Soil, basin and soil–building–soil interaction effects on motions of Mexico City during seven earthquakes

Abstract

Starting with the destructive 1985 Michoacan Mw8·0 earthquake, the lake zone of Mexico City has been experiencing ground motions bearing the effects of strong soil amplification at specific site-dependent periods. Last year's Mw = 7·1 Puebla earthquake, although less damaging, with different orientation and mechanism, and at a much shorter distance than the 1985 event, nevertheless produced records bearing similar soil amplification effects and even rather similar motions on rock. This paper concentrates on four sites, distinguished mainly by the thickness of the soft clay, on which three to seven seismic events (with Mw > 6·5) have been recorded. Using the corresponding records on two hilly zone sites as base rock motions, amplification functions in terms of ratios of top-over-base recorded acceleration response spectra are studied. Although the dominant periods of each amplification function confirm the occurrence of resonance at the fundamental natural period of each particular soil stratum, differences between the amplification functions of the east–west and north–south components are evident. Using the actual soil profile of two of these stations, along with the G:γ and ξ:γ curves from the general literature and from site-specific measurements, one-dimensional wave propagation analyses are conducted. Reasonable agreement with the records is generally found in terms of response spectra. However, weak motions recorded in some events (where soil behaved essentially linearly) exhibit a number of beating cycles which last for almost one additional minute beyond 60 s, and which are barely noticeable in the computed motions. An oversimplified (in geometry and properties) two-dimensional basin, 6 km long, containing 60 m of Mexico City clay, is analysed with finite elements, excited by the rock outcrop hilly zone records. Wave propagation in this basin reveals the generation of Rayleigh waves at the edge, which move towards the centre and interfere with the incident and reflected S and P waves. The resulting motions away from the edge resemble the recorded motions at the CAO site in a 1999 earthquake better than the one-dimensional analysis, but are still far from satisfactorily reproducing the beating. An additional factor to qualitatively explain reality is generically examined: the oscillations of a tall (15-storey) building and the feedback of its free oscillations into the ground (soil–building–soil interaction) are found to increase free-field accelerations slightly.