Text S3. Mean Stress and Stress Transfer

As an indicator of elastic strain energy storage and release, we calculate gridded mean stresses (σm) throughout the system at different stages (Morgan, 2015; Wang and Morgan, 2019) of the simulation. The stress transfer and the corresponding energy release during earthquake unloading is approximated by plotting the cumulative change in σm between State 2 and State 1 of the simulation (Figure S3). A reduction (negative in blue) in σm indicates the release of elastic strain energy, whereas an increase (positive in red) indicates an increase in energy. No change in stress is indicated by white.
To demonstrate the role of the frictionally stronger outer wedges in modulating fault slip, we plot corresponding model-derived stress changes that accompanied the simulated earthquakes (Figure S3) for each selected models (Figure 3). In all cases, the earthquake causes a reduction in inner wedge stress, some of which is transferred toward the toe, resulting in an increase in stress within the outer wedge.
For the Maule rupture segment, the simulated coseismic stress change (Figure S3a) shows that unloading of the inner wedge transferred significant stress into the outer wedge, around 80 km, near the boundary with the inner wedge, whereas the toe of the wedge experienced essentially no change in stress. This demonstrates that the outer wedge resisted megathrust slip, limiting rupture propagation to the trench, which is consistent with expected velocity-strengthening behavior and observations (Contreras‐Reyes et al., 2010; Delouis et al., 2010; Moreno et al., 2010; Maksymowicz et al., 2017).
For the simulated coseismic stress change for the Valdivia rupture segment (Figure 3b), the wedge experienced primarily coseismic stress drop, except very near the toe of the wedge, where a stress rise is observed above the strong outer wedge fault. However, despite the outer wedge resistance, our simulations suggest that the toe experienced more than 30 m of slip (Figure 3b).