Discrimination of local seismic events in Panama by means of higher order statistics

The analysis of seismicity need to be restricted to earthquakes. The estimated seismic energy release and the spatial distribution of seismic events will be erroneous and biased if the data catalogue includes artificial events such as explosions. Therefore, it is necessary to separate earthquakes and explosions prior to the compilation of the seismicity in a particular region. In the canal area of Panama, the events are equally distributed with signal energy contents in the same range, making localization and magnitude estimation not effective as discriminators. In this study we use a method based on master-event correlation's for filtering out the explosions in the seismological catalogue. We use a library of time series with the a priori knowledge of the source type. The method is based on comparison of unknown events with the library by second, third and fourth order cross-correlations. Studied time series are divided into three parts, the first phases of the P and S-wave and a window from the preceding noise. Twenty known and twenty-one unknown events are tested with master-event correlation in second, third and fourth-order domain. The second-order based method discriminate correctly 40% of the known events while the third and fourth-order based correlations succeed in 75% respectively 80% of the known cases.