In this work, an advanced machine learning technique named diffusion maps is applied for array-based earthquake-explosion discrimination. We rely on prior work that utilizes the diffusion map-based discrimination approach for data collected from a single seismometer. The discrimination task is an essential component of the Comprehensive Nuclear-Test-Ban Treaty verification regime and since many of the International Monitoring System (IMS) stations consist of arrays, the extension to array based processing is of interest. The proposed method includes a pre-processing step, which constructs time–frequency representations of the P-wave and S-wave seismograms followed by a non-linear dimensionality reduction step. Discrimination is performed in the low-dimensional space. The performance of the presented algorithm is demonstrated on a data set from Southern Israel, recorded at the IMS seismic array of Mt. Meron (MMAI). We show that the diffusion maps-based approach enables to enhance the discrimination capabilities of seismic arrays, even when processing low-magnitude events.
Bibliographical notePublisher Copyright:
© 2020, Springer Nature Switzerland AG.
- Seismic array
- diffusion maps
- machine learning
- seismic discrimination
- seismic monitoring and Test-Ban Treaty verification