- - - - - - - - - - - - -

Sunday, December 30, 2007

NZ government scientist insists week-long earthquake swarm is unconnected and random: "Mimi, mi, mi, mimimi, mi"

Is it any wonder that Muppet Labs can't predict earthquakes when they insist that all clusters of the phenomena they study are always random? A scientist in any other field who ignored this localised cluster data and the place it has in the system that causes it (ie. plate tectonics) would be drummed out of their profession... but not if you're a seisemologist, apparently.

GNS Science says a spate of earthquakes in the North Island are not connected.
A quake measuring 5.6 on the Richter Scale occurred at 7.03am on Saturday, 80km below the surface of Lake Taupo. It was felt across much of the central North Island.
An earlier quake measuring 5.5 on the Richter scale occurred at 9.07pm on Thursday. It was located 10km northwest of Turangi at a depth of 120km.
Seismologist Brian Field says it is unlikely the latest earthquake has caused any significant damage.
says a spate of recent earthquakes in the North Island are not connected and are completely random.
There were two quakes on Boxing Day: a 4.4 magnitude earthquake at 5.18pm, 40km south of Hawera, and another measuring 3.5, 60km northeast of Gisborne at 7.22am.
A 6.8 magnitude earthquake struck the Gisborne region at 8.55pm on Thursday, 20 December [...]
- 2:29pm on 29 Dec 2007

"Not connected and completely random"

Which would naturally, of course explain this:

There was an earthquake measuring 4.1 on the Richter scale 50km south east of Wairoa today.
GNS Science said the earthquake occurred at 11.34am and was at a depth of 25km.
- 12:18PM Sunday December 30, 2007

The only thing unconnected in this story is the seismologist from his brain.



At 30/12/07 8:27 pm, Blogger Rich said...

Sorry Tim, I didn't know you were a seismologist? Or are you just a believer in the traditional NZ attitude that a bit of applied stupidity (sorry, common sense) trumps the result of years of painstaking study?

There are over 1000 magnitude 3.5+ earthquakes a year and 24 magnitude 5+, which equates to a 5+ every 16 days and several 3.5+ quakes a day.

So six *reported* quakes over a 10 day period is perfectly to be expected and doesn't signify anything (there were actually 49 3.5+ quakes in that period).

Random events *do* form apparent clusters - if you sit tossing a coin for long enough, it will come up heads several times in a row.

The reason you *heard* about lots of quakes is that the news media will report small quakes after a newsworthy large one - especially at Christmas with a lack of other (domestic) news. Media aftershocks, not seismic ones.

I got my numbers from this searchable database of quakes:

At 30/12/07 9:52 pm, Blogger Stephen said...

Yeah, bullshit Tim. You're full of crap on this one.

Random noise does come in bursts (and actually, the bursts come in bursts too; grab anything by Mandelbrot and check it out) and in fact if your data was more or less even, that would be far more weird and worthy of investigation.

If it weren't for the fact that for once a reasonably well-settled place got something noticeable you would be none the wiser.

At 31/12/07 12:51 am, Blogger Tim Selwyn said...

Thanks for the link. I will use it to post some images to demonstrate what I'm talking about.

The problem is going to be to convince you that it is not "random" and "unconnected". To do this I must "establish a connexion" and "show a pattern". My premise is based on plate tectonic theory.

If that theory is correct then the entire plate boundary is constantly causing earthquakes because it is in motion. The way in which it moves will probably be predictable because of the constant pressure. Because of widespread monitoring since the 60s the data should be enough for an observer to notice something.

If you look at the map of NZ, the ridge of deeper quakes lies to the west - over time this is demonstrably true. Now if you took a snapshot of any day it might look random. It is accepted that the longer timescale is necessary to prove it is not random. So it should also be accepted that to prove they are connected and are a pattern we should look at a shorter timescale.

At 31/12/07 1:07 am, Blogger Tim Selwyn said...


Earthquakes also often occur in volcanic regions and are caused there, both by tectonic faults and by the movement of magma in volcanoes. Such earthquakes can serve as an early warning of volcanic eruptions.
Sometimes a series of earthquakes occur in a sort of earthquake storm, where the earthquakes strike a fault in clusters, each triggered by the shaking or stress redistribution of the previous earthquakes. Similar to aftershocks but on adjacent segments of fault, these storms occur over the course of years, and with some of the later earthquakes as damaging as the early ones. Such a pattern was observed in the sequence of about a dozen earthquakes that struck the North Anatolian Fault in Turkey in the 20th century, the half dozen large earthquakes in New Madrid in 1811-1812, and has been inferred for older anomalous clusters of large earthquakes in the Middle East and in the Mojave Desert.

At 31/12/07 1:04 pm, Blogger Stephen said...

This kind of thinking convicts people of murder: Losing the lottery

At 31/12/07 2:48 pm, Blogger Tim Selwyn said...

You people are deniers.

I'm sure you will call the after-shocks from a major quake that continue for days afterwards "after-shocks" that are after the major shock and are connected to it and are part of a pattern - and also a predictable pattern. So you should also recognise the possibility of pre-shocks that are before the major shock and are connected to it and are part of a pattern - and also a predictable pattern.

At 31/12/07 6:04 pm, Blogger Stephen said...

Denier? I certainly deny that you are on to anything.

Your intuitions are simply wrong. You can find apparent patterns, clusters, trends, etc in all kinds of random data. You actually hit on it yourself when you stressed "predictable." A pattern is only interesting if we can make predictions from it. I would be amazed, absolutely amazed if seismologists haven't already been looking for such patterns.

I don't trust expert predictions in any field very much, precisely because they have track records no better than chance and a nasty habit of weighing the wrong factors the wrong way. But I would still trust a seismologist to be more on top of the state of the art than you, who doesn't even seem to grasp what "random" means.

At 1/1/08 1:41 pm, Anonymous Anonymous said...

Pre-dictions based on past data, you should get a job with the Reserve bank.
yes they form a pattern, no they are not connected in that one quake leads to another. They all occur beacuse of the location on the plate boundary, that is the only "connection".
Check out 'random walk' if you want to know about random events that seem to form a pattern


Post a Comment

<< Home