Slip partitioning during the initiation phase preceding the 2014 Iquique, Chile, earthquake.

Abstract:

How earthquakes start is a long-standing question but yet still outstanding. Two models are usually proposed: the cascading model where one earthquake randomly triggers other events and so on until the mainshock occurs, and the slow-slip model where a phase of progressively accelerating aseismic slip precedes the mainshock. Thus, when a nucleation phase is observed, it is key to evaluate how much of the slip on the fault is seismic versus how much is aseismic in order to discriminate between these two models. In this study, we attempt to quantify the evolution of seismic vs. aseismic slip partitioning along with the associated uncertainty during the preparatory phase of the 2014 Iquique earthquake in Chile. To do so, we analyse time series of surface displacements recorded by Global Navigation Satellite System (GNSS) receivers in the Iquique region. First, we obtain the spatiotemporal evolution of slip prior the mainshock using a Bayesian algorithm we have developed describing the temporal evolution of slip on the fault using a set of overlapping I-spline functions. We find that ~15 days prior the mainshock a slip patch develops above the level of uncertainty. Over the same time period, ~40 earthquakes with magnitude > 4.0 are recorded. To assess how much slip is contaminated by these foreshocks, we attempt to remove their contribution on the surface displacements time series. To do so, we calculate for each foreshock the expected contribution at the surface and subtract it from the time series while accounting for uncertainties on the foreshock location and focal mechanisms. The corrected time series are then used to invert once again for the spatiotemporal evolution of slip on the fault. By comparing the uncorrected slip evolution with the corrected one, we are able to quantify the partitioning between the seismic and aseismic slip during the nucleation phase of the Iquique earthquake.