ring hit

Wed, 15 Jun 2011 19:48:00 +0100

Was that a "hit" ?

The big earthquake just been. Did Ken Ring's predictions "hit" it?

According to my previous betting schedule, I wagered these bets:

3 chips on ±1 day surrounding the lunar equinox [ paying 14:1 ]
3 chips on ±1 day surrounding the lunar perigee [ paying 14:1 ]
2 chips on ±1 day surrounding the point of maximum declination [ paying 14:1 ]
2 chips on ±1 day surrounding the full moon [ paying 14:1 ]
2 chips on ±1 day surrounding the new moon [ paying 14:1 ]
1 chip on ±1 day surrounding the point of first quarter [ paying 14:1 ]
1 chip on ±1 day surrounding the point of third quarter [ paying 14:1 ]
2 chips on ±1 day surrounding the lunar apogee [ paying 14:1 ]

Let's just assume for now that the chance of the earthquake is completely random (it isn't; there is a 2% extra chance at lunar perigee, but let's not worry about this for now).

The earthquake was (in GMT/UTC) at 13 June 2011 at 02:20.

The lunar perigee (closest point to earth in its orbit) was at Jun 12, 2011 01:43:00 AM GMT - miss by all of 37 minutes

The point of southern lunar declination was at Jun 15, 2011 8:54am GMT - miss by 1d 6h 34m.

The full moon was at June 15, 20:14 GMT (very shortly, along with the eclipse of course) - miss by 1d 18h 34m.

Darn. Didn't win any pots.

Let's widen the odds to ±2 days:

3 chips on ±2 days surrounding the lunar equinox [ paying 7:1 ]
3 chips on ±2 days surrounding the lunar perigee [ paying 7:1 ]
2 chips on ±2 days surrounding the point of maximum declination [ paying 7:1 ]
2 chips on ±2 days surrounding the full moon [ paying 7:1 ]
2 chips on ±2 day surrounding the new moon [ paying 7:1 ]
1 chip on ±2 days surrounding the point of first quarter [ paying 7:1 ]
1 chip on ±2 days surrounding the point of third quarter [ paying 7:1 ]
2 chips on ±2 days surrounding the lunar apogee [ paying 7:1 ]

The paid odds reduce, because the windows are wider.

The 3 chips on the perigee paid off, returning 21 chips. All other bets failed, returning nothing. 16 chips were laid down. Total profit: 5 chips.

What about widening the odds to ±3 days, making the odds 3.5:1?

Then, the 3 bet on the perigee, the 3 on the full moon, and the 2 bet on the declination all pay back. Again 16 chips were laid down. This pays back 28 chips, a more substantial profit, in fact almost doubling the outlay for this month.

I guess when gambling, sometimes you win. To enjoy it the most, you need to forget all the times that you went home empty-handed instead.

ring function

Fri, 11 Mar 2011 22:00:00 +0000

Ravings to Hypothesis: a stab at a "Ring function"

As others have pointed out, Ken Ring makes a lot of predictions. As Ring does not attempt to turn his viewpoints into rigourous studies, the best that work like this normally deserves is referencing the closest rebuttal of similar ideas along with some reasoning as to why it is sufficiently similar. Not every crackpot idea deserves a full scientific investigation! But the supermoon hypothesis is now gaining momentum, at least in the media.

Ring does make some quantitative predictions, let's look at perhaps this section from a book of Ring's, Moon & Weather Lore:

Earthquakes mostly occur when the perigeal or apogeal moon is at either declination(stitial colure) or crossing the equator (lunar equinox), and within one or two days of either of these. A detailed glance at any earthquake gathering station will reveal that around these dates the numbers of quakes rise steeply and then dropoff afterwards as the moon moes out of those declination zones. Close perigee and full Moon or new moon adds to the potential for increased earthquake activity along the moon's path between the latitudes.

And later, from Predicting the Weather by the Moon:

There is evidence that moonquakes increase when the Moon is closest in its orbit to the Earth. Correspondingly, we might expect an increase in Earthquakes at that time, (the perigee) too. Earthquakes are triggered by the moon in its monthly movement north and south of the quator and its orbit around the Earth. The word ‘triggered’ is used here because the Moon may pass over a danger point many times until the stress on a fault becomes too great, after which the fault may give in one sudden movement.

...

One of the main danger times is when the Moon is crossing the equator during the monthly declination cycle. This is the time while the Moon is moving quickly between the hemispheres. When the Moon is at the maximum 28° declination, it will cross the equator twice each month at about seven degrees in a day which gives considerable pull on the planet. At minimum 18°, it crosses at about four degrees in a day and the effect is less positive.

The other danger point is while the Moon is at either of the maximum declination positions north and south of the equator. The Moon is at these positions for about three days and does place considerable strain on the techtonic plates while there. It must be remembered that the Moon is always on the move and a quake can happen at any time.

...

In G.A. Elby's book “Earthquakes” (Heinemann 1980), 209 earthquakes dating back to 1505 were recorded with their dates. We can check each quake against Moon phases. 96%. of these quakes recorded which were above 6 on the Richter Scale, occurred exactly on or within a day of one extreme feature of the Moon cycle, that is, New Moon, Full Moon, Apogee or Perigee. 75% involved two factos; when the say, the Perigee plus Full or New moon were on the same day.

So what's the summary of this?

Ring actually predicts Earthquake risk as a function over time and space. Each of the major events in the cycle of the moon's orbit represent increased risk factors; a confluence of extremes, further risk again.

This might not exactly correlate with past studies into lunar earthquake relationships; a study which fails to find a pattern only disproves its own method - it does not exclude other studies which might use a different method.

That being said, it's not a good idea to ignore studies which produce negative results; they shave away at the idea, approaching it from many different angles, until the balance of probabilities is that the idea is considered disproven.

Testing Ring's theory scientifically.

Perhaps the simplest approach to testing a prediction theory would be a gambling game, similar to roullette.

The goal of this is to show, scientifically, that a certain limited sets of dates have an increased risk of earthquake. If a significant relationship is found, these dates could potentially be used as quasi-arbitrary dates for civil defence planning.

Ladies and Gentlemen, bets please.

As the model makes predictions, the ‘house’ pays odds on those predictions based on the odds of that prediction occurring according to best accepted theories.

You are allowed to place this pot on future times, and possibly, specific regions (which would lengthen the odds dramatically).

Where the earthquake events occur, they pay back to the bet placed on them, times the released energy in units of, say, log E, where E is the energy released by the earthquake in MegaJoules, such that getting a hit on a very large earthquake pays back handsomely.

Odds would shorten after an earthquake for aftershocks in the region, according to current established theory. For some types of analysis, it may be more useful to simply remove aftershocks from the input data to avoid having to make prediction functions include aftershock predictions.

The nice thing about this is it allows people to "play" real-time, as well as being able to test the past success of the forecasting abilities (a technique known as hindcasting). Thus people who are not able or willing to share their methods can participate and be judged fairly.

Gambler's ruin

A function which simply places bets on all outcomes is not useful and will result in a high score. There are several ways around this:

the betting agent has a finite pool of chips which they can only bet until they have run out; the house scores in terms of paid back chips, and functions are compared by the amount of chips they have in their pool. The nice thing about this is that it is somewhat self-regulating. However it is more difficult to reason with.
placed bets are scaled so that they represent a predicted energy release function, with the total energy release matching the energy release of the period. The "overlap" is the score of the function.
A statistical significance test. Feed the prediction function or predictions with random input data (with known aftershock behaviour built into it) and see how many times it gets such a high score. If the score run against the real data is never matched by the prediction function with random input data, over say 100 runs, then this may be enough to show statistical significance at the 1% level.

A concrete "Ring Function";

The function will bet on a series of extremes and mid-points:

3 chips on ±1 day surrounding the lunar equinox
3 chips on ±1 day surrounding the lunar perigee
2 chips on ±1 day surrounding the point of maximum declination
2 chips on ±1 day surrounding the full moon
2 chips on ±1 day surrounding the new moon
1 chip on ±1 day surrounding the point of first quarter
1 chip on ±1 day surrounding the point of third quarter
2 chips on ±1 day surrounding the lunar apogee
3 extra chips on the lunar perigee bets covering the 2 closest lunar approaches every 18.6 years
3 extra chips on the lunar apogee bets covering the 2 furthest lunar distances every 18.6 years
3 extra chips on the bets covering the maximum declination on years where the declination is at its maximum 28° (as in 2004/5)
3 extra chips on the best covering the lunar equinox, on the bets that surround the maximum declination times (ie, the points where the change in declination peaks).

The above should be able to test the claim from Predicting the Weather by the moon - though obviously, there is still a lot of mathematical work involved in getting the rules of the game ironed out, and expressing the functions mathematically.

To give the man a chance, the values of the odds, and the width of the bets, are to be fine-tuned using a genetic algorithm; ie, tweak values one by one to get better results.

Anyone want to write the site/thesis?

The Christchurch Daily Energy Release page has a chart which is something towards the above goals.

However a full system to incorporate existing theories, a betting/odds engine (the minimal approach above), as well as collecting/calculating all the relevant information and writing the functions in terms of those sources is just not a small job. Ken Ring, of course, could fund this, if he was genuine about testing his theory and not just after selling predictions. Then again, it's perhaps not in his best interests financially to do this - as he writes in his book above, he's already confirmed to his own satisfaction that there is a pattern. So he'll keep his "competitive edge" rather than try to solve the massive problem of proving a hunch scientifically.

I produce this text in the hope that those who do try to take up the task of proving or disproving Ring's predictions have something to work with, without being forced to purchase or read his books, which are full of crazy theories. And I'm more than happy to send my copies of the books to someone seriously taking this challenge on.

But myself, I have theories to develop in my own field.

pages tagged eq