A matched-filter technique with an objective threshold

Area of Science:

• Geophysics and Seismology
• Signal Processing
• Statistical Analysis

Background:

• Detecting seismic signals in continuous waveform records is crucial for seismological studies.
• Traditional methods often rely on subjective thresholding, leading to potential inaccuracies.
• Objective and automated approaches are needed for robust seismic event identification.

Purpose of the Study:

• To develop an objective method for determining thresholds in cross-correlation coefficients for seismic signal detection.
• To automate the detectability assessment using Akaike's Information Criterion (AIC).
• To validate the method's efficacy on extensive continuous seismic waveform data.

Main Methods:

• Empirical distribution analysis of cross-correlation coefficients among seismic waveforms.
• Objective threshold determination guided by Akaike's Information Criterion (AIC).
• Application of extreme value statistics to model the distribution of maximum cross-correlation coefficients.

Main Results:

• The proposed method successfully detected seismic signals from two years of continuous waveform records.
• Maximum network cross-correlation coefficients were found to follow extreme value statistics.
• A parametric probability density function of maxima was derived, enabling a reasonable outlier criterion.

Conclusions:

• The developed objective thresholding method provides a robust framework for seismic signal detection.
• Integration of AIC and extreme value statistics enhances the reliability of outlier identification.
• This approach offers an automated and data-driven solution for analyzing continuous seismic data.