The source mechanisms of low frequency seismic events on volcanoes

Volcanoes generate a variety of seismic signals. One specific type, the so-called low frequency (LF) event, has proven to be crucial for understanding the internal dynamics of the volcanic system. While many endeavours have concentrated on the nature and cause of the seismic coda, the actual trigger mechanism of these events is still poorly understood. Several conceptual source models have been developed, ranging from magma-water interaction, stick-slip motion of magma plugs, magma flow instabilities, repeated release of gas-ash mixtures into open cracks, magma wagging, to brittle fracturing of magma. All but one trigger model, namely brittle failure of magma in the glass transition as response to the upwards movement of magma, fail to explain all observed characteristics of LF volcanic seismicity. Here, a spatially extended source, the ring fault structure, is developed to mimic the proposed source mechanism. The extended LF source is modelled as an arrangement of 8, 16 and 32 double couples (DCs) approximating a 30 m, 50 m and 70 m wide circular ring fault bounding the circumference of the volcanic conduit. Due to (partial) destructive interference, P-wave amplitudes of a ring fault structure are greatly reduced compared to single double couples and compensated linear vector dipoles (CLVDs), by 350% and 470%, respectively. It is shown here that these seismic amplitude differences may result in the underestimation of average slip and thus magma ascent rate by a factor of up to 3.5 when using an over simplified point source. To resolve the driving forces of LFs, synthetic seismograms representing both point and spatially extended sources were inverted for the apparent physical source mechanism using moment tensor inversion techniques (MTI). If original input parameters were unknown, MTI results of the ring fault would indicate a combination of 63% isotropic and 37% CLVD components. The proposed moment tensor strongly resembles that of a real CLVD case. The results of this study give evidence that slip along the conduit walls yields the same MTI results as a subhorizontal tensile crack, and the importance of knowledge about the source nature becomes eminently significant. Spatially extended source geometries describe an alternative to point dislocation sources. Additionally to the ring fault structure, this study provides a catalogue of further complex source scenarios involving new spatially extended sources, such as slip along a dike, conduit segments, two simultaneously acting ring fault structures, and helix-like flow patterns. P-wave amplitudes and waveforms vary largely with source geometry, stressing that source geometry is key for source interpretations and thus it is not sufficient to assume a point source nature of the processes involved to generated the observed seismicity.