Statistical analysis op earthquake data

Abstract

The Statistical properties of earthquake data from 14 different areas have been studied by considering the earthquake occurrence as a one-dimensional stationary point process. A review of the main properties of the Point processes is given and some counting and interval properties of the mutually exciting processes are derived. As a result of an exploratory analysis the Poisson and renewal models are not found adequate to describe the earthquake occurrence and any kind of periodicity is not well established. The Meymann-Scott model with mixed exponential decay is found more suitable than the one with single exponential for describing the earthquake occurrence. The fit of the mutually exciting process is as satisfactory as the fit of the Weymann-Scott with mixed exponential from the spectral analysis viewpoint but not from the viewpoint of the interval analysis. As a result of the interval analysis a four-variate mutually exciting process is proposed for describing the earthquake occurrence, which also takes into account the differences according to depth. Finally an attempt to classify the areas under investigation is made and some ideas about the study of the earthquake phenomenon as a multidimensional point process are put forward.