Spatial and temporal measurements from satellite geodesy observations provide fundamental constraints for investigating the mechanisms that control earthquake-related processes over the seismic cycle. The surface motions, in particular those time-dependent observed following large earthquakes (in the postseismic period), can be used to examine both the frictional behaviour of the fault and the rheological properties in the lithosphere-asthenosphere system by means of geomechanical models. During the postseismic period, afterslip on the fault interface and viscous relaxation in the lithosphere-asthenosphere system are the dominant mechanisms. However, the relative contribution of these processes is not entirely understood, which is primarily due to the challenge of modelling the rheological behaviour of the lithosphere-asthenosphere system following megathrust earthquakes. Accordingly, models of the postseismic period have commonly assumed the whole crust as an elastic material above a viscoelastic upper mantle with linear or non-linear (viscosity) rheology. In this dissertation, I integrate state-of-the-art geomechanical-numerical models, Global Positioning System (GPS) observations, and aftershock seismicity to investigate the underlying deformation processes controlling the postseismic deformation induced by the 2010 Mw 8.8 Maule earthquake in Chile. I particularly focus on investigating the fundamental discrepancies in the resulting postseismic deformation between linear and power-law rheologies. In contrast to previous works, I use, for the first time, a forward model considering temperature-dependent power-law rheology. Furthermore, I implement a novel approach to discriminate competing postseismic simulations, which incorporates the positive correlation between afterslip and aftershock activity. The resulting stresses from the coseismic and postseismic deformation transferred to the northern segment of the 2010 event rupture area, where the Mw 8.4 Illapel earthquake occurred in 2015, are also studied.
The geomechanical-numerical models consider constitutive equations to simulate the elastic and viscous rock responses of the crust and upper mantle. For the power-law rheology case, the spatial and temporal viscous distributions are modulated by the dislocation creep parameters and the temperature field in the crust and upper mantle. On the other hand, the linear rheology case consists of a material with homogeneous and linear viscosity in the upper mantle, while the whole crust is fully elastic. I employ the Finite Element Method (FEM) to solve the involved partial differential equations, for discrete elements representing the study area. The GPS observations and aftershock seismicity consist of published data spanning six years after the 2010 Maule event.
By using a two-dimensional (2D) model approach, my results reveal crucial variations in the modelled surface postseismic displacements due to the choice of model rheology. When comparing these results with the GPS data, I find that the GPS displacement patterns are notably better explained by the power-law rheology model, particularly the six-years cumulative uplift in the volcanic arc and the high-rate transient displacements in the first two years following the mainshock. The primary deviations in the cumulative patterns are produced because of the location of the viscous relaxation. In the linear rheology case, most of the viscous relaxation occurs in the continental mantle wedge, beneath the forearc. Conversely, in the power-law rheology case, the viscous relaxation mainly occurs in the lower crust, beneath the volcanic arc, due to dislocation creep processes.
In a subsequent study, I extend the 2D model approach to 3D. Here, I choose the preferred postseismic simulation from an innovative approach that accounts not only for the best fit to the GPS observations, but also incorporates the spatial correlation between inverted afterslip distributions and aftershock activity. My results reveal that afterslip inversions strongly depend on the choice of rheology, especially at greater depths (> 60 km depths). I also show that a simulation that exhibits non-linear viscous relaxation, in the continental lower crust, considerably reduces the deep afterslip, which is in agreement with observations of relatively less aftershock moment release and the apparent lack of interseismic locking, at greater depths along the plate interface. Conversely, the linear rheology case results in large afterslip at greater depths. Similar to the 2D model outcomes, I favour a 3D simulation in which non-linear viscous relaxation mostly occurs in the continental lower crust and, to a lesser extent, in the continental upper mantle beneath the volcanic arc, since its better fit to the GPS data and, distinctly better correlation between afterslip and aftershock moment release. Therefore, my results challenge the common belief that the continental crust responds only elastically after megathrust earthquakes.
Finally, I calculate for the first time the transfer of stresses to the Illapel segment due to afterslip and non-linear viscous relaxation associated with the Maule event under the Coulomb Failure (CFS) theory. I show that the patterns of predicted horizontal surface displacement are opposite when using a model with linear and power-law rheology. Predictions from the power-law rheology case agree better with the GPS data. My results reveal that most of the CFS changes are due to the coseismic deformation. I also find that a direct triggering of the Illapel earthquake due to the Maule event is unlikely, since the small, albeit positive (∼ 0.05 bar) CFS values calculated at the Illapel hypocenter. Conversely, seismicity Mw ≥ 5.0 between these two events, in the southern region of the Illapel segment, occurs in areas of CFS > 0.2 bar, suggesting a mechanical triggering.
I conclude that geomechanical-numerical models incorporating temperature-dependent power-rheology with dislocation creep processes in the crust and upper mantle can be used to examine the underlying processes controlling the postseismic deformation. Furthermore, these models can produce deformation patterns that are more consistent with the physical concepts of strength distribution with depth and aftershock activity than earlier models that impose linear and non-linear rheologies without consideration of temperature.
