Complex seismic sources in volcanic environments: Radiation modelling and moment tensor inversions

Long period (LP) signals are special seismic events observed at volcanoes, which comprise both a high frequency onset due to brittle failure and a more energetic low frequency part due to resonance in a fluid-filled conduit. They are critical for volcano monitoring since they can be used as a volcanic forecasting tool. Classic seismology assumes planar faults for seismic sources; however, there is increasing evidence that suggests different fault shapes such as dyke faults and ring faults. We consider in this study narrow dykes and conduits rather than large calderas, hence, we model these complex sources by superposing vertical single double couple (DC) sources arranged along narrow fault structures with inner upward movement. We calculate seismic radiation patterns and synthetic seismograms for a rupture along a dyke, three different partial ring ruptures and a full-ring rupture. Results show that planar faults are the most effective at radiating energy. The more the source geometry deviates from a planar fault the smaller become the amplitudes and therefore the Moment Magnitudes. For example, the amplitudes decrease to 2.4% of the planar radiation for a full-ring rupture and to 0.7% for a dyke rupture. The waveforms produced by partial ring ruptures are in accordance to what is expected in the far field, representing the derivative of the source displacement and emulating radiation of a DC with different azimuths; however, the dyke and full-ring sources produce waveforms that appear to represent the second derivative of the source displacement and negative first onset polarisations. Moment Tensor Inversions support similarities between DC ruptures and partial ring ruptures; however, they show ambiguous solutions for the other sources. This point source assumption can lead to misinterpretations of slip history on the fault and a consistent underestimation of magnitudes which has direct implications for magma ascent estimations derived from seismic amplitudes.