When evaluating the success of an earthquake prediction method one should always consider the prior probability that an event will happen in the prediction time, space, and magnitude windows based on past catalogs. If the probability of success of each prediction is the same, one can determine if a statistically significant result has been obtained by using the binomial distribution. When the probability of success for the predictions is different for different predictions, it is not straightforward to evaluate if a statistically significant result has been obtained. We present a method for scoring predictions to test their success.

Many people...

You do not currently have access to this article.