Are the earthquake locations reliable in caldera structures?

Abstract

Increasing evidence suggests that non-planar faults have been activated during caldera collapse events linked to basaltic eruptions around the world. Despite advances in the description of the source location and processes, there is inconclusive evidence regarding the quality of the locations, both hypocentres and centroids, and their relationship with the geology and laboratory experiments. For example, trapdoor and piston collapses are not expected to activate faults within the caldera boundaries, unlike piecemeal or funnel collapses. Small earthquakes at ring faults present small rupture areas, hence, the curvature does not affect the location of the events and the source can be approximated as a planar fault. For larger events, the curvature becomes critical and the centroid migrates towards the centre of the ring while the arc rupture increases. Consequently, centroid locations within the caldera rims might be either real events in the corresponding location or an artefact due to the curvature of the source. In this work, I test the well-known equation of the centroid location at a ring fault with uniform slip and contrast it with a Gaussian-like distribution. Then, I perform numerical simulations assuming three cases of rupture velocity to compare hypocentre estimations with the analytical expressions for centroid location. For instance, hypocentres of an instantaneous rupture behave as the centroid of a uniform slip distribution, a 6 km/s rupture shows hypocentres congruent with the centroid of a Gaussian-like slip distribution, and finally, a 0.9vS rupture velocity shows the most likely locations at the rims but showing large uncertainty. I applied these results to two cases; hypocentres calculated at Bárðarbunga and centroids calculated for Sierra Negra caldera collapse earthquakes. By using this methodology, both cases give a first order approximation on the quantification of the artefact affecting locations due to curvature.