7% confidence that the recent earthquakes show that Apple is destroying planet Earth with its new iOS devices - or whatever will be the preferred explanation of the earthquakes.

In particular, the Japanese earthquake was caused by iPad 2 which is being released exactly today while iPad 1 was released exactly one year ago, i.e. 2 weeks after the Chilean earthquake. At the time of the Indonesian tsunami, Apple changed its suppliers and opened its first store in Europe (London) and sold its 200-millionth song via iTunes. The science is clearly settled. My 300-to-1 proofs are even stronger than the proofs of AGW which are based on 3-to-1 odds.

To try to avoid a lawsuit from Apple, I have to emphasize that the previous two paragraphs were a joke: well, some people won't believe me it's a joke, anyway, but I have tried, Steve.

As I have repeatedly emphasized, 3 standard deviations is not a terribly strong piece of evidence. It's because one can often fine-tune the "disciplines" in which strong "bumps" may be found by accident - and because there are so many disciplines, it's not such an accident if you find a three-sigma bump.

The average year in which the 16 largest earthquakes in 1900-2011 took place is 1964.8. The middle of the interval is just 1955.5. The probability that the average is 1964.8 or higher, instead of the expected 1955.5, is 12 percent or so. Not a terribly strong signal. Thanks, Wolfram Mathematica.

You know, statistics happens. Events and combinations of events whose probability is nonzero will ultimately occur if they're given a sufficient number of attempts when they may occur.

However, I don't suggest that the earthquakes occur randomly in the sense of Poisson distribution - random beeps that you may hear from the Geiger-Müller counter. It seems pretty clear to me that the earthquakes are correlated with each other - both spatially and temporarily - a little bit like fractals with some characteristic scaling laws describing the autocorrelation.

The series of earthquakes in the 1960s is unlikely to be an accident and there exist good theoretical reasons why one earthquake could make it easier for its followups to take place as well.

Tsunami have been smaller than expected. Well, I guess that the proximity of the epicenter to a big island is what doesn't allow the waves to grow much.

A NOAA visualization of the wave's height at some moment (or at the maximum-height moment chosen separately for each point? It sounds more likely given the radial shapes), via Andrew Revkin.

1 comment: