Can earthquakes be predicted?

Total Pageviews

Search This Blog

Tuesday, October 01, 2013

Earthquake Prediction for October 2013

Hi
 Here are the probable dates and magnitude for probable earthquakes in October 2013

1) 4th/5th Oct.............6.4,,,( from 12 Noon UTC on 4th to 12 Noon UTC on 5th)

2) 10th /11th.............6.7

3)18th......................6.6

4) 24th....................6.3


As a typical test case the time and location of 10th Oct quake is given here

10th Oct 2013.......23.06 UTC.......80 degree west and/or 100 degree East

window period  for sr no 1 to 4 is + or - one day

magnitude window is + or - 0.3

Amit

27 comments:

AMIT said...

Hi
please see the accuracy of prediction for 4/5th October,both date and magnitude
6.4 quake on 4th at Mid Atalantic ridge
Amit

AMIT said...

hi
Now some statistical data regarding 6.4 quake prediction for 4/5th October

Between 1st Oct 2004 to 1st Oct 2013 ie 9 years ie 3240 days there are 104 earthquakes of exactly 6.4 magnitude
ie odds are

104/ 3240 no of quakes /no of days for a date hit prediction

ie 0.032 odds
ie nearly 1 in 30

Is this below average ?
remarks are invited
Roger is requested to Comment as an expert in earthquake evaluation
Amit

Roger Hunter said...

Amit;

Out of all the predictions you have made, how many were hits?

That's what matters, not whether one hit.

Roger

AMIT said...

Roger
you are requested to do one thing

-search my blog for "date hit prediction " and find for your self how many of them are date hits

Amit Dave

Roger Hunter said...

Amit;

I have all your predictions to date but my quake file only goes to 2013/09/04.

During that period you made 193 predictions for quakes of mag 6+ in a 3 day window.

The odds on a hit are 0.683

You got 135 hits and the expected number is 131.8 so you got 3 more hits than expected.

This is not significantly better than chance.

Roger

Anthony Boucouvalas said...

Amit Roger is right on this one. Come on now...

AMIT said...

Roger
Thanks for feedback. Few things I would like to state here
1) It is good enough to beat the odds for 6+ quake predictions over several years of consistant predictions.135 hits out of 193 predictions is good ,as far as my understaning.
2) Odds takes in to account all quakes ie aftershocks at same place also besides taking clusters in account.so odds are always high
3)If these are true odds 5 out of 10 should be able to reach with their random dates. I doubt even one in tencan reach odds with random dates
Amit

Roger Hunter said...

Amit;

1) It's not good, it's just chance and (without location) useless.

2) The odds are for having one or more 6+ quakes in a 3 day window and that's all you are predicting. You don't predict how many will happen.

3) That's like saying a coin tossed 1000 times will not be heads half the time.
In fact, it won't and there's a distribution curve that tells us how often it will come out at different values. And that curve tells me that your results are not significant.

Roger

AMIT said...

Roger
1)If anybody can achieve with random dates, at least 5 in 10 person should be able to achieve it by giving random dates
It is not fair to say "anybody can achieve the odds" ,and also to say "it can not be proved practically"

Let ten person give random dates and observe how-may can reach the odds.

2) can you run random dates on software and find out one in how may attempts the odds (6+ ,with 3 day window) can be achieved?

Amit

Roger Hunter said...

Amit;

I'm working on it now.

But will you accept the results and admit you were wrong?

Roger

Roger Hunter said...

Amit;

I've tested random predictions with 3 day windows against 6+ quakes and find that they do better than chance and in fact better than your real predictions. This is with trials of up to 100,000 repetitions.

The ratio was about 60% better to 40% worse.

Mathematically this makes no sense and I found that slight changes in the odds could reverse the ratio.

But I have no basis for tinkering with the odds.

All I can say at this point is that my evaluation is as good as I can make it but there's an area of uncertainty in my results such that to be taken seriously you have to pass by a wide margin.

Which actually supports my insistence on a 99% significance level.

Roger

AMIT said...

Roger
The bottom line is there is no time tested method for evaluating earthquake predictions,
It is surprising to see the indifference of seismology departments in general and USGS in particular,that no such proven method is officially approved by any authority. In fact, it is so strongly believed,at all leve, that,the earthquakes can not be predicted and hence no need of such futile exercises
I will believe that,my predictions are below odds/averages if at least 5 out of ten people can beat my predictions with their random dates,over a considerable period of time
Amit

Roger Hunter said...

Amit;

If the difference between your predictions and randomly chosen dates is small, why do you think you're succeeding?

You have a method which is sometimes correct but more times not. That strongly suggests chance at work.

A good method would work most of the time, not some of the time.

Roger

AMIT said...

Roger
One more aspect ..
I predict quake with +or - 0.3 mag window. Each prediction should be evaluated with its own odds, ie if predicted 6.5 the odds to be considered should be only pertaining to quakes between 6.2 and 6.8.
Thus each prediction odds are different.Hence 6+ quake odds can not be applied to all predictions. How these all prediction can be evaluated with single odd is not known.
Amit

AMIT said...

Hi
One more date ,time ,magnitude and locations are given here to be observed
(These are personal views)

11th October 2013...22.30 UTC
and 6.9 magnitude
locations .. 75 degree West and /or 105 degree east

windows

Date ...nil
time....+ or - 6 hrs
magnitude.....+ or -0.3
location..+ or -10 degrees
Amit

Roger Hunter said...

Amit;

The location is too vague. Latitude is required too. Right now you've specified two bands of longitude, each 20 degrees wide and 180 degrees tall.

Roger

AMIT said...

Hi
watch accuracy
Two quakes hours apart on10/11 th Oct 2013.karmadic island 6.3 and venezuala 6.1
Are these by chance or fluke?
There is no proven method in stats to evaluate predictions.Hence on pratical ground let 5 out of 10 peoplee beat my score with their random dates
Amit

Roger Hunter said...

Amit;

Why 5 out of 10? What are you basing that on?

Do you think you have a 50% chance on your predictions?

Roger

AMIT said...

Roger
1) Yet another quake of 6.4 at Greece
Three 6+ quakes at different places.Is this just a chance?
2) I am not an expert in stats.Nevertheless I know that all theoritical hypothesis have to be proved practicaly,including..God particle.
Hence your odds (which are supposed to be easily achieved by random dates)
should be achieved at least by 50% person.In fact two third person should achieve it,but for equal opportunity I prefer 50%
Amit

AMIT said...

hi
I would like to be more specific for remaining two dates of October. As we have not entered window period as of now (15th ,14.00 UTC now) ,I would like to give specific details of these dates

1)18th October....23.00 UTC ...6.6

2)23rd October ...01.15 UTC...6.7

window period is reduced to + or - 12 Hrs
place not predicted

Amit

AMIT said...

hi
Pl note 19th quake of 6.5 at Gulf of California (17.54 UTC)
Though, + or -12 Hrs accuracy could not be maintained, + or - one day accuracy along with + or -.03 Mag accuracy is maintained
Now ,how can we say,these are all by chance or by fluke

Amit

Roger Hunter said...

Amit;

You're making 6 to 8 predictions per month with a chance expectation of 30%.

Why are you claiming success when one or two are correct?

That's how chance works.

Roger

AMIT said...

Roger
If I am giving 8 date per month ,it would be 100 odd in a year and 600 odd in 6 years
where as actual dates are 200 odds
Amit

Anonymous said...

amit, you're full of shit. you're not a predictor of earthquakes. you just have no life.

AMIT said...

Anonymous
Thank you so much

regards

Amit

AMIT said...

hi
watch the date hit 6.5+ prediction
6.7 quake at south sandwich island on 24th October 2013
How many 6.5+ date hits are required to prove the point?
Amit

AMIT said...

Roger
watch one more 7.3 at Japan ,on 25th at 17.10 UTC
How many hits in window period
You may say ,I have not specified two in window. You also know ,there is no point in such argument.
You are aware that,my theory basically states that the date has potential of 6.5+ quake. As the earth moves /rotates ,various fault lines comes under influence of pull and there could be more quakes.
The point here is to see the potency of date to produce such quakes,on diffrent plate boundaries and not related to each other. watch also my indicators if you wish
Jupiter closer having higher declination amd Moon as trigger Joining Jupiter now
Amit