Insights into driving mechanisms of volcanic seismic high frequency tremor above 10 Hz on Mount Etna

Seismic tremor plays a crucial role in eruption forecasting and is therefore monitored extensively on volcanoes around the world. However, the use of volcanic tremor for eruption forecasting purposes requires improving our present understanding of its source processes. This has proven a challenging task.
Traditionally, the generation of volcanic tremor is attributed to processes associated with magma transport or linked to fluid-induced resonance (e.g. gases or hydrothermal systems) within the volcano plumbing system. In contrast, other studies suggest that fluids may not be required to generate tremor but the weak, unconsolidated, materials that make up volcanic edifices can experience diffusive failure patterns causing non-localised, low-amplitude seismic events merging into tremor. Small departures from the background stress levels would be sufficient to generate low-amplitude, small-stress-drop events for materials near the brittle-ductile boundary that still support seismicity, as demonstrated by numerical models and laboratory experiments. Changes in stress could be caused by variable magma flow or gas influx or simply linked to gravity impact on the edifice. Even if magma flow or gas influx drive stress level changes the subsequent failure of material would be dry mechanically.
Here, we investigate high-frequency tremor, in the frequency band 10-20 Hz, from data recorded on the summit of Mount Etna during a large seismo-acoustic deployment during the summer of 2022. High-frequency seismic signals, with energy at frequencies >10 Hz, experience rapid attenuation and are affected by extensive scattering making their analysis particularly challenging. We show how insights into the driving mechanisms of the episodic, high-frequency, tremor at Etna can be gained from the analysis of the seismo-acoustic energy ratio, which shows significant variations across different tremor episodes; this suggests different conditions for tremor generation. Additionally, we are able to locate the high-frequency tremor using multi-array beamforming and 3D grid-search algorithms; our results reveal the presence of different source regions from where tremor is radiated, including areas associated with extensive degassing. We also carry out synthetic tests to assess the reliability of the localisation results. Finally, frequency-magnitude distribution of tremor episodes is explored to investigate the hypothesis that tremor may result from sequences of multiple small-magnitude, very small-stress-drop, individual seismic events.