Enhancement of Seismic Phase Identification Using Polarization Filtering and Array Analysis

ABSTRACT Seismic arrays play a crucial role in identifying weak signals in the seismic wavefield based on their expected slowness and backazimuth values. However, their resolution power is limited when studying phases with similar horizontal slownesses and arrival times, such as receiver-side or source-side reverberations and converted phases. Therefore, we investigate the benefit of applying polarization filtering to three-component seismograms before stacking to remove undesired signals and increase the signal-to-noise ratio of the array. Customized polarization filters enable more sophisticated wavefield separation and robust phase identification on vespagrams. However, selecting the suitable polarization filter requires a balance between noise reduction and the preservation of desired signals. We find that degree-of-polarization filters generally excel in suppressing incoherent noise. On the other hand, some filters, for example, based solely on ellipticity, do not yield notable enhancements for body waves and may even produce adverse effects, specifically for phases that arrive late in the seismogram. We demonstrate these findings using data recorded by AlpArray and surrounding permanent stations.