The best model that allows for slip on one plane prefers the slip to be located above the upper portion of the rupture zone. The data do not permit slip on the downdip extension. If we restrict the slip such that it occurs only on the downdip extension of the fault, the misfit of the residuals (c2/dof) is large (~ 5). If we free up various parameters, the slip is either located above 18 km, or the fit is unrealistic (e.g., extremely large slip on an extremely narrow fault). Similar results occur for slip on the updip extension of the fault; the afterslip does not extend updip from the rupture plane.
In a second set of inversions, we solved for an auxiliary fault while fixing and freeing various parameters. This heuristic approach was dictated by the limited degrees of freedom and data strength afforded by the sparse data set. In each of these inversions nearly pure thrust afterslip occurs on one fault that fits the entire mainshock rupture plane. No slip extends above or below the rupture plane, and with the addition of a second fault the slip does not extend beyond the rupture plane laterally. We present the best model in this paper; however, it should be taken as broadly representative of the possible postseismic mechanisms. We experimented with dozens of models by varying the initial values and fixed parameters. Afterslip on the main fault and additional slip on an auxiliary fault emerge as the best fitting models. The auxiliary fault does not correspond to any mapped fault but may rather be indicative of general deformation of the upper crust as a result of the mainshock. This "fault," which is more likely representative of broad deformation in the upper crust, coincides with shallow aftershocks that are also interpreted as deformation of a quasi-elastic material (Unruh et al., 1997).
