Earthquake forecasting

like a weather forecast, this would have to be probabilistic i.e. There may be about. a magnitude 2 earthquake in this district in the late afternoon. Sort of like tornado warnings.

So far earthquake prediction as been either
1 ) where destructive events have left there mark in the environment in a periodic fashion, then it is reasonable to predict similar levels of destruction and periodicity in the future.
2 ) the idea of earthquake storms, where one earthquake increases the stress at the next locked point along the same fault, increasing the probability that that will by the site of the next earthquake, and after that one the stress gets passed to the next locked point, and so on. This technique only works on some well mapped / characterised faults such as the North Anatolian fault
these two methods are only suitable for longer term planing, such as not moving to Istanbul Seattle or Vancouver.

Here is an idea that might bring earthquake prediction up to about the accuracy of a 1950's weather forecast

a) The earth is not getting bigger, or smaller, over time!
(most) Earthquakes might be expected to cause either an increase or decrees in the circumference of the earth along the great circle along which the earth moves.- so by tracking earthquakes globally and calculating there effects upon other faults on there grate circle it aught to be possible to extend the earthquake storm model to cover most earthquakes. Because there are thousands of earthquakes every day a picture of exactly where on the globe the stresses are changing will be quickly built up. This information by its self dose not provide the prediction has to when a particular fault will rupture, to get that information it would be necessary to run the model with 50 or so years worth of historical data. At first the model will be more useful for excluding earthquakes, because a particular earthquake fault had been under grater stress than this in the passed and did not rupture. going forward it will become increasingly good at predicting earthquakes, particularly in those areas that generate a lot of data.
b) Liberation. When the earth is struck by a large earthquake, it rings like a bell. Further earthquake are more likely whilst the earth is ringing / vibrating.
c) Although tidal forces are too small to cause an earthquake it is possible that they may play a part in triggering some earthquakes. This will become apparent as the model is run, if it is a real effect

The 90 east ridge and other similar poorly described seismic zones pose a problem, as it will not be clear if an there earthquake increases or decreases the stress along the appropriate grate circle Running the model with historical data should make it obvious which it is. I suspect that the 90 east ridge is a pressure ridge, smiler to those that form on sea ice.

I used the word (most) above because very shallow earthquakes need not necessarily be strongly linked to the underlying tectonic plates.