Broad-band strong motion simulations coupling k-square kinematic source models with empirical Green's functions: the 2009 L'Aquila earthquake
DEL GAUDIO ; CAUSSE ; FESTA
Type de document
ARTICLE A COMITE DE LECTURE REPERTORIE DANS BDI (ACL)
Langue
anglais
Auteur
DEL GAUDIO ; CAUSSE ; FESTA
Résumé / Abstract
The use of simulated accelerograms may improve the evaluation of the seismic hazard when an accurate modelling of both source and propagation is performed. In this paper, we performed broad-band simulations of the 2009, M 6.3 L'Aquila earthquake, coupling a k2 kinematic model for the seismic source with empirical Green's functions (EGFs) as propagators. We extracted 10 EGFs candidates from a database of aftershocks satisfying quality criteria based on signal-to-noise ratio, fault proximity, small magnitude, similar focal mechanism and stress drop. For comparison with real observations, we also derived a low-frequency kinematic model, based on inversion of ground displacement as integrated from strong motion data. Kinematic properties of the inverted model (rupture velocity, position of the rupture nucleation, low-frequency slip and roughness degree of slip heterogeneity) were used as constraints in the k2 model, to test the use of a single specific EGF against the use of the whole set of EGFs. Comparison to real observations based on spectral and peak ground acceleration shows that the use of all available EGFs improves the fit of simulations to real data. Moreover the epistemic variability related to the selection of a specific EGF is significantly larger (two to three times) than recent observations of between event variability, that is the variability associated with the randomness of the rupture process. We finally performed 'blind' simulations releasing all the information on source kinematics and only considering the fault geometry and the magnitude of the target event as known features. We computed peak ground acceleration, acceleration Fourier and response spectra. Simulations follow the same trend with distance as real observations. In most cases these latter fall within one sigma from predictions. Predictions with source parameters constrained at low frequency do not perform better than 'blind' simulations, showing that extrapolation of the low-frequency description of the rupture front as inferred by the kinematic inversion may introduce some bias in the final simulations.
Source
Geophysical Journal International, num. 1, pp.720 - 736 p.
Editeur
Oxford University Press
