Seismic Velocity Model From Site Response Measurements and Influence of Lithological Variations Down to 500-m-depth on Ground Motion Amplification (Israel's Inner Coastal Plain)

The last strong destructive earthquake in Israel occurred in 1927 when 300 people were killed out of a population of 200-300 thousand. Today the population is about 8 million and the bothering question is how many losses are expected, if such an earthquake will struck again. The present research focuses on the inner coastal plain of Israel, where much of its population is concentrated. The distance to the Dead Sea seismogenic zone is ~70 km and in order to estimate the ground motion expected during an EQ, amplification related to near surface sediments must be considered. One practical obstacle in estimating site effect in Israel is the lack of strong motion recordings. Another problem, typical to the area studied here, is the lack of hard rock sites that can be used as reference stations. Thus, the present study relies on the Nakamura technique, which determines the dominant frequency and the amplification factor of a single site from the ratio between horizontal and vertical components of ambient vibrations. Three advantages were most helpful here. One is the large amount of subsurface data available from hundreds of boreholes and seismic lines. The second is the simple measurement technique that allowed hundreds of results within a few years - the present study analyzes 700 provided with the courtesy of the Geophysical Survey of Israel. The third advantage is the consistent relationships found between the thickness of soft sediments overlying the hard Judea Group and the measurements. These relationships (valid to thickness smaller than 500 m) allow quick approximation of the resonance frequency and the amplification factor directly from the top Judea structural map in areas that were not surveyed. The advantage of this approximation is the availability of such a map in high resolution in most of the country; the disadvantage is the low accuracy of plus minus 50 percent. In order to improve this approximation the soft sediments overlying the Judea Group were subdivided to 4 units with known thicknesses. A power law function describing the increase in S-wave velocity with depth was found for each unit in a way that best fits calculated frequencies and amplification factors to measurements. After calibration of the velocity model to 700 measurements, amplification was recalculated and better maps were prepared for areas still not surveyed. The maximal error now is 35 percent for the dominant frequency and 45 percent for the amplification factor, though for most sites the error is much smaller. Except for the specific importance of these maps, it is emphasized here that lateral variations of 0.3 to 10 Hz in the resonance frequency and 2-10 in the amplification factor are caused by lithological variations at depths of 0-500 m. Surface geology, which is quite monotonic in this case, clearly cannot explain the results alone.