Author David Nabhan has written a book stating that quakes are most likely to happen within 3 hours of dawn or dusk and are also more likely within 36 hours of new or full moons.
I looked at all the 7+ quakes in the NEIC catalog from 1973 to 2013.
There were 638 of them.
124 were within 36 hours of a new or full moon.
279 were within 3 hours of 6:00 AM or PM.
Impressed?
Well, new/full moons happen every 2 weeks. 36 hours is 1.5 days or 3 days in all (+/-) so those windows are 52*6 days which is 312/365 so the odds on a hit are 0.855. Pretty hard to miss. He should have 545 hits, not 124.
3 hours +/- is 6 hours twice a day which is half a day so the odds on a hit are 0.50 so half the quakes should be hits.
He had 279 hits and should have had 319 so that's false too.
Now he was writing about the west coast so I looked at only quakes within the -115 to -130 longitude range. There 60% of the mag 7+ quakes were within +/- 3 hours of dawn or dusk.
The problem is that there were only 10 of them so while 6/10 is statistically significant it's too small a sample to be considered reliable.
Roger
(06-05-2014, 12:26 AM)Roger Hunter Wrote: [ -> ]Author David Nabhan has written a book stating that quakes are most likely to happen within 3 hours of dawn or dusk and are also more likely within 36 hours of new or full moons.
Yes, but does he say which dawns or dusks, and which new or full moons?
Even if there's greater chance at these times it's not very useful if you still don't know which of the 730 dawn/dusks and/or 24 new/full moons per year.
Brian
Roger,
I think your math is wrong. You have "52*6 days which is 312/365 "
But it is 6 days per month, not 6 days per week, so should be 12*6=72,
and odds should be 72/365.25
But, first post from you in a long time, so Hello.
Chris
[quote='Roger Hunter' pid='313' dateline='1401928017']
Well, new/full moons happen every 2 weeks. 36 hours is 1.5 days or 3 days in all (+/-) so those windows are 52*6 days which is 312/365 so the odds on a hit are 0.855. Pretty hard to miss. He should have 545 hits, not 124.
(06-07-2014, 11:57 AM)Island Chris Wrote: [ -> ]I think your math is wrong. You have "52*6 days which is 312/365 "
But it is 6 days per month, not 6 days per week, so should be 12*6=72,
and odds should be 72/365.25
Aaaaaagh!
You're right. I broke my own rule about letting the computer do the math
Within 3 hours is 6 hours. Twice a day is 12 hours.
Odds on a hit are 0.500
Expected number of hits is 319
There were 638 quakes and 279 were within 3 hours of dawn or dusk.
Quakes per hour (local time)
33 21 26 27 35 22 30 30 16 21 26 28 33 29 33 33 22 21 22 27 27 30 23 23
There were 1014 new or full moons during this period.
Odds on a moon hit are .203
Expected number of moon hits is 130
There were 638 quakes and 124 were within 36 hours of a new or full moon.
Quote:But, first post from you in a long time, so Hello.
back at'cha!
Roger
(06-05-2014, 12:26 AM)Roger Hunter Wrote: [ -> ]Well, new/full moons happen every 2 weeks. 36 hours is 1.5 days or 3 days in all (+/-) so those windows are 52*6 days which is 312/365 so the odds on a hit are 0.855. Pretty hard to miss. He should have 545 hits, not 124.
Roger, that make a lot more sense. With a lot of moons and sunrises/sunsets, if there is no correlation, should be close to chance. At first you suggested a hugely significant anti-correlation, which if true would have meant something.
Chris
(06-08-2014, 12:15 PM)Island Chris Wrote: [ -> ]Roger, that make a lot more sense. With a lot of moons and sunrises/sunsets, if there is no correlation, should be close to chance. At first you suggested a hugely significant anti-correlation, which if true would have meant something.
Chris
Pursuing this investigation a little further, I added mag limits to the program.
I found that mag 6-10 gave prefect chance results whereas mag 5-6 gave extremely negative results.
This makes perfect sense. The database was California quakes and the only tidal correlation ever found was for small shallow angle thrust faults at LOW tide whereas Full and New moons are HIGH tide..
Roger
Roger, as worded your post does not sound right. Maybe your concise wording is ambiguous and I mis-read. For Ocean tides full and new moon tides are both low and high. It is just that the magnitude (range) of the tides is larger. I would think it would be the same for earth tides.
Chris
(06-22-2014, 07:59 PM)Roger Hunter Wrote: [ -> ] (06-08-2014, 12:15 PM)Island Chris Wrote: [ -> ]Roger, that make a lot more sense. With a lot of moons and sunrises/sunsets, if there is no correlation, should be close to chance. At first you suggested a hugely significant anti-correlation, which if true would have meant something.
Chris
Pursuing this investigation a little further, I added mag limits to the program.
I found that mag 6-10 gave prefect chance results whereas mag 5-6 gave extremely negative results.
This makes perfect sense. The database was California quakes and the only tidal correlation ever found was for small shallow angle thrust faults at LOW tide whereas Full and New moons are HIGH tide..
Roger
(06-27-2014, 11:57 AM)Island Chris Wrote: [ -> ]Roger, as worded your post does not sound right. Maybe your concise wording is ambiguous and I mis-read. For Ocean tides full and new moon tides are both low and high. It is just that the magnitude (range) of the tides is larger. I would think it would be the same for earth tides.
Once again, you're right. Thank goodness you read these posts!
Yes, I was ignoring the diurnal cycle since Nabhan was claiming effects from new and full moons. I added the quarter moons and found no correlation with them either and once again the difference is highly significant negatively.
This only applies to mag 5 quakes. Mag 6+ are pure chance and I can't account for the difference.
Roger
Highly significant negatively is just as interesting as highly significant positively. And, maybe, just as unlikely. You may want to look hard at what you are doing, and/or post the details here.
Chris
(06-30-2014, 10:57 AM)Island Chris Wrote: [ -> ]Highly significant negatively is just as interesting as highly significant positively. And, maybe, just as unlikely. You may want to look hard at what you are doing, and/or post the details here.
Chris
Yes Chris, I quite agree but I don't know what to do about it.
My program makes a list of days and adds quakes, moon phases and perigees. The odds are the number of each element times the days covered divided by the total number of days. Moon phases have a 36 hour window (+/-) so get 3 days each. Perigees get two days total.
The expected number of hits is number of quakes times odds.
Mag 6+ quakes get the expected number of hits. Mag 5 quakes get too few. It's easily seen looking at the big table.
Roger