Figure 5. Best fit afterslip only models for published coseismic models (a) Elliott et al. (2022)model. (b) Liu et al.(2022) model (c) Ye et al. (2022) model. The error ellipse show 95% confidence. The region of the rupture areas is > 1m slip. Dashed light grey lines outlines the depth contours from the Slab2 model (Hayes et al., 2018). The white barbed line shows the plate boundary between the Pacific plate and the North American plate.
the goodness of the post-seismic fit, we the utlilize near-field GPS continuous sites except for site AC13, while focus on three peninsula sites (AB13, AC21, and AC40), for these three sites have large displacements and are most sensitive to the downdip distribution of afterslip. We are not evaluating the fit to site AC13 for two reasons. One is that it is not sensitive to the downdip afterslip (see the downdiponly afterslip model, Figure S2 a), and the other is that it is located at the edge of the rupture area, and its fast seaward motion might also include a contribution from other post-seismic mechanisms such as poroelastic rebound, which is beyond the scope of this paper. The sites the Shumagin islands (AC41, AB07, AC28, and AC12) have relatively small displacements and might be affected by the postseismic processes of the July 21, 2020, Mw7.8 Simeonof earthquake, or by alternative assumptions about the distribution of velocity-strengthening material, so we will consider multiple misfit metrics to determine the best model.
The Ye et al. (2022) model differs from the other models in ways. They assumed a deeper than the Slab2.0 geometry used by others, based on a seismic reflection study (Kuehn, 2019). Also, their slip model is more compact in the downdip direction, as a result of an assumption they made about the maximum possible depth of slip. Additionally, they added a patch of shallow slip near Chirikof Island in order to better explain the tsunami. This added slip patch, which was added to the slip model by those authors after their initial slip inversion, also makes their model predict the AC13 postseismic data better.
We estimated a new coseismic model by shifting the Elliott et al., (2022) model fault plane to be 6.5 km deeper, similar to the Ye et al. (2022) model geometry, but found that the coseismic slip (Figure S3i) and post-seismic displacement patterns (Figure S4) did not change much. However, when we tested a narrower fault model, by restricting the downdip extent of the model fault plane, we found that the fit to the postseismic data improved dramatically while the fit to the coseismic observations was nearly unchanged (Figure S3 b, Figure 6 b). These tests indicated that it is the narrower downdip width of the slip distribution that makes the Ye et al. (2022) model be a better of the postseismic afterslip. A narrower fault model with a more abrupt decrease of the coseismic slip at the downdip end of the rupture plane results in afterslip being located farther offshore.
(a) (b)