Using ambient noise measurements in the process of assessing earthquake hazards in Urban areas: Examples from Israel

Ground motion amplifications due to soft soils, common in urban areas, are a major contributor to increasing damage and number of casualties. The great variability in the subsurface conditions across a town/city and the relatively high costs associated with obtaining the appropriate information about the subsurface, strongly limit proper hazard assessments. Direct information from strong motion recordings in urban areas is usually unavailable. Such is the situation in Israel which is small and its population centers are in close proximity to the seismically active Dead Sea Fault system, capable of generating earthquakes with magnitude as high as 7.5. Heavily limited with relevant local recordings of strong ground motions, we adhere to the use of simplified modelling of the earthquake processes. More precisely, we generate synthetic spectra of the expected ground notions by implementing the so called Stochastic Approach (e.g. Boore, 2000), in which we integrate analytical models to determine the nonlinear response of the site under investigation that requires modelling of subsurface. Then, Monte-Carlo simulations are used to obtain the uniform hazard, site-specific acceleration spectrum. Over the years, we have conducted site investigations in several thousands of sites across Israel. These investigations demonstrate the usefulness of using horizontal-to-vertical (H/V) spectra of ambient noise measurements to identify sites with high potential for being vulnerable to amplification effects and characterize the sites with respect to their expected resonance frequencies and the corresponding H/V levels. This information, together with any available geological, geotechnical and geophysical information, helps building a reliable model of the subsurface, which is then integrated in the processes of the seismic hazard assessment. Modeling the subsurface and assessing the earthquake hazards in urban areas involves systematic ambient noise measurements on a grid with spacing of 500 m. At instances where high variations are observed, the spatial density of measurements is significantly increased. In doing so, we are able to develop a regional subsurface model, which is systematic with all additional information we compile, i.e., geological maps, borehole information, seismic refraction surveys etc. In order to reduce the scatter in H/V observations, the processing scheme involves continuous recording of ambient noise for about 1-2 hours and careful selection of time windows from which H/V functions are calculated. To assure stability in H/V observations, measurements are repeated in different times and dates. At several occasions, while measuring the ambient noise or when we deliberately aimed at recording seismic events, it is possible to support ambient noise spectral ratios with spectral H/V observations from earthquakes (often considered transfer functions) and explosions recorded by accelerometers and seismometers. In all cases, we obtained similar H/V spectral ratios from the different data sets. At the final stage of the hazard assessment process, we divide the study region into zones of similar hazard characteristics, which are used for earthquake scenarios and better represent the design acceleration spectra for safer buildings.