A study of earthquake locations under Somma-Vesuvius volcano using different 3D models

Anthony Lomax and Jean Virieux
UMR Geosciences Azur, 250 Rue A. Einstein, 06560 Valbonne, France

Raffaella Montelli
UMR Geosciences Azur, 250 Rue A. Einstein, 06560 Valbonne, France; Department of Geosciences, Guyot Hall, Princeton University

A. Zollo and P. Capuano
Dipartimento di Scienze Fisiche, Universita' di Napoli, Italy; Observatorio Vesuviano, Naples, Italy


The Somma-Vesuvius volcano and its surroundings are characterised by significant topography and strong 3D structural variations. This structural complexity must be taken into account when monitoring the volcano seismicity in order to get reliable estimates of the absolute epicenters, depths and fault mechanisms for events beneath the volcano. We have recently developed a reference, 3D, P-wave velocity model for Vesuvius by interpolation of 2D velocity sections obtained from non-linear tomographic inversion of the Tomoves 1994 and 1996 active seismic experiment data.

In this paper we first examine events from 1989-1998 under Vesuvius located using this reference 3D model and the NonLinLoc (http://www.alomax.net/nlloc) global-search, 3D location software The events with highest quality locations fall in a depth range of about 1 to 3.5 km below sea level, significantly shallower than has been determined in previous studies. The events are concentrated directly under the active cone, in the upper 2 km of the Mesozoic carbonate basement underlying the Somma-Vesuvius complex. We obtain the best quality locations using a relatively high Vp/Vs ratio of 1.90; this ratio reflects primarily physical properties of the volcanic edifice, such as abundant micro fracturing or the presence of fluids. The results of this study do not exclude a more typical Vp/Vs ratio for the carbonate basement and surrounding outcrops. The first-motion mechanisms give a weak indication of predominantly N-S to near-vertical directions for the tension axes, and ESE-WNW to near-vertical directions for the compression axes.

We next relocate the 1989-1998 events in new 3D models obtained with fully 3D tomographic inversion of the Tomoves data. We compare the locations obtained with the various 3D models and discuss the implications to absolute event location and depth uncertainties and to the reliability of features in the 3D models.