Design and Development of the Intelligent Event Identification System: DesignConsiderations and Processing for Regional Event Identification

摘要

This report describes the design of the Intelligent Event Identification System(ISEIS) and the key processing algorithms of the system. ISEIS will characterize seismic events utilizing discriminants to classify source types, explosion or earthquake, and by comparing waveform features with reference events to estimate how similar the events are to previously observed events. Various graphics interfaces will provide explanations to the analyst about the event characterization results as well as various graphics displays of the data processing results. ISEIS will initially be developed to identify small events at regional distances, recorded at the regional arrays. A discrimination study of regional discriminants at NORESS for mine explosions and earthquakes in Norway is presented. This study has revealed that ratios of regional-phase amplitudes at frequencies above 6 Hz provide the best discrimination capability. Spectral discriminants, such as Lg spectral ratio, appear to be less useful. However, the identification of time-independent spectral modulations effectively identifies most mine explosions which are ripple fired. A process called Multiple Event Recognition System (MERSY) is described which automatically identifies ripple-fired explosions using cepstral methods. Finally, the Dynamic Time Warp (DTW) algorithm is described which will be used in ISEIS to do waveform matching of regional seismic codes.