Research

My research interests lie in using seismic source imaging techniques such as tomography to infer tectonic structure in the crust and the dynamics of fault zones. I also investigate the effects of complex fault systems on the rupture path during an earthquake using dynamic rupture simulations and generate broadband ground motion simulations for seismic hazard assessment.

The Mw7.0 2010 Haiti earthquake caused more than 200,000 deaths and left more than one million people homeless. The earthquake was not unexpected, given geodetic measurements showing strain build up on the Enriquillo Fault, the major fault in southern Haiti. We used a compilation of data from temporary seismic stations deployed after the earthquake to obtain high-resolution aftershock relocations. Cross sections through the aftershock distribution show several dipping fault planes. The western cluster delineate a south-dipping structure which correlate to a previously mapped fault called the Trois Baies thrust fault. The central and eastern clusters of aftershocks delineate a main north-dipping rupture plane with slightly different dip. These two clusters correlate with the locations of slip patches from previous source inversion study. To learn more, see Douilly et al. (2013).

We performed a joint inversion for a 3-D P and S velocity structure and hypocenter location by using arrival times from 595 aftershocks following the 2010 Haiti earthquake. We considered a 10 × 10 km grid spacing and used the damped least squares computer algorithm SIMULPS14 to simultaneously invert for 3-D P velocity, Vp/Vs ratio, and hypocenter location. The model contains strong lateral variation onshore and offshore of the peninsula near the Trois Baies fault (TBF). The offshore low velocity anomaly resides primarily at 5 km depth and correlates with the location of thick sediments in the submarine basin. At 10 km depth, we found a sharp N125° low velocity zone running along the Petit-Goâve-Jacmel fault (PGJF) that extends offshore to the Trois Baies fault. The Petit-Goâve-Jacmel fault, which was a suspected fault based on geological observations, seem to be a major fault according to the tomography result. To learn more, see Douilly et al. (2016)

We use a finite element model to simulate propagation of rupture on the Léogâne fault, varying friction and background stress to determine the parameter set that best explains the observed earthquake sequence, in particular, the ground displacement. The dynamic rupture simulations of the 2010, Mw 7.0, Haiti earthquake show that planar fault geometry derived from the precise aftershock relocation study of Douilly et al. (2013), together with the appropriate regional stress and friction model, can successfully replicate a rupture propagating from east to west and generate a finite ground displacement consistent with geodetic observations. As inferred from finite fault inversions, rupture does not propagate to the neighboring Enriquillo or Trois Baies faults because, given their geometry and orientation (strike and dip) with respect to the rupture propagation on the Léogâne fault, shear stress on those two faults did not reach failure. To learn more, see Douilly et al. (2015).

In this study of fault interactions, we use dynamic rupture simulations to investigate the effects of complex fault geometry and background stress field on rupture of the San Andreas fault east of San Gorgonio Pass, with a focus on determining the conditions that could allow the rupture to propagate on the Mission Creek, Banning, or Garnet Hill strands. The top figure is the finite element mesh and the bottom figure shows the slip on the faults. The results reveal that fault geometry has a significant impact on throughgoing rupture along this branched fault system in Southern California. To learn more, see Douilly et al. (2020)

We generated broadband ground motions in the vicinity of a hypothetical Mw 7.1 earthquake consistent with a 52-km-long rupture on the plate boundary fault segment adjacent to, and to the east of, the 2010 Léogâne Fault rupture in Haiti. Field observations have shown evidence of recent earthquake ruptures and geodetic measurements have attested that this fault segment is currently accumulating elastic strain likely to be released in future earthquakes. We used dynamic rupture simulations to generate the low-frequency ground motion and the specific barrier model (in the context of the stochastic modeling approach) to generate the high-frequency ground motion. The two independently derived ground motion components were then combined using matched filtering at a crossover frequency of 1 Hz to generate broadband ground motion simulations. Results showed that the mean peak ground acceleration in Port-au-Prince is ~0.45g, which is about twice as much as the estimated peak ground acceleration during the 2010 Haiti earthquake. To learn more, see Douilly et al. (2017)

Ongoing project of multi-cycle earthquake simulations in the Caribbean using RSQSim. The video above is showing slip distribution for M > 7 earthquakes