GPS constrained finite element models of inter-rifting deformation: A case study from the Main Ethiopian Rift

The initiation and evolution of continental rifting involves both faulting and magmatism. However, the relative importance of these two processes is still under debate. Currently, two end-member conceptual models are proposed, one where rifting is driven by tectonic forces only (i.e. tractions at plate boundaries and buoyancy forces), the other where magmatic intrusions play a significant role. These two conceptual models predict different patterns of strain rate distribution across and along rifts, with a more focused pattern of deformation in the magma-assisted models versus a broad region in the tectonic stretching model. Due to this difference, GPS observations at active continental rift segments should allow us to distinguish the underlying processes. Here we present results of a study designed to develop a better understanding of continental rifting by using inter-rfting (between rifting episodes) GPS observations as constraints on 2-D finite element models of rifting cycles to determine the role of magmatic versus tectonic processes and constrain parameters associated with rifting architecture. GPS data from multiple rift zones (e.g. the Main Ethiopian Rift (MER) and the Western and Eastern volcanic zones in Iceland) reveal strain accommodation to be strongly focused within a narrow, ∼50-km-wide shear zone in each rift zone. Previous models have found that the narrowness of the shear zone can be explained either by a buried, inflating dike within a purely elastic half space or a model of regional extension incorporating a viscoelastic layer underlying an elastic layer only 3 km thick. Neither of these models are geologically viable because the former does not incorporate the influence of viscoelastic flow before and after rifting episodes and continuous diking at depth seems unlikely, and the latter contradicts seismicity observed to depths of 15 km in the MER. Here we use two-dimensional Abaqus finite element models to test models incorporating more realistic parameters including viscoelastic rheology, recurrence intervals for rift cycles, the timing of the last rifting event, and the amount of dike opening and fault slip. Each model is tested against observed present day strain rates at the MER. Our modeling results reveal that slip on MER border normal faults is not a likely explanation of the observed narrow shear zone. Whereas, models that are driven by 1) dike intrusions and 2) plate spreading consistent with kinematic plate motion models are able to explain the inter-diking GPS observations. The models consider rheologic partitioning (divided into 3 vertical layers) with the dike cutting through the lithosphere to a depth of 8km at a 200 year recurrence interval at a rate equal to that of plate spreading. This model of diking episodes provides a more plausible explanation for the rifting process at the MER than previous studies.