OBSERVED VERSUS PSHA-ESTIMATED EXCEEDANCE RATE OF GROUND MOTION IN MEXICO
Earthquake hazard at a site is expressed in terms of exceedance rate, ν, of some measure of ground motion, for example peak ground acceleration (PGA). Thus, ν(a) is the number of times the ground motion is expected to exceed PGA of a at the site per unit of time. If the site of interest has been instrumented for sufficiently long time, then the estimation of exceedance rate reduces to number counting. This may be called the observed exceedance rate. This, however, is rarely the case since few sites of interest are instrumented for sufficiently long time. In these cases, recourse is taken to the probabilistic seismic hazard analysis (PSHA). The method requires an earthquake catalog. Based on seismotectonic considerations, the seismicity of the region around the site is divided to different areas and volumes. Appropriate attenuation relations (preferably site specific, although this is rarely feasible) are used in the calculations. The analysis is based on several assumptions such as the validity of seismicity models and the ergodic assumption of ground motion prediction equations.
Obviously, it is of critical importance to know how well observed and PSHA estimations of exceedance rates agree. Beginning 1985, accelerographic network of Mexico has undergone rapid expansion. At many sites, the recording has been continuous with minor interruptions although the quality of accelerographs has improved. On the other hand, Servicio Sismológico Nacional (SSN) began installing and operating broadband seismic stations in 1992. These stations are equipped with both velocity and acceleration sensors. Even though the time of operation of these networks is somewhat short, it seems worthwhile to compare observed and PSHA estimations of exceedance rates.
We have constructed observed exceedance rates at many of the stations of the accelerographic and broadband networks which have been operating for more than 20 years (~ 50 stations). Our preliminary analysis shows large difference in the observed and PSHA estimations at many stations. We discuss whether the difference is consistent. Is the difference attributable to inadequate attenuation relations used? Does the site effect, which is not included in PSHA calculations, the primary cause? How about seismic model used in the PHSA-computed exceedance rate?