Large Aftershocks Prediction Results In Eastern And Central Iran Using Artificial Neural Networks (ANNs)

In this study, aftershocks of Birjand-Ghaen earthquake of May ۱۰, ۱۹۹۷ (N۴,---۷,۳) in eastern Iran and Golbaf-Kerman earthquake of June I ۱, ۱۹۸۱ (N۱,-۶.۷) in central Iran were used for predicting large aftershock by two methods. In the first method, using modified Omori function, aftershock activity shows an appreciable decerease form the level expected from this formula, before the occurrence of the large aftershock and then recovers to the normal level or even increase beyond the mormal level shortly before the occurrence of a large aftershock which might be comparable with the main shock, if we keep watch on the change of the aftershock activity immediately following the main shock_ Also the place of the large aftershock can be roughly predicted by following the hypocenter location of aftershocks occurring just before it. Our goal in this article is to compare tha above nonlinear method with recent RBF (Radial Basis Function network) functional approximation, in the neural network literature. These networks often used to predict the outcome of a future event based on current observations of the state of the environment