A Real-Time Discrimination System of Earthquakes and Explosions for the Mainland Spanish Seismic Network

Resumen

Different waveform-based discrimination parameters were tested using multivariate statistical analysis to develop a real-time procedure for discriminating explosions from earthquakes at regional distances in the Iberian Peninsula. This work enabled a purge of the Spanish National Seismic Catalogue for the period 2003-2014. The training data consisted of waveform-based signal properties in the time and frequency domain for events (earthquakes and explosions) recorded during the selected time period by the Spanish Broadband National Network and Sonseca short-period Array of the Instituto Geografico Nacional (IGN). For each station and its associated training dataset, a discriminant function was defined as a linear combination of the measured variables. All station-specific discriminant functions were then combined with a weighting scheme to test the training events, revealing that 86 % of the events were consistent with the analysts' judgement. The application of this method to the whole of the IGN's seismic database for the studied period gave an 83 % success rate; however, a 91 % success rate is reached if events are classified using at least three stations and 100 % confidence levels.