Enhancement of Seismic Phase Identification using Polarization Filtering and Array Analysis

Single-station polarization analysis allows us to extract wave parameters, such as inclination, azimuth, and ellipticity angle, directly from a recorded seismic signal theoretically. In reality, however, seismic data are not purely polarized in the finite analysis window due to varying noise levels, complex wavefield interactions, and calibration errors. Hence, this would potentially influence the observation window of phases of interest. In order to minimize these systematic errors, the involvement of arrays and array processing techniques can further increase the signal-to-noise ratio of coherent signals in a wavefield, which allows us to identify different seismic phases, especially the weaker phases that are usually difficult to observe in a single waveform, even after filtering for a desired wave type. In this study, we present a new approach that combines polarization analysis and filtering in the time-frequency domain using the S-transform with conventional array analysis such as beamforming to enhance seismic signals and distinguish different phases based on their expected slownesses and backazimuth. We apply this approach on AlpArray data and demonstrate wavefield separation in vespagrams using various polarization filters. We also discuss the benefits of our approach especially on small amplitude inner core phases (e.g., PKIKPPKIKP) and their applications for advancing seismological study of Earth’s inner core.