Locating volcanic tremor using azimuth coherence of cross-correlation

Accurate localization of volcanic tremors can provide important information about volcanic activities. However, the seismic waveform of volcanic tremor is complicated, resulting in that its location cannot be accurately determined from traditional traveltime inversion methods due to the unidentifiable first arrivals. The cross-correlation-based back-projection method, which stacked the back-projected spatial images utilizing the cross-correlation data from different station pairs, is considered as one of the most effective waveform stacking location methods for the volcanic tremor. In this method, the obtained coherence from different station pairs can improve the signal-to-noise ratio of projected energy distribution, and then constrain the potential source location. Nevertheless, not all back-projected images of station pairs are helpful to enhance the positioning effect in practice because the resolution of the energy peak in the back-projection method is affected by the selection of used station pairs. In this study, we optimized the back-projection method by establishing some criteria for the selection of the station pairs, including the distance between two stations in each pair, and the calculated azimuth angle, the cover area, and the central location between each two of all station pairs. In addition, the phase information in cross-correlation data has been also used in our method to limit the energy divergence in the resulting map. Taking the Katla volcano as a real example, we demonstrate that the proposed second-order azimuth coherent location method can increase the accuracy of locating volcanic tremors compared to the traditional methods, resulting in an improvement of approximately 70 % in location resolution.