3:30 PM - 3:45 PM
[SSS05-06] Complex rupture propagation of the MW 7.7 2023 southeast of the Loyalty Islands earthquake
It’s been recently reported that an earthquake rupture can become irregular, involving such as back-propagating rupture and consecutive ruptures across different fault segments, especially when it occurs in the oceanic crust. For earthquakes with those rupture propagations, it is difficult to construct a source process model that explains the observed waveforms in a conventional finite-fault inversion scheme that uniquely prescribes the rupture direction and fault geometry. One such candidate earthquake is the 2023 Southeast Loyalty Islands Earthquake (MW 7.7), which occurred on 19 May 2023 in the outer rise of the curved Vanuatu subduction zone at the boundary between the Indo-Australia and the Pacific Plates. The U. S. Geological Survey constructed a source process model of the 2023 Southeast Loyalty Islands earthquake using the finite fault inversion, which assume the expanding rupture direction and the unified fault geometry. However, the theoretical waveforms from their model do not explain the features of the observed waveforms, which suggests that the 2023 Southeast Loyalty Islands earthquake may include an irregular rupture propagation that cannot be represented by the assumption of a simple rupture scenario. We here apply the recently developed Potency Density Tensor Inversion (PDTI) to the teleseismic P waveforms of the 2023 Southeast Loyalty Islands earthquake to discuss the source process in detail.
The PDTI introduces the uncertainty in the Green's function to stably estimate a high degree of freedom source process model that includes information of fault geometry. We allow a model to freely release the potency after a virtual rupture front is reached at each point of the model plane until a total rupture duration so that we can represent potential irregular rupture behaviors such as the back-propagating rupture. The PDTI does not require fault geometry assumption (fault plane) because the fault slip is projected on a model plane. However, if the adopted model plane deviates significantly from the fault plane, modeling errors will increase. Thus, we set the model plane with a strike of 280° and a dip of 48° parallel to the trench axis with reference to the Global Centroid Moment Tensor solution and the aftershock distribution.
The result shows that the rupture propagates southeast from the epicenter until 6 s, and then propagates bilaterally toward east and west. The eastern propagation first avoids a region located 30 km east of the epicenter (region B1) during 6–10 s, but eventually ruptures the B1 during 10–20 s, where we observe the highest potency-rate centered on the B1. The western propagation ruptures a region extending parallel to the trench axis (region B2) 5 km south of the epicenter during 10–16 s, and then terminates. A normal fault mechanism is obtained throughout the entire rupture, but different strikes are obtained for each rupture episode as discussed below. The theoretical waveforms calculated by our model explain the observed waveforms well.
Our result shows that the B1 is not immediately ruptured, but is eventually ruptured after the minor rupture encircles, suggesting that the B1 can be acted as a barrier with high strength, and the rupture that apparently avoids the B1 during 6–10 s may have led to a sufficient concentration of shear stress, leading to the delayed rupture of the B1. The changes in strike during the B2 rupture are observed at 260°~280°during 6–12 s and 280°~300° during 12–18 s. The strike changes associate with the timing where the rupture front switches its direction from west to northwest from the epicenter, which shows an apparent back-rupture propagation. The strike changes and the associated switch of the rupture front suggest that the B2 rupture can be made through the different fault segments with different geometry.
The PDTI introduces the uncertainty in the Green's function to stably estimate a high degree of freedom source process model that includes information of fault geometry. We allow a model to freely release the potency after a virtual rupture front is reached at each point of the model plane until a total rupture duration so that we can represent potential irregular rupture behaviors such as the back-propagating rupture. The PDTI does not require fault geometry assumption (fault plane) because the fault slip is projected on a model plane. However, if the adopted model plane deviates significantly from the fault plane, modeling errors will increase. Thus, we set the model plane with a strike of 280° and a dip of 48° parallel to the trench axis with reference to the Global Centroid Moment Tensor solution and the aftershock distribution.
The result shows that the rupture propagates southeast from the epicenter until 6 s, and then propagates bilaterally toward east and west. The eastern propagation first avoids a region located 30 km east of the epicenter (region B1) during 6–10 s, but eventually ruptures the B1 during 10–20 s, where we observe the highest potency-rate centered on the B1. The western propagation ruptures a region extending parallel to the trench axis (region B2) 5 km south of the epicenter during 10–16 s, and then terminates. A normal fault mechanism is obtained throughout the entire rupture, but different strikes are obtained for each rupture episode as discussed below. The theoretical waveforms calculated by our model explain the observed waveforms well.
Our result shows that the B1 is not immediately ruptured, but is eventually ruptured after the minor rupture encircles, suggesting that the B1 can be acted as a barrier with high strength, and the rupture that apparently avoids the B1 during 6–10 s may have led to a sufficient concentration of shear stress, leading to the delayed rupture of the B1. The changes in strike during the B2 rupture are observed at 260°~280°during 6–12 s and 280°~300° during 12–18 s. The strike changes associate with the timing where the rupture front switches its direction from west to northwest from the epicenter, which shows an apparent back-rupture propagation. The strike changes and the associated switch of the rupture front suggest that the B2 rupture can be made through the different fault segments with different geometry.