← Back to Sessions

Stress Shadows in Physics-based Forecasts of Aftershock Locations

Description: 

The spatial distribution of aftershocks is an important component of earthquake forecasting following a mainshock, and in current operational forecasts an empirical decay with distance from the mainshock is often used. In physical models, in contrast, aftershock triggering is commonly attributed to static Coulomb stress changes from the mainshock, in combination with rate-and-state friction, which produces spatial kernels with multiple lobes of forecasted seismicity rate increase and decrease. However, the static Coulomb stress change kernels usually don’t outperform a simple decay with distance in testing. One challenge is that some aftershocks occur in “stress shadows”, where a decrease in static Coulomb stress is modeled to suppress earthquake occurrence. We investigate what causes aftershocks in the stress shadows, and how to improve the physics-based spatial kernels.

We examine several hypotheses that reconcile the aftershocks in stress shadows with the static Coulomb stress change model, including inaccuracy in modeling, effects of receiver fault complexity, variability in fault friction, secondary triggering from prior aftershocks or afterslip, and continuing background earthquakes. When tested on standard network catalogs and machine-learning focal mechanism catalogs for the 2016 Kumamoto, Japan, and 2019 Ridgecrest, California, aftershock sequences, none of these hypotheses can explain most of the aftershocks in the stress shadows, and taken together they can explain only about half of these aftershocks.

Alternatively, we consider dynamic stress triggering from the passing seismic waves. We find that the spatial and temporal distribution of aftershocks in the stress shadows are consistent with the expectations of dynamic triggering: the aftershocks mainly occur in a burst over the first few days to weeks, and decay with distance like near-field body waves. We find that a hybrid spatial kernel, combining static Coulomb stress lobes with dynamic-stress decay with distance in the stress shadows, performs better than a static-stress-only kernel in pseudo-prospective testing.

Session: Improving the State of the Art of Earthquake Forecasting Through Models, Testing and Communication - I

Type: Oral

Date: 4/15/2025

Presentation Time: 02:15 PM (local time)

Presenting Author: Jeanne

Student Presenter: No

Invited Presentation: 

Poster Number:

Authors