Stability of fault plane solutions for the major N-Italy seismic events in 2012

Highlights

• We compare fault plane solutions retrieved inverting at different cutoff periods.

• Short cutoff period inversion of moment tensor for shallow events is preferred.

• The retrieved solutions agree with independent geodynamic models of the area.

Abstract

We propose a critical analysis of the moment tensor solutions of the major seismic events that affected northern Italy in 2012. Inverting full waveforms at regional distance using the non-linear method named INPAR, we investigate period dependent resolution that affects in particular the solutions of shallow events. This is mainly due to the poor resolution of M_zx and M_zy components of the seismic tensor when inverting signals whose wavelengths significantly exceed the source depth. As a consequence, instability affects both source depth and fault plane solution retrieval, and spurious large Compensated Linear Vector Dipole components arise. The inversion performed at cutoff periods shorter than 20 s reveals in many cases different details of the rupture process, which are supported by independent geodynamical arguments. Thus we conclude that inversion of full waveforms at cutoff period as short as possible should be preferred.

Introduction

In the critical analysis of the moment tensor solutions of the recent seismic sequence in Emilia–Romagna, retrieved by the non-linear method named INPAR (Šílený et al., 1992), the reliability of each solution is discussed in terms of accuracy of source time function (STF), percentage of Compensated Linear Vector Dipole (CLVD) component and statistical significance of the retrieved mechanism. The theoretical and practical limits of linear moment tensor retrieval from both body waves (Dufumier, 1996) and surface wave spectra (Dufumier and Cara, 1995) have been widely discussed, in particular for very shallow earthquakes, suggesting to use full waveforms at regional distance for moment tensor inversion. Reduction to zero of the isotropic component does not ensure stable solutions, since it does not affect off-diagonal elements M_zx and M_zy. The poor resolution of these elements, which excite Green functions vanishing at the free surface, causes instability in moment tensor solutions for crustal earthquakes whose source depth is significantly exceeded by the dominant wavelengths of data used in the inversion. As a consequence spurious large CLVD components arise and pairs of solutions with nodal planes showing a 180° rotation around the vertical axis (i.e. ϕ = ϕ + 180°, where ϕ is strike angle) are possible, as shown analytically by Henry et al. (2002). The less well determined the M_zx and M_zy components are, the more likely is the existence of a pair of well-fitting solutions. The conditions for the existence of such a pair are more likely satisfied when dealing with earthquakes close to vertical strike–slip or near-dip–slip mechanism, whose nodal plane dips ~ 45°. Similar conclusions are reached by Bukchin (2006) and Bukchin et al. (2010), who prove that the focal mechanism of a seismic source can be uniquely determined from records of surface waves with lengths significantly exceeding the source depth only if the dip angle of one of its nodal planes is sufficiently small.

The real meaning of the retrieved CLVD component is still debated, since it can be even an artifact of the inversion arising from both sparse distribution of stations and inaccurate Earth models (Henry et al., 2002, Panza and Saraò, 2000). The occurrence of sub events with different pure double-couple mechanisms very close in time may lead to the retrieval of a relatively large CLVD component (Guidarelli and Panza, 2007). In this case, different intervals of the source time function can be separately investigated in order to assess how the rupture mechanism changes in time.