4.3 3-Month AfterslipOnly Models with Different Fault Width Assumed
Because even the viscoelastic models with maximum displacement have relatively small amplitudes compared to the data, it is useful to start by considering afterslip-only models. The models with the minimum potential viscoelastic relaxation contribution subtracted are equivalent to aftersliponly models, because the viscoelastic relaxation contribution for that viscoelastic model is negligible.
The first important question to address for the afterslip-only models is where the afterslip allowed to occur The edge of the west portion of the Chignik earthquake rupture plane is adjacent to the coseismic rupture area of the Mw7.8 Simeonof earthquake. If that part of the fault has velocity-weakening friction, there should be no afterslip allowed within the coseismic rupture region of the S earthquake. However, we find that allowing afterslip there or not does not significantly affect our predictions the sites outside of the Shumagin Islands (Figure S6). Thus, we first considered three scenarios of afterslip forward models: allowing the afterslip to occur up-dip only, down-dip only, and fully surrounding the coseismic rupture zone, using the coseismic model of Elliott et al. (2022), and we focus on the predicted displacements on the Alaska Peninsula. None of these models predict the displacement of site AC13 well, although the Ye et al. (2022) model with its large patch of shallow slip very close to AC13 does fairly well
Figure S2a shows that the downdip-only afterslip model can easily explain the displacements along the Alaska Peninsula. Downdip afterslip contributes almost nothing to the signal at AC13, which must be explained by some combination of updip afterslip and perhaps poroelastic relaxation. Figure S2b shows that the updip-only afterslip model predicts displacements at Peninsula sites AB13 and AC40 that are much smaller than the observations, because the total stress change is not able to generate enough afterslip to match the observed displacements there. Thus, for simplicity our preferred model is that afterslip is allowed to fully surround the coseismic rupture zone. Changing the updip frictional parameters would have only a minimal impact on our model predictions, except at AC13.
We then use the coseismic rupture models with different assumed fault widths that we obtained from section 4.1 as input for a series of afterslip-only models. We find that using the coseismic model with a fault width of 10km significantly reduces the azimuthal misfit of the two peninsula sites AB13 and AC40 with the best fit model having a minimum reduced \(\chi^{2}\) of 16.27 when considering all data (Figure 8). The same model has the minimum misfit whether we consider only the Peninsula sites or include the Shumagin sites as well. A slightly narrower model (110 km width) minimizes the angular misfit of the displacements for AB13 and AC21. The frictional parameters we find for this preferred model are a \(\sigma\) =0.60Ma, \(V_{0}\) = 0.8m/yr. Figure shows the best fit afterslip only models for coseismic models with different fault widths. The postseismic fit is hugely improved by reducing the fault widths, so that afterslip occurs further offshore (Figure ). This indicates that the postseismic give new constraints to the coseismic slip model.