Deep long-period earthquakes (DLPs) are often detected near volcanoes from the crust down to the upper mantle. Exhibiting coincidence with some volcanic eruptions, DLPs are recognized as potential precursors to volcanic activities yet their detection remains challenging. Meanwhile, their relation to volcanic activities and specific source mechanisms remains uncertain. In this study, we first classify earthquakes into volcano-tectonic (VTs) and long-period (LPs) earthquakes at 10 Alaskan volcanoes by frequency index (FI). The consistent differences in FI distribution between LPs and VTs across volcanoes suggest it can be leveraged for automatic classification. Presence of both VTs and DLPs in some regions suggests frequency content differences mainly arise from source effect. We then analyze 12 years of continuous waveforms by template matching and detect 20,979 DLPs. No repeating DLPs are identified below the depth of 10 km except ~1% of DLPs at Spurr Volcano, indicating source processes of DLPs at Alaskan volcanoes are primarily non-repetitive. Detections of mantle DLPs and positive correlations between DLPs and volume change rate in the magma reservoir support the involvement of magmatic activities in some DLP sources, though there is limited correlation in DLP depth distribution with magmatic water content and kinematic parameters of plate motion. In addition, only ~12% of DLP bursts occur simultaneously with VT bursts, suggesting most DLP rate anomalies are not directly linked to shallow volcanic unrests. Finally, since on average only ~26% of DLP rate anomalies coincide with eruption episodes, it is still challenging to foretell eruptions uniquely based on DLP occurrences.

Both volcano-tectonic (VTs) and deep long-period earthquakes (DLPs) have been documented at Akutan Volcano, Alaska and may reflect different active processes. In this study, we perform high-resolution earthquake detection, classification, and relocation using seismic data from 2005-2017 to investigate their relationship with underlying magmatic processes. We find that the 2,787 VTs and 787 DLPs are concentrated above and below the shallow magma reservoir respectively. The DLPs’ low-frequency content is likely a source instead of path effect considering its uniformity across stations. Both VT and DLP swarms occur preferentially during inflation episodes with no clear migration. However, the largest VT swarms occur during non-inflating periods, and only VT swarms contain repeating events. Therefore, we conclude that the VTs represent fault rupture triggered by magma/fluid movement or larger earthquakes, while the DLPs are directly related to unsteady magma movement through a complex pathway or represent slow fault ruptures triggered by magma movement.