Ground motion simulations of a major historical earthquake (1660) in the French Pyrenees using recent moderate size earthquakes

International audienceIn regions where only small- to moderate-size events have been recorded, it is important to be able to anticipate the effects of a large event by simulating the ground motion it may generate. Using the very good records of two small earthquakes that occurred in the central French Pyrenees (2007 November 15, Mw= 3.6; and 2006 November 17, Mw= 4.5), we simulated the ground motions generated by a magnitude 6.1 earthquake, equivalent to a historical event that struck the region in 1660. This major earthquake caused severe damage and reached a maximum macroseismic intensity of IX on the MSK scale. The simulation is based on the empirical Green's function (EGF) method, which allows simulations in a broad frequency range, and accounts for both propagation and site effects. We first validated the method by reproducing the records of the Mw= 4.5 earthquake using the Mw= 3.6 earthquake as an EGF. A careful analysis of corner frequencies and spectral ratios revealed a clear directivity effect of the rupture process for the larger event (Mw= 4.5). When this directivity is taken into account in the simulation, a very good reproduction of the Mw= 4.5 ground motion parameters is obtained at all the stations. We then used the records of the Mw= 3.6 earthquake as an EGF for simulating the historical event. At 11 stations we computed a large number of synthetic accelerograms that aim to account for the possible source variability of an Mw= 6.1 earthquake. The comparison between the computed ground accelerations and those given by three different empirical ground motion prediction equations (GMPEs) reveals that simulation is quite successful. Finally, our simulation results are compared with intensity data of the 1660 event, by use of three different Ground Motion Intensity Conversion Equations (GMICEs). We found that the intensity levels predicted from ground motion simulations are systematically lower than the reported macroseismic intensities. To explain this difference, the hypothesis of a possible underestimation of the magnitude of the 1660 event is discussed