Validation of Teleseismic Inversion of the 2004 Mw 6.3 Les Saintes, Lesser Antilles, Earthquake by 3D Finite-Difference Forward Modeling

Abstract

International audienceWe study the 21 November 2004 Mw 6.3 Les Saintes earthquake that occurred south of the island of Guadeloupe at a shallow depth, damaging some buildings on the island. The objective of this work is to assess the potential of a teleseismic source study to reproduce local ground motion. The velocity model of this area is not currently well known, and the near-field seismograms are affected by paths and site effects. We first analyze this earthquake as a point source and then as an extended fault using teleseismic broadband waveform inversion approach from the P and SH waveforms at stations well distributed in azimuth. We obtain two stable and reliable solutions of the spatiotemporal history of the earthquake rupture on each nodal plane. The models show a rupture duration of 6 sec and a maximum slip of about 1 m but have different locations of the main asperity. We then carry out the forward modeling of the seismic wave propagation at the regional scale, coupled with the obtained kinematic source models taking into account the topography, the bathymetry, and the sea layer. We investigate if one of the fault models could produce synthetics that match the local observed ground motions. The simulation results provide a better understanding of the uncertainty of source, path, and site effects even if our knowledge of the underground geological structure is limited to the 1D stratified model. It is inferred that the 1D model does not seem to be appropriate in the northern direction (middle of Guadeloupe Island) due to the unknown basin structure, while it is relatively sufficient in the eastern and the northwestern directions from the source. By careful comparison of the synthetic and observed displacement seismograms in terms of main phase forms and arrival times at low frequency, especially at the closest station GBGA, the fault model with northeast faulting is estimated to be the more relevant to the measured local ground motions

The Mw = 6.3, November 21, 2004, Les Saintes earthquake (Guadeloupe): Tectonic setting, slip model and static stress changes

“…The previous model determined by Salichon et al [2009, Figure 4] has almost the same geometry (strike/dip/rake = 327/55/−89.8) and displays also slip zones on both sides of the hypocenter, with little slip at rupture initiation. However, their south‐eastern rupture patch is much reduced in comparison to ours, and we found that slip propagated from this zone toward the surface, a characteristic absent from their model.…”

“…We present here a new source model for the main shock determined from the joint inversion of teleseismic and strong motion records (data from the French national strong motion permanent network (RAP) available at http://www-rap.obs.ujf-grenoble.fr/, see details in Appendix A). A previous model was proposed by Salichon et al [2009] from the inversion of the teleseismic data only. Here we add the available strong motion records at distances less than 50 km in a joint inversion to increase the resolution on the rupture slip model and determine the optimum focal mechanism.…”

Quantification of Gravitational Mass Wasting and Controls on Submarine Scarp Morphology Along the Roseau Fault, Lesser Antilles

“…Left-lateral transtensional deformation along the inner arc is partially accommodated by a right-stepping series of trench-parallel, en-échelon, oblique faults that generally trend between 130 and 160 degrees (Figure 1b; Feuillet et al, 2010Feuillet et al, , 2011. One such fault is the Roseau normal fault, which was the source of the damaging 2004 M w 6.3 Les Saintes earthquake and associated tsunami (Figure 1b; Bazin et al, 2010;Escartín et al, 2016;Feuillet et al, 2011;Le Friant et al, 2008;Salichon et al, 2009). The ∼40 km long Roseau fault system forms the western boundary of the Les Saintes graben (Figure 2a).…”

Simulation of the 2004 tsunami of Les Saintes in Guadeloupe (Lesser Antilles) using new source constraints

“…Its hypocentral location is compatible with the mean dip of 50 • of the Roseau fault, the only candidate with a plane model in agreement with the focal mechanism. More complex source models have been determined from the inversion of teleseismic data (Salichon et al 2009) and the joint inversion of teleseismic data and strong motion records (Feuillet et al 2011a). The latter one consists of a single fault segment 31.5 km long and 19.5 km wide, subdivided into 273 sub-faults measuring 1.5 km along strike and dip (Table 2 and Fig.…”