Large earthquakes generate aftershocks, the spatiotemporal patterns of which reveal details of how the strain energy in the Earth’s crust relaxes following a main rupture. Prior research has established that aftershocks correlate with regions of low co-seismic slip – slip which occurs at the instant of an earthquake – and of increased shear stress. Aftershocks occur preferentially in relatively low mainshock slip areas, and near the periphery of large slip asperities. However, researchers still require more detailed quantitative studies of earthquake triggering to allow better forecasting. In understanding the relationship between aftershocks and mainshocks, one key analytical challenge in particular is to exclude those aftershocks which are triggered by other aftershocks, and focus only on those triggered by the mainshocks. The most fruitful comparisons should be made between these “directly triggered aftershocks” and mainshock slip distributions.
In a new paper, LML External Fellow Jiancang Zhuang and colleagues pursue this end by using a new earthquake clustering model to disentangle overlapping family trees among earthquakes. In the study, they fit the statistical 3D-FS ETAS model – a generalisation of the standard Epidemic-Type Aftershock Sequence (ETAS) model – to the Southern California earthquake catalogue. When estimating the triggering relationships between mainshocks and aftershocks, this model incorporates the 3D fault geometries of large earthquakes. In the analysis, four mainshocks with magnitudes larger than M6.5 were taken as finite sources, which include three strike-slip events: The Landers, Hector Mine, Ridgecrest earthquakes, and one thrust event: the Northridge earthquake. The authors use early aftershocks as constraints for source boundaries of mainshocks, and introduce an iterative algorithm to estimate the spatial distribution of aftershock productivity within the mainshock source.
Compared to earlier modelling results, the 3D-FS ETAS model corrects underestimated aftershock productivity of large mainshocks. For an earthquake of magnitude 7.0, the model predicts a five-fold increase in the expected number of direct aftershocks. The reconstructed direct aftershocks occupy 21% to 41% of all the events within the source regions after the mainshocks. Aftershock productivity within source regions shows a strong spatial heterogeneity with high values mainly distributing in depth layers shallower than 15 km. Aftershocks of the three strike-slip mainshocks mostly distribute along the rupture faults, particularly concentrating at the ends of mainshock ruptures. For the Northridge earthquake, many aftershocks composed of all faulting types occur in the shallow layers.
In summary, the new study demonstrates that the 3D-FS model enhances the triggering ability of large earthquakes, and offers an improved ability to reproduce directly triggered aftershocks within the mainshock ruptures. By incorporating the constraints from early aftershocks and co-seismic slip distributions, the authors suggest, this model should help to improve aftershock forecasting performance.
The paper is available at https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2020JB020494