What is the current status of the reliability of predicting earthquakes? Last I heard, it was little better than pseudo-science, based mostly on a failed quake prediction for Parkfield, CA.
In the current state of knowledge, it’s essentially impossible to predict when a specific earthquake will occur, or how large it will be.
On a scale of 1-10, 10 being easiest to predict, Nate Silver gives earthquake prediction a 2.
It is possible to do some general forecasting regarding earthquakes, such as where they are more likely to occur, and their potential size. But specific earthquake prediction has never been shown to be successful.
Wikipedia’s article is comprehensive and well-cited.
It is possible to do statistical modelling and probability analysis of quakes. But it is a pretty blunt instrument. What is wanted is a date, time, location and intensity. What you really get is data like this: Canterbury region long-term probabilities(scroll down to table). (Note als the really interesting graph at the top of the page. Well worth opening the large version and looking at in detail.) Here is how the prediction compares with what actually happened:Canterbury region aftershock statistics and forecasts (Scroll down to table).
While this kind of analysis is obviously useful and can help communities prepare – especially in terms of engineering codes and specifications, it does little in the way of actual quake prediction.
Where an area is known to be earthquake prone it is possible to measure strain in the crust and estimate how much the stresses are building. It is still guesswork as to whether all of that energy will be released in one hit, when that will occur and on which fault system the stresses will be relieved. You can look at some stress data and not know whether an earthquake will happen next week or next century. If there is a long time between significant quakes then stresses can build to a higher level and the resultant quake is likely to be more energetic. However, even this is by no means a certainty.
In the case of the Christchurch quakes, all of the activity took place on a previously unknown system in an area that was not known for earthquakes in the historical past. Really difficult to make any kind of meaningful prediction under those circumstances and pretty much impossible to prepare people for it. The one that levelled the city on 22 Feb 2011 was actually an aftershock of the Mag 7 Darfield quake of Sept 4 2010. Unfortunately it was pretty much a perfect storm of circumstances that led to it being far more destructive that could have been imagined. IIRC, the ground acceleration exceed that of the Mag 9 quake responsible for the Tsunami that hit Fukushima. In other words, it is not just a matter of predicting the quake itself – the effects of the quake can be a bit of a dice throw as well.
Ideas on prediction are out there and people are always looking for something that will give them even a few minutes warning. Here is one of the more recent ideas to surface. Whether it holds any real promise is yet to be seen. (My guess is probably not.) Good minds are working on the prediction problem but it is likely to remain at least as difficult as predicting lightning.
According to this article, the Japanese can predict an earthquake 12-15 seconds in advance, which is enough time to slow down or stop the Shinkansen using automatic braking systems. That’s one kind of earthquake prediction that’s currently in use.
Thats more detection than prediction. They detect an earthquake and can then send out alerts.
I’ve heard of this technology before. It works because the alert signal travels at the speed of light, much faster than the quake itself. Does anyone know if this tech is being used outside of Japan?
Also, while Google is being unkind this morning, I did find this “earthquake forecast” site which has a bunch of numbers and terror-color coded numbers, but I can’t figure out what they exactly mean?