Finite-Fault Source Inversion Including Surface Topography Effects and 3D Velocity Structure: Application to the Norcia, M6.5, 30 October 2016, Central Italy Earthquake

Session: What Can We Infer About the Earthquake Source Through Analyses of Strong Ground Motion?
Type: Oral
Date: 4/29/2020
Time: 02:45 PM
Room: 115
Description: 

The reliability of the earthquake source parameters retrieved from kinematic finite fault inversion of strong motion data depends, among other factors, on the proper velocity structure adopted to generate the Green’s functions. The conventional approaches usually include Green’s functions for 1D layered seismic velocity models with flat surface topography. However, many regions in the world are characterized by significant topography and lateral heterogeneities in the crustal model. To improve earthquake source inversion and derive valuable constraints on earthquake rupture processes, we have modified the non-negative least-square kinematic inversion method of Dreger et al. (2005) by including the effects of topography and 3D velocity models on the adopted Green’s functions. The new Green’s functions have been generated with the numerical Spectral-Element Method (SEM) code SPECFEM3D for two structures: the 1D-crustal NNCIA velocity model with topography and 3D-crustal IMAGINE_IT velocity model with topography (Casarotti et al., 2016).

The computational cost of Green’s functions for such models is very expensive, for this reason we have decided to assume an already published fixed fault geometry allowing only for heterogeneous rake variation on the fault plane and inverting for rupture velocity, rise time and slip distribution. We have tested the new procedure on the 2016, Mw 6.5, Norcia earthquake for which none of the several studies in literature have adopted 3D layered seismic velocity models or have taken into account the significant topography of this area. To this goal, we have inverted the strong motion records of 34 three-component digital accelerometers belonging to the National Accelerometric Network of the Italian Department of Civil Protection and to the National Seismic Network of INGV. Finally, we have compared these new rupture models with Scognamiglio et al. (2018)’s model obtained with the same data and fault geometry but 1D velocity model and no topography.

Presenting Author: Laura Scognamiglio

Authors