← Back to Sessions

Spatial Clustering of Aftershocks Impacts the Performance of Physics-Based Earthquake Forecasting Models

Session: Constructing and Testing Regional and Global Earthquake Forecasts III

Type: Oral

Date: 4/22/2021

Presentation Time: 10:00 AM Pacific

Description: 

I explore why physics-based models of earthquake triggering, such as those based on Coulomb stress changes, rarely outperform statistical models in prospective testing, outside of limited spatial-temporal windows. I consider the hypothesis that statistical models like the Epidemic Type Aftershock Sequence (ETAS) model are able to perform as well as physical models because of the tight spatial clustering of aftershocks. In many aftershock sequences, while most aftershocks may occur within the Coulomb-stress-increase lobes, the aftershocks do not uniformly fill these lobes, and instead cluster in a few locations. Pseudo-prospective tests on suites of synthetic aftershock sequences show that the level of spatial clustering of the direct aftershocks impacts the relative performance of physical versus statistical models. The synthetic sequences are generated from generalized “physical” triggering models, superimposed on background heterogeneity that controls the level of clustering. The statistical ETAS model performs relatively better the more clustered the direct aftershocks, while the true generalized “physical” model performs relatively worse. ETAS must successfully, although incorrectly, model the spatial clusters as secondary triggering. Real aftershocks appear to be sufficiently clustered to allow ETAS to perform as well as or better than physical models. A likely cause of the spatial clustering of direct aftershocks is heterogeneity of the background physical conditions, which typically isn’t modeled in physics-based forecasts. This implies that the forecast performance of physical models could be substantially improved through a better understanding of the interaction between earthquake stress changes and variable background physical conditions such as stress state, fault strength, and fluid pressure. The performance of physical models can also be improved through the incorporation of ETAS-like features, such as modeling secondary triggering from aftershocks.

Presenting Author: Jeanne L. Hardebeck

Student Presenter: No

Authors