No abstract

The relative locations between mainshocks and their aftershocks have long been studied to characterize the mainshock-aftershock relationships, yet these comparisons may be subjected to biases inherited from various sources, such as the analysis method, data, and model parameters. Here, we perform both a relocation analysis of interplate events to obtain accurate relative centroid locations and a slip inversion analysis of the mainshock slip relative to the relocated events, with some of the relocated events used as empirical Green's functions, to retrieve the spatiotemporal slip features of the mainshock relative to all of the relocated events. We perform these analyses on the large ( 6.0 ≤ M W ≤ 7.3 ) interplate earthquakes that occurred near four giant ( M W ≥ 8.5 ) megathrust earthquakes: the 2007 M W 8.5 Bengkulu (Indonesia), 2005 M W 8.6 Nias (Indonesia), 2010 M W 8.8 Maule (Chile), and 2011 M W 9.1 Tohoku-Oki (Japan) earthquakes. Most of the spatiotemporal slip features of the mainshocks are consistently recovered using different empirical Green's functions. We qualitatively and quantitatively demonstrate that the large interplate aftershocks within 5 years of the four analyzed mainshocks are largely located on the periphery or outside of the large-slip regions of these four giant megathrust earthquakes.

Accurate earthquake locations are important for understanding earthquake processes. Recent advances in earthquake relocation methods have resulted in relative centroid locations with smaller uncertainties using cross‐correlation‐based methods. However, relative hypocenter locations, which are traditionally determined using the onset times of event waveforms, generally possess larger uncertainties than that of the centroids. Here, we develop a new cross‐correlation‐based relative hypocenter relocation method based on the network correlation coefficient method, whereby we incorporate an additional step to search for a suitable window to cross‐correlate the event waveform onsets. We then perform a joint inversion on the obtained relative hypocenter and centroid locations to directly compare their relative locations under a unified coordinate. We apply this method to three regions where repeating earthquakes have been detected, and reveal both the repetitive rupture of similar centroids by these repeating events and newly illuminated hypocenter hotspots that serve as the onset region of earthquakes that grow into different but selective magnitudes. Our joint analysis of hypocenter and centroid locations enables quick estimations of rupture propagation directions without the need to perform slip inversion analyses. We estimate the location uncertainties via bootstrapping, with the 95% confidence intervals yielding sub‐100 m horizontal uncertainties in both the centroid and hypocenter locations with the best station coverage and event availability. These results point to limited but existing correspondence between the hypocenter and centroid locations and the earthquake magnitudes, as well as complexity in the hypocenter distributions, even within repeating earthquake sequences whose rupture areas largely overlap.

No abstract