Prediction of the ground motion in Lebanon by the empirical Green functions techniques and the ground motion prediction equations
Résumé
Historically, Lebanon experienced strong earthquakes produced by the Yammouneh fault that crosses its territory from the north to the south dividing it into two parts. An example is the 1202 A.D. earthquake, estimated to 7.6’s magnitude and felt from Armenia to Egypt. The present study intends to perform a site-specific estimation of the ground motion caused, within the heavily populated Beirut area, by such potential large earthquakes on the Yammouneh fault. A multi-step procedure was adopted in that aim: The first one is to simulate the effect of a M6.5 event at a Broad Band rock station using the empirical green function method EGF. After reviewing the list of recorded events along the fault during three years, a digital seismogram of a representative small event of Md 3.1 is chosen. The variability of the source, directivity and roughness is assessed. Furthermore, due to the lack of accuracy in the determination of many parameters of the empirical Green function, the uncertainties of the corner frequency, the epicentre distance, the magnitude and the depth are also evaluated. Finally the median and the standard deviations of the ground motion are estimated on the reference rock station BHL. The second step consists in evaluating the same ground motion M6.5 by a set of ground motion prediction equations GMPE. Five models representing different regions of the world with different databases are used. The predicted ground motions are then compared to the simulated event by EGF. The rock ground motion prediction equation that gives results in good agreement with the simulated earthquake by the EGF is considered as the model that represents the best the region of Lebanon. This estimated rock ground motion is then transferred to various other sites with different subsoils within the greater Beirut area, using empirical transfer functions derived from a temporary seismological survey with the standard spectral ratio technique. The response spectra, associated to a M6.5 earthquake, are calculated in all the sites and the amplification factors are deducted. The final step consists in estimating the effect of a M7.5 event that could occur on the ammouneh fault. The "best suited“ rock ground motion prediction equation as identified in the step 2, is used to predict the ground motion on the reference station BHL. By applying the amplification factors, the response spectra are also calculated on the sediment and the rock sites of Beirut. Finally, a comparison of these response spectra with the actual response spectra used in the Lebanese regulations is presented.