The numbers below are guestimates, but given all that we don't know, I wonder whether some modeling and rigorous analysis would be any better. I am aware that I am not doing probability correctly...This page is a hobby...I don't do earthquake hazard analysis for a living (or at all):
OK...let's guess that there is a 75% chance of the 1833 Sumatra rupture area failing in the next 100 years. So, on average, there is 1 chance in 48,700 (call it 50,000) that it will fail on any given day. But, let's say that the date of the future quake is unknown, but the stress changes will advance that quake towards present by 10 years (educated guess). And, it has not occurred yet. And, the subduction zone just south of the equator may continue to weaken for the next month or 2 by delayed stress transfers and increasing pore pressure....and then continue to weaken by plate motion loading until the quake occurs. So, I'll guess (semi-educated) that there is a 25% chance that it will fail in the next 3 years (meaning odds of it failing between 3 and 100 years are less than 75%). So, 1 chance in 4383 (call it 4400) on any given day. Then, say that the rate of weakening is a little faster in next couple of months than it will be after that. I'll just guess that the odds would increase to 1 chance in 3000 that the quake would occur on any given day from now through July (for example, 1 chance in 100 that it would occur in June 2005), then dropping off to less likely after that.
Let's say the quake did not occur in the next 3 months. Then, the odds of it
occurring in the following 3 months would be slightly higher than they would be estimating forwards today between 4 and 6 months...due to the fact that the quake had not occurred and that the main weakening would then just be plate motion...but, probably less likely than 1 in 3000.
So....my post above should not be read to say that I am predicting a quake for June 30....it should be read that the odds for a quake that day might be 1 in 3000, as compared to odds of 1 in 50,000 for the average day over the next 100 years.....or, 17 times more likely that quake will occur on June 30 2005 than June 30 2105.
The concepts discussed here could be useful for planning, but the numbers I made up......but I'd like someone like John to say whether they believe these odds are radically wrong, or on-track.
Chris