Abstract:
Several source models have been proposed to explain the enigmatic 2009 Tonga-Samoa earthquake. The long-period data require a composite source model and can be fit with a normal-faulting subevent followed by one or more reverse-faulting subevents. The short-period data, in contrast, indicate a more compact rupture pattern around the epicenter. The lack of a unified source model reflects the complexity of the event. We analyze the spatiotemporal evolution of this earthquake with P wave back-projectionfromgloballydistributedstationsindifferentfrequencybands(lowfrequency:0.05–0.2Hz, high frequency: 0.2–2 Hz) and a multiple moment tensor inversion. The rupture propagation revealed by back-projection exhibits frequency-dependent behavior, with two branches of high-frequency-enriched bilateral rupture around the epicenter and a high-frequency-deficient rupture branch at the subduction interface. A composite source model with one Mw 8.0 normal-faulting earthquake east of the trench axis (seaward) followed by one Mw 8.1 reverse-faulting earthquake along the subduction interface west of the trench axis (landward) can explain the very long period data (200∼500 s). Combined with high-resolution swath bathymetry data, the back-projection images show that the azimuth of rupture branches east of the trench axis were controlled by the geometry of bending-related faults on the Pacific plate and that the rupture branch west of the trench axis may correlate with the along-strike fore-arc segmentation. The rupture along the subduction interface was triggered by the seaward rupture and a partially subducted normal fault may have played a key role in facilitating the triggering. The apparent normal-reverse faulting interactions pose a higher seismic risk to this region than their individual strands at the northernmostcorneroftheTongasubductionzone.
