CRUSTAL WAVE-PROPAGATION ANOMALY ACROSS THE PYRENEAN RANGE - COMPARISON BETWEEN OBSERVATIONS AND NUMERICAL SIMULATIONS

Lg records analysis and numerical modelling of Lg propagation are used to find out to what extent this phase can be seen as a marker of unidentified structural anomalies in the crust. This study is based on Lg propagation through the Pyrenean range from earthquakes located in Spain. We have first evaluated the mean value of the S-wave quality factor for central Spain. We have computed simultaneously the seismic station responses and the source functions. The correction for propagation effects, assuming a homogeneous attenuation and the theoretical calculation of the Lg excitation, lead to the seismic moment of each event. The moment magnitude obtained correlates well with the magnitude proposed by the local networks. This gives a confirmation of the Q model in the low-frequency range (1-5 Hz). As we intended to compare traces of different Spanish earthquakes recorded in France at different epicentral distances, we had to make amplitudes independent of propagation and source effects. Therefore, we corrected the spectral amplitudes for geometrical spreading, anelastic attenuation and normalized them to equal seismic moment. We then plotted the records as a function of group velocity, in order to make up a fan profile along the Pyrenean axis. The resulting section reveals that in the central and the eastern parts of the range, neither the North Pyrenean Fault, nor the Moho jump deduced from seismic-refraction experiments and vertical seismics, seem to affect Lg propagation. However, there is an extinction of the Lg phase in the western part of the chain. The lateral extent of this area is correlated with a zone of positive gravity anomaly, probably linked to the presence of dense material of mantle origin. A numerical simulation in the low-frequency band indicates that the Moho topography inferred from deep seismic soundings does not explain the strength of the observed attenuation. Ray-tracing seismograms show that, at high frequency, the conclusion is the same. The attenuation effect due to lateral variation of structure should not be so strong. We, therefore, think that attenuation of guided waves is not due to large-scale geometry effects, but is due to local properties of the crustal materials, possibly apparent attenuation due to scattering on small-scale heterogeneities.