3D Tomography based on the aftershock sequence of the 2016 Mw 7.8 Pedernales, Ecuador earthquake

Abstract

Based on seismicity recorded by the permanent Ecuadorian seismic network and our large emergency network installed shortly after the 2016 Mw 7.8 Pedernales earthquake, I derive a seismic 3D velocity model for central coastal Ecuador based on local earthquake tomography (LET). I manually analyzed seismic waveforms recorded on our combined amphibious network to determine high quality arrival times of seismic phases. I inverted the seismic travel times simultaneously for earthquake locations and three-dimensional (3D) velocity structure ($Vp$ and $Vp/Vs$) using a staggered approach that increases complexity from 1D to finally obtaining a detailed 3D model. While our 1D velocity model highlights the first-order structures of the studied margin through the analysis of the relocated seismicity, station correction terms and regional moment tensors (RMT), the resulting three-dimensional tomographic images show an area that is widely affected by the subducting topography and small- to large-scale faults at the surface. The oceanic Nazca plate is well imaged down to $\sim$40 km depth by an eastward dipping high $Vp$ velocity feature. I also identify a low $Vp$ ($\sim$5.5 km/s) region along strike in the marine forearc, which I interpret as a thermal anomaly that might be caused either the rocks coming from the Galapagos Hot Spot or by shallow serpentinization. The marine forearc region also shows differences to the North and to the South of the Equator line, with a prominent seamount, imaged by a $Vp\sim$5.0 km/s, coming from the Atacames seamount chain in the northern region and, based on the elevated $Vp/Vs$ ratios (>1.84), a deeply fractured oceanic crust in the South caused by the subduction of the Carnegie Ridge. The elevated $Vp/Vs$ ratios (>1.84) also suggests a large presence of fluids. In the oceanic crust, this is associated to a combination of extensional faults formed prior to subducting and highs in topography that contribute to fracturing within the downgoing plate. In contrast, elevated $Vp/Vs$ ratios (>1.84) in the upper continental crust are expressions of a highly fractured marine forearc and the presence of small- to large-scale faults that help fluids to circulate along the overriding plate. The relocated seismicity shows several clusters, mostly organized along the plate interface, but also occurring in both, oceanic and continental plates. Clustered aftershocks in the oceanic crust are located on the flanks of a subducting seamount which promotes active faulting. On the other hand, seismicity observed in the upper plate is related to the (re)activation of several crustal faults following the 2016 Pedernales earthquake. The imaged three-dimensional seismic velocity structures were associated with the seismotectonical and geological context of the Ecuadorian margin to give insights about the main features controlling the occurrence of large megathrust earthquakes in the region. I also explore the relation between the observed physical properties of the rocks along the slab interface and the spatial distribution of the coseismic slip of the Pedernales earthquake. Our findings suggest a strong correlation between domains with normal $Vp/Vs$ ratios (1.78-1.84) and the areas where the rupture propagated. In contrast, the areas with elevated $Vp/Vs$ ratios (>1.84) can be colocated with the up- and down-dip limits of the Pedernales earthquake and might have contributed to stop the rupture. Furthermore, a grid search analysis shows that the 2016 event occurred in the only area of the margin capable to host an earthquake with the observed magnitude. Finally, this study contributes to a better understanding of the processes occurring in subduction zones. Especially, I remark the importance of having a complete seismic velocity structure that includes both $Vp$ and $Vp/Vs$ ratios which complement with each other in order to give a full interpretation for the features observed along the study margin. Furthermore, the analysis of the seismic velocities of the rocks along the seismic interface, together with information derived from geodetic studies and the rupture area grid search approach designed in this work, can provide valuable data neccesary for the estimation of seismic hazard.