Broadband ground motion simulation in the vicinity of a fault
Supervisor : Mathieu Causse, ISTerre (mathieu.causse univ-grenoble-alpes.fr)
Skills : quantitative seismology and/or engineering seismology, signal processing, programming in Matlab
Key words : seismic hazard, strong-motion data, seismic rupture, Green’s function, uncertainty, variability
The prediction of synthetic ground-motion time histories due to a hypothetical future earthquake is a fundamental stage to anticipate damage. Ground motion depends on the seismic energy released on the ruptured fault (rupture process) and on the geological structures crossed by the waves (Green’s functions). The prediction of ground-motion in the vicinity of a fault, where damage is expected to be the highest, is particularly challenging because : (1) ground motion strongly depends on the details of the rupture process, which is a priori unknown and (2) the wave propagation can strongly vary between the different parts of the fault and the target site.
The candidate will contribute to the development of a recently implemented simulation code combining a generator of heterogeneous rupture models and empirical Green’s functions (Dujardin et al. 2018). This means that the wave propagation in the geological structures is not modeled using numerical techniques but based on small earthquake recordings, naturally containing this information. In low to moderate seismicity areas, the number of available small earthquakes is however generally too small to properly sample the fault and capture the spatial variability of the wave. Some corrections are then required (distance, radiation pattern, etc), the effects of which need to be carefully quantified before an application to ground motion prediction of a potential future event. In addition, the lack of a priori knowledge about the rupture process of a potential rupture event (rupture velocity, stress drop, position of the rupture nucleation, etc) implies the generation of a large population of physically realistic rupture scenarios. The resulting ground motion variability needs to match the recent observations based on analyses of large ground motion databases. This internship proposes two applications :
- The M6.5 2016 Norcia strong motion dataset. The candidate will test the ability of the simulation code to predict a posteriori the strong motion data recorded during the 2016 Norcia event (Central Italy). The very large number of small aftershock recordings makes it an ideal case to analyze the impact of the small earthquake dataset (number of events, magnitude, processing, etc) on the simulations.
- Excepted ground motion at CEA-Cadarache for a M6 event ? The candidate will next simulate ground motion for a potential event on the Middle Durance fault system at the CEA-Cadarache site (Sinaps@ project extension, 2018-2019). The empirical Green’s function simulations will be combined with Green’s functions computed based on seismic noise and/or 3D simulations techniques, so as to enhance the low-frequency predictions (< 1 Hz).