Effet de site sismique, méthodes d'évaluation: application au site de la ville de Tunis

Résumé.

Tunis n'est pas située dans une zone sismique très active; toutefois, une grande partie de la ville est construite sur un bassin quaternaire et la qualité géotechnique faible des terrains l'expose à des effets de site sismiques. Deux campagnes de mesures comparent les réponses dynamiques de plusieurs sites obtenues à partir de l'enregistrement de la séismicité et de mesures de bruit de fond sismique. Les résultats montrent que les fréquences de résonance peuvent être déterminées au moyen du bruit de fond. Les mesures réalisées en 300 sites de la ville mettent en évidence de très basses fréquences (0.4 à 1.5 Hz) dans des sites à remplissages sédimentaires. Les effets de site les plus marqués sont attendus dans les sites à remplissage peu consolidé. Une modélisation numérique attribue les modes propres de ces sites aux couches sous-consolidées superficielles, du moins pour les sites éloignés des reliefs.

Abstract.

The seismic site effect reflects the magnification versus frequency undergone by a seismic wave when crossing the subsurface layers. The methods most frequently used to obtain estimates are geotechnical structure modelling (assuming the information is available) or recording the actual site seismicity compared with the seismicity recorded on a reference rock. The latter method, also called the sediment to bedrock spectral ratio, provides the best experimental estimation as the required parameters are measured directly. However, it involves major practical difficulties, particularly the long duration of the measurements (and hence high cost) and the difficulty of undertaking seismic measurements in towns due to the ambient noise. The "H/V background noise" method (H/V, horizontal/vertical) consists of recording the site seismic background noise and computing the spectral ratio between the signals obtained on the vertical and two horizontal components. This ratio is known to give a good estimation of the fundamental soil resonance frequency.

The main part of Tunis city is built in a Quaternary basin. The presence of saturated silty and plastic clayey formations up to 50 m thick, overlying a more compact clayey–sandy layer, suggests the likelihood of seismic site effects. No seismic microzonation was available for Tunis city. To prepare such a zonation, two field surveys have been carried out to evaluate the dynamic site responses. Both of the methods discussed above were used. The experimental validity of the H/V background noise was checked and very low fundamental resonance frequencies were detected in the basin and on the higher ground. These appear to be caused by a deep reflector, probably the top of the limestone and marls of Eocene age. There are strong site amplitudes in areas with a silty infill. The amplification factor increases with the thickness of the deposits.

Non-linear responses were also modelled using the computer program EERA which uses the equivalent linear theory. Such modellings are based on the unit weight, shear wave velocity and thickness as well as the shear modulus of the soil layers. The results in the central basin, in zones some distance from the higher ground, show that the eigenmodes are mainly controlled by the upper unconsolidated formations present above the relatively sandy–clayey compact layer. However, for the zones near or extending into the higher ground where wave focusing phenomena would be expected, it is likely that the two-dimensional approach is necessary.