Therefore, we investigated a series of alternative coseismic models with different model fault widths. In all cases, we fixed the up-dip limit of the fault plane to be the same as the Elliott et al. (2022) model, but varied the down-dip extent of the fault plane. We considered models with different downdip model fault plane widths of 100km, 110km, 120km, 130km, 140km, 150km, and 160km (the last is equal to that used by Elliott et al. (2022)). All models apply zero slip conditions beyond the edges of the model fault, and the bottom row of the subfaults also has a zero slip condition, so we report the width of the part of the fault plane that is actually allowed to slip. The models with narrower assumed fault widths force the slip distribution to be more compact and located farther offshore, and in general have a different character at the downdip limit of the coseismic rupture, with the models having narrower widths producing a more abrupt downdip limit of slip (Figure 7).
Figure 7. Average slip along strike of coseismic models with different fault width. Blue and green solid line outlines the comparison of the Elliott et al. (2022) model and our preferred coseismic model with 120km fault width.
We find that all of these models fit the coseismic data almost equally well (Figure S3 ag), indicating that the coseismic data alone do not constrain these details of the coseismic slip distribution, due to limited model resolution arising from the sparsness and the spatial distribution of the data. As discussed later in sections 4.3 and 4.4, our preferred coseismic slip model (Figure 6) has a model fault 10 km wide in the downdip direction, similar to the Ye et al. (2022) model. The narrower fault models do not predict the coseismic vertical displacements as well as the wider models, but our tests indicate that making the model fault deeper could improve the fit to the vertical (Figure S3h). However, for simplicity we continue to use the Slab2.0-based fault geometry in the rest of this study.