Time series analysis of earthquake sequences by means of information recognizer

Abstract

Three seismic sequences of several thousand earthquakes each are analyzed by means of a tunable information recognizer known as wlzip. These sequences are different both in the geographical coverage and the time span, including earthquakes of magnitude larger than 8.0. The main variable under scrutiny here is the time interval between consecutive events. Two parameters (mutability and interval dilation) are defined for each sequence, which relate to the information contained in it. In this way it is possible to characterize different regimes in the seismic activity. For instance, mutability increases before large earthquakes and decreases sharply immediately after each of these events. On the other hand, interval dilation reaches a clear maximum several months before major earthquakes, while it decreases to its lowest possible value after such earthquakes during the aftershock regime. Extensions of the application of this new method to other problems in seismicity are mentioned.