SITE EFFECT AND SEISMIC HAZARD MICROZONATION ACROSS THE TOWN OF TIBERIAS

Transcription

1 SITE EFFECT AND SEISMIC HAZARD MICROZONATION ACROSS THE TOWN OF TIBERIAS February, 009 Report No 0/46/09 Principal Investigator: Dr. Y. Zaslavsky Collaborators: M. Gorstein, M. Kalmanovich, I. Dan, N. Perelman, D. Giller, G. Ataev, T. Aksinenko, V. Giller and A. Shvartsburg Prepared for Geological Survey of Israel

2 CONTENT LIST OF FIGURES... 3 LIST OF TABLES... ABSTRACT... 6 GEOLOGICAL AND TECTONIC CONTEXT... 8 Stratigraphy and lithology... 0 BRIEF REVIEW OF SEVERAL EXPERIMENTAL METHODS FOR SITE EFFECT ASSESSMENT... 3 MICROTREMOR RECORDING AND PROCESSING... Site response in Tiberias estimated by H/V spectral ratio from microtremor... 0 Comparison of H/V spectral ratios from microtremor and seismic events... 4 DISTRIBUTION OF THE RESONANCE FREQUENCY AND ITS ASSOCIATED H/V AMPLITUDE OVER THE STUDY AREA... 8 ESTIMATION OF SHEAR-WAVE VELOCITY MODELS AND RECONSTRUCTION OF SUBSURFACE STRUCTURE... 3 SEISMIC HAZARD MICROZONATION CONCLUSIONS... 7 ACKNOWLEDGEMENTS... 8 REFERENCES... 9

3 3 LIST OF FIGURES Figure. Tiberias - the New and Old together Figure. Geological map of the study area compiled from Schulman (966) and Sneh (008) with locations of the refraction profiles TB-, TB- and TB-3 (Ezersky, 008); R- and R- (Shtivelman, 99) and profiles and for constructing cross sections Figure 3. Location of the measurement sites in the study area. Numbers indicate the sites used as examples. TB-, TB- and TB-3 - refraction survey profiles (Ezersky, 008); R- and R- refraction survey profiles (Schtivelman, 99); TVR, TVR and POR accelerometer locations; Profile and Profile profiles for reconstructing subsurface structure Figure 4. Examples of seismometers locations during various sets of the site investigations in different geological conditions Figure. Examples of seismic station locations in Tiberias Figure 6. Examples of average Fourier spectra (top) and H/V spectral ratios (bottom), which yield a single peak. The black line indicates a vertical spectral component; the grey line indicates the average of NS and EW horizontal components of motion Figure 7. Examples of average Fourier spectra (top) and H/V spectral ratios (bottom), which yield two resonance peaks. The black line indicates a vertical spectral component; the grey line indicates the average of NS and EW horizontal components of motion.... Figure 8. Fourier spectra (top) and H/V spectral ratios (bottom) obtained at sites located on the exposure of the Cover Basalt. f0 indicates the fundamental frequency of the measurement site; f is an artificial frequency. H/V ratios at site T6 obtained by processing and reprocessing of the measurement data are shown by dashed and solid lines respectively.... Figure 9. H/V spectral ratio obtained at sites located on the outcropped Bira Fm. (T70 and T) and alluvium (T0 and T0) Figure 0. Comparison between the H/V spectral ratios obtained near the Tiberias Town Hall (T6 and TVR) and at site T3 at different times Figure. Accelerograms of the earthquakes recorded at site Tiberias Hotel (TVR): (a) earthquake occurred in the Dead Sea (004 0-, 08:, M L =., R=0 km) and (b) earthquake occurred in the Dead Sea fault ( , 4:3, M L =4.7, R=89 km);... Figure. Accelerograms of the earthquake occurred in the Dead Sea (004 0-, 08:, M L =., R=0 km) and recorded at site Tiberias Town Hall (TVR).... 6

4 4 Figure 3. Accelerograms of the earthquakes recorded at site Poriya Hospital: (a) earthquake that occurred in the Dead Sea basin (004 0-, 08:, M L =., R=0 km) and (b) earthquake that occurred in the Dead Sea fault ( , 4:3, M L =4.7, R=89 km) Figure 4. Comparison of different estimates of site amplification based on H/V spectral ratio techniques applied to earthquakes and microtremor recordings at sites TVR (Hotel) (a); TVR (Town Hall) (b) and Poriya Hospital (c) Figure. Distribution of the fundamental resonance frequency over Tiberias Figure 6. Distribution of amplitude associated with the fundamental frequency Figure 7. Comparison between the analytical transfer function (grey line) and experimental H/V spectral ratio (black line) obtained at two sites along TB-3 refraction profile Figure 8. Comparison between the analytical transfer function (grey line) and experimental H/V spectral ratio (black line) obtained at four sites along TB- refraction profile Figure 9. Comparison between the H/V spectral ratios obtained from microtremor measurements in 99 (grey line) and 008 (black line) near refraction profile TB Figure 0. Comparison between the analytical transfer function (grey line) and experimental H/V spectral ratio (black line) obtained at two sites along refraction profile TB Figure. Comparison between the analytical transfer function (grey line) and experimental H/V spectral ratio (black line) obtained well Kineret 6 (site ) Figure. Comparison between the analytical transfer function (grey line) and experimental H/V spectral ratio (black line) obtained well Kineret 0-B (site ) Figure 3. Schematic geological NS cross section beneath profile... 4 Figure 4. H/V spectral ratio (black line) and analytical transfer function (grey line) for representative sites of profile Figure. Schematic geological EW cross section along profile Figure 6. H/V spectral ratio (black line) and analytical transfer function (grey line) for representative sites of profile Figure 7. Seismic microzoning map of Tiberias presenting zones of common site effect characteristics Figure 8. Examples showing influence of thin upper soft layers on spectral accelerations computed for two sites located on the Cover Basalt.... 0

5 LIST OF TABLES Table. Stratigraphic table of the geological map of Tiberias (Sneh, 008)... Table. Brief description of wells located in the Tiberias region... Table 3. Parameters of earthquakes recorded by accelerometer stations used in this study. Distance is to the surface projection of the rupture.... Table 4. Geophysical and analytical models for calculating transfer functions at points located along TB-3 refraction profile Table. Geophysical and analytical models for calculating transfer function at sites located along refraction profile TB Table 6. Geotechnical data obtained from refraction surveys carried out in 99 and Table 7. Soil-column model for sites along refraction profile TB Table 8. Geotechnical data and soil-column for well Kineret Table 9. Geotechnical data and soil-column for well Kineret-0B Table 0. Ranges of S-wave velocities for litho-stratigraphycal units represented in the study area and used in calculating site response Table. Soil column models for representative sites of zones, their transfer functions and spectral accelerations....

6 6 ABSTRACT To quantify the seismic hazard across the town of Tiberias we used a methodology in which horizontal-to-vertical spectral ratio from microtremor (the Nakamura s technique) obtained on a dense measurement grid is utilized to assess the site-specific uniform acceleration spectra. This process of hazard assessment involves: a detailed mapping of the fundamental and other natural frequencies and amplitudes of H/V spectral ratios; compiling geological, geophysical and borehole data and integrating it with H/V observations to develop models of the subsurface at many sites across the study area. The subsurface model serves as an input for computing the expected Uniform Hazard Site-Specific Acceleration Response Spectra at the investigated sites. The final stage is generalizing the hazard by mapping zones that feature similar seismic hazard functions. Microtremor measurements were carried out at 7 sites, which are characterized by amplification from up to 8 in the frequency range Hz. The receiver function, which is horizontal-to-vertical spectral ratio obtained from earthquakes (shear wave) confirms the results obtained from microtremor records at three acceleration locations. H/V ratios, geological data and information from S-velocity refraction profiles enables construction of geological cross sections. Certain sharp differences in the H/V ratios have been interpreted as being associated with a subsurface discontinuity, i.e. fault. By comparison of the Uniform Hazard Acceleration Spectra calculated for probability of exceedance of 0% during an exposure time of 0 years and a damping ratio of % at more the 0 sites and in consideration of the constructed subsurface models, we subjectively divided the study area into eleven zones. The linear spectra for eight zones significantly exceed the design spectra required in the same area by the current Israel Standard 43 (IS-43) in the period range sec.

7 7 INTRODUCTION Figure. Tiberias - the New and Old together. Tiberias, famed as a city in the region where Jesus preached, as the capital of Herod Antipas, the seat of the Sanhedrin, and the place where the Jerusalem Talmud was written, is so rich in antiquities that archaeologists in Israel call it the City of Treasures. Tiberias now is a relatively small town (about 40,000 inhabitants), situated on the western shore of the Sea of Galilee on the seismically active Dead Sea Fault system, capable of generating earthquakes with magnitude as high as 7.. The long documented history of destructive earthquakes in Israel shows that the whole area, where modern Tiberias is now located, is subject to strong earthquakes, which have in the past caused considerable damage and many casualties. In present millennium several worth mentioning earthquakes occurred: for example the 033 in the Jordan Valley (massive destruction at Tiberias), 79 (walls of Tiberias collapsed, seiche on the Sea of Galilee), 837 the "Safed earthquake" (8% of the population of Tiberias were killed and city walls destroyed) and 97 (Tiberia suffered damage) according to Amiran, D.H.K (96). In order to mitigate earthquake risk and assess the site specific seismic hazard in urban areas, we must estimate the possible consequences of strong earthquakes, i.e., implement our accumulated experience of past earthquakes to present a scenario of an eventual earthquake. It is known that local ground conditions played an important part in the amount of damage suffered at any particular locality. Most examples from several destructive earthquakes during the two past decades, for example, in Mexico-City, 98 (Singh et al., 988; Reinoso and

8 8 Ordaz, 999), Spitak, Armenia, 988 (Borcherdt et al., 989), California, Loma Prieta, 989 (Hough et al., 990) and Northridge, 994 (Hartzell et al., 996), Kobe, Japan, 99 (Iwata, et al., 996), Kocaeli (Izmit), Turkey, 999 (Ozel et al., 00) Algeria, 003 (Hamdache et al., 004) have clearly shown that local site conditions can greatly increase ground shaking during an earthquake. The greater damage in Tiberias was, at least in part, due to the fact that it was founded on unconsolidated alluvium, which produced an exaggerated response. A better assessment of the expected ground motions inside the town is thus a key element for urban and civil protection planning. In the present study we used a three-step process for evaluating site effects and estimating their influence on seismic ground motion (Zaslavsky et al., 00). At the first step, we performed microtremor measurements on a dense spatial grid and H/V spectral ratios, from which we obtained a spatial distribution of the frequencies at which amplification is likely to occur and the expected level of amplification at those frequencies. H/V spectral ratios of S-waves, often known as receiver functions, generated by earthquakes and recorded at three accelerometer locations are considered in the analysis. At the second step, all available geological information, geophysical and well data are collected and incorporated as an aid to construct subsurface models for different sites within the investigated area. Finally, one-dimensional analytical models are used to predict site-specific acceleration response spectra from future earthquakes. The application of this methodology makes possible reliable assessment of disaster from different earthquakes, especially in the regions where big earthquakes present a long return period, but which exhibit a high seismic risk according to historical reports, population distribution and its socio-economic importance. GEOLOGICAL AND TECTONIC CONTEXT Figure presents the geological map of Tiberias at a scale of :0,000 compiled from Sneh (008), and Bogoch and Sneh (008) with an overlay of faults after Schulman (966). The town of Tiberias is located on the western shore of the Tiberias lake at the foot of the structural high of Poriya tilted block. The present Tiberias lake is a remnant lake that evolved from the ancient water bodies filling the Tiberias basin (the northern part of the Jordan Valley during the Pleistocene Holocene periods). There is continuous exposure of the Cover Basalt from the

9 9 elevation of -0 m at the south-eastern corner of the town to the Tel Ma on hill in the west, at +0m. Figure. Geological map of the study area compiled from Sneh (008), and Bogoch and Sneh (008) with an overlay of faults according to Schulman (966), and with locations of the refraction profiles TB-, TB- and TB-3 (Ezersky, 008); R- and R- (Shtivelman, 99) and profiles and for constructing cross sections. With the exception of the Upper Cretaceous rocks exposed in the structural highs of Poriya and Fuliya blocks, all the formations on the geological map are part of the Neogene. From

10 0 bottom to top these are: the Miocene Hordos Fm. and the Lower Basalt; the Neogene Bira Fm., Gesher Fm. and the Cover Basalt. The investigated area is dissected by two normal fault systems: the WSW-ENE transversal system with the down throw to the north, and the SE-NW system of step-faults with the down throw to the northeast. The two transversal faults in the south are of a Neogene pre-cover basalt age. They were rejuvenated in the Pleistocene. The NW trending stepfaults are of Pleistocene post-cover Basalt age. Along the greater part of their traces they bring basalt against basalt. Only at the southeastern termination of two of them, where they abut against a transversal fault, Neogene sediments rise to the surface. Here the throw of the two stepfaults is the greatest. A fourth step-fault is inferred within the lake and parallel to its shore. A significant feature is the considerable vertical displacement at the NE corner of the titled block, a result of the cumulative effect of the two fault systems. In the Upper Pliocene, the site of the town and its lakeshore were structurally higher than Tel Maon in the west (Schulman, 966). Schulman (966) proposed Ron et al. (984) supported that the middle to upper Miocene sediments and basalts underwent intensive deformation by horizontal shear in a compressive stress field which operated during the end of the Miocene and early Pliocene times. Stratigraphy and lithology The stratigraphic units are given in explanatory table to the geological map of Tiberias compiled and edited by Sneh (see Table ). Upper Cretaceous sediments. The upper part of the Sakhnin, Bina and Menuha fms. crop out in two areas of the lake shore: at Tel Raqat (Hirbet Fuliya) and foothills at Mt. Hordos (Berenice). According to Golani (96), these formations are about 60 m, m and 0-60 m thick accordingly. They consist of grey hard dolomite and lithographic limestone and chalk respectively. Upper Cretaceous sediments are penetrated by several wells situated at Tel Raqat, Mt. Hordos and Hamei Teveria. Information on depth of the Judea Gr. available from the wells is shown in Table. Eocene Chalky-Limestone Complex represented by the Avedat Gr. is exposed to the north of the study area, at Mt. Arbel. The Neogene deposits are divided into three formations Hordos, Bira and Gesher crop out along the Tiberias lakeshore. Fluviatile-lacustrine sediments of the Hordos Fm. including also the Hugog Cgl. comprise alternating red mudstone, sandstone, limestone and conglomerate.

11 Table. Stratigraphic table of the geological map of Tiberias (Sneh, 008) Thickness of the Hordos Fm. according to Shaliv (99) reaches 70 m in the Poryya escarpment. Six basalt flows are intercalated within the Hordos Fm. They thicken southward and form a continuous basalt section the Lower Basalt. Based on data from HZORM- well one can presume an increasing thickness of the Lower Basalt to the southwest as well. The rock is olivine basalt, usually porphiric. The basalt is intensely jointed with calcite-filled cracks.

12 The Bira and Gesher Fms. overly with slight unconformity the red beds of the Hordos Fm. in the mountain scarp along the shore of the lake. The Bira Fm consists of marly clay, siltstone or calcarenite and has a thickness of about -70 m. The Gesher Fm. up to m thick consists of chalky limestone. These sediments are overlain by the thick Cover basalt (Michelson, 987). The basalt flow is discordantly resting on the erosion surface of the Gesher beds (Heimann, 993) or the Bira Fm. It consists normally of hard olivine basalt. Basalt on the surface is weathered with sporadic patches of clay-alteration products of basalt. The thickness of the Cover Basalt increases from the east to the west from 30 0 m in the easternmost block of Tiberias and m up to 7 m thick west of it. Quaternary sediments are distributed along the lakeshore, the western part of the town of Tiberias, the Poriya escarpment and southwestern part of the study area. They are represented by anthropogenic (archeological) deposits along the lakeshore and in the old part of town, silty clay, conglomerate, mostly basaltic components, poorly cemented with layers of clay. Such a composition is typical for landslides found in the old part of town and Poriya escarpment. Table. Brief description of wells located in the Tiberias region EW NS Name Depth to the Elevation Judea Gr. m TD D ? D-96 > D-966 >3-98 > D-967 > ? D-968 > D > D-974 4? -00? D ? D-990/97 > H.TVRIA- > HZORM J- 8? -73? J- 0-60? K K K. / K0b 84-08? K

13 3 BRIEF REVIEW OF SEVERAL EXPERIMENTAL METHODS FOR SITE EFFECT ASSESSMENT Various empirical techniques have been used to detect locations where site effects are likely to occur. - S-Wave spectral ratio with respect to reference site The most common technique for estimating site response is the standard (classic) spectral ratio procedure first introduced by Borcherdt (970). This approach considers the ratio between the Fourier spectra of a seismogram recorded in the site of interest and the spectrum of a seismogram recorded at a reference site, which is usually the rock outcrop. This ratio can be considered as the transfer function between the bedrock and the surface assuming that the two recordings correspond to the same source, the same path effect and that the reference site has a negligible site effect. It is very difficult to implement all these assumptions in real conditions. First, in many cases we do not have a nearby bedrock site and therefore the condition that the path of the propagating seismic waves is the same is not fulfilled; second, it is known (e.g., Steidl et al., 996, Zaslavsky et al., 00) that weathered and cracked bedrock site exhibits a significant site effect, associated with frequency-selective ground motion amplification; third, there are many cases in Israel, when nearby bedrock outcrop is not the same rock at the base of the soil layer which is responsible for amplifying seismic waves amplitudes. It should also be noted that performing simultaneous measurements at two sites is often relatively costly. Nevertheless, when all the conditions are observed, this method maybe considered the most reliable estimate of the empirical transfer function of site. Many investigators used this method and evaluated site response functions from moderate to weak motion recording of earthquakes (Tucker and King, 984; McGarr et al., 99; Field et al., 99; Liu et al., 99; Carver and Hartzell, 996; Hartzell et al., 996; Steidl et al., 996; Zaslavsky et al., 000 and others). - Horizontal-to-vertical S-wave spectral ratio (Receiver Function) In this technique applied by Lermo and Chávez-García (993) the receiver function can be obtained from ratio between horizontal and vertical amplitude spectra computed at the same investigated site from S-waves, respectively. Receiver function was introduced by Langston (979) to determine the velocity structure of the crust and upper mantle from P-waves of teleseisms. Langston made the assumption that the vertical component of motion is not influenced by the local structure, whereas the horizontal

14 4 components, owing to the geological layering, contain the P to S conversion. In the spectral domain this corresponds to a simple division of the horizontal spectrum by the vertical. Many studies report that the frequency dependence of site response can thus be obtained from measurements made at only one station at the analysed site (Lermo and Chavez-Garcia 994; Malagnini et al., 996; Seekins, et al., 996; Theodulidis et al., 996; Castro et al. 997; Yamazaki and Ansary, 997; and others). Their results confirm the validity of the method to estimate S-wave site response. We obtained similar conclusion in our investigations (Zaslavsky et al., 000). Nevertheless, the implementation of this approach still requires a rather frequent occurrence of earthquakes. This requirement becomes an obstacle in regions of low seismicity. - Microtremor spectral ratio with respect to reference site Kagami et al. (98) proposed that the ratio of the spectra of the horizontal ground motions of the microtremor at the investigated site to those of a reference site can be used as a measure of the site response function. This method can be successfully applied for long period microtremors with period ranging from.0 to 0 sec. When higher frequencies are of interest, the distance between the measured sites should not exceed few hundred meters. The reliability of this method depends on whether or not the simultaneously measured motions at each site are from the same source and propagation path. This technique is widely used for site response estimates (Lermo et al., 988; Field et al., 990, 99; Rovelli et al., 99; Dravinski et al., 99, 003; Gaull et al., 99). However, experimental study of site effect by sediment-to-bedrock spectral ratio in urban and suburban regions can be successful only under particular circumstances, because microtremor would be influenced by local artificial sources generated by human activities which essentially change from place to place. - Horizontal-to-vertical microtremor spectral ratio Nakamura (989) proposed the hypothesis that site response function under low strain can be determined as the spectral ratio of the horizontal versus the vertical component (H/V) of motion observed at the same site. He hypothesized that the vertical component of microtremor is relatively unaffected by the unconsolidated near-surface layers. Hence, the site response is the spectral ratio between the horizontal component of microseisms and vertical component of microseisms recorded at the same location. Many authors, among them Lermo and Chávez-García (994), Seekins et al. (996), Toshinawa et al. (997), Chávez-García and Cuenca (998), Enomoto et al. (000), Shapira et al.

15 (00), Mucciarelli and Gallipoli (004), Murphy and Eaton (00), Maresca, (006), show that the H/V spectral ratio technique can be a useful tool for the assessment of ground motion characteristics on soft sediments. However, other authors (for example, Bonilla et al., 997; Horike et al., 00; Satoh et al., 00) conclude that whereas the predominant peak of H/V ratio is well correlated with the fundamental resonance frequency, the amplitude of this peak is not necessarily the amplification level as obtained from sediment-to-bedrock spectral ratio of earthquake records. MICROTREMOR RECORDING AND PROCESSING Microtremor measurements were carried out during the period from June to September 008 at 7 sites in an area of about 3 km. Measurements are conducted using portable instruments (Shapira and Avirav, 99) consisting of a multi channel amplifier, Global Positioning System (GPS) for timing and a laptop computer with 6-bit analogue-to-digital conversion card to digitize and store the data. In our experimental set-up, each seismograph station consists of three (one vertical and two horizontal) L4C velocity transducers (Mark Products) with a natural frequency of.0 Hz and damping ratio 70% of critical. The recorded signals are sampled at 00 samples per second and band-pass filtered between 0. Hz and Hz. All the equipment: sensors, power supply, amplifiers, personal computer and connectors are carried in a vehicle, which also serves as a recording centre. The seismometers are fixed on levelled metal plate placed directly on the ground. To study the characteristics of spectra of the microtremor signals, we compute Fourier spectra and spectral ratios. The record length (time window) used for spectral calculations depends on the fundamental frequency. The basic criterion is to choose the minimal time window which yields spectra that practically do not change when increasing the record length. We have concluded that at sites with fundamental frequencies of Hz (or more) we should use a record of at least 30 sec. At sites with lower frequencies, the time window should be increased to 60 sec. The selected time windows are Fourier transformed, using cosine-tapering ( sec at each end) before transformation and then smoothed with a triangular moving Hanning window. More precisely, we apply window closing procedure (see Jenkins and Watts, 968) for smart smoothing of spectral estimates so that any significant spectral peaks are not distorted.

16 6 The H/V spectral ratios are obtained by dividing the individual spectrum of each of the horizontal components [S NS (f) and S EW (f)] by the spectrum of the vertical component [S V (f)]: A S NS f NS f A f S f V f f S EW EW () S The average spectral ratio for each of two horizontal components is computed. If the curves of average spectral ratios of the two components are similar then the average of the two horizontal-to-vertical ratios is defined as: A f f f n S n S NS i n i S i i S V EW V V f i f i The measurement sites in Tiberias were designed with variable grid spacing. Different surface sedimentary deposits, thickness of sediments and shear wave velocity contrast between sediments and bedrock were considered in the design stage. In the process of accumulating the data and understanding the general picture of site effect distribution, we made operative decisions as regards changing the grid to gain reliability of the results obtained. Sharp changes in frequency over a short distance, disagreement with geological data and equivocal measurement results are the reasons for additional points and a denser grid. The densest network was deployed inside the old part of town with remains of historical buildings and walls. Unlike the area of the old town that is covered mainly by soft sediments, the greater part of Tiberias is covered by Pliocene basalt. Therefore, the spatial density of the measuring sites was decreased to a grid spacing of 00 meters. Distribution of the 7 measurement sites within the study area is shown in Figure 3. The local topography ranging from -00 m above sea level to +0 m and inaccessibility of some sites led to changing the spatial density of the measurements planned in the design stage. Figures 4 and present examples of the seismic station locations in the different geological and urban conditions. (6)

17 7 Figure 3. Location of the measurement sites in the study area. Numbers indicate the sites used as examples. TB-, TB- and TB-3 - refraction survey profiles (Ezersky, 008); R- and R- refraction survey profiles (Schtivelman, 99); TVR, TVR and POR accelerometer locations; Profile and Profile profiles for reconstructing subsurface structure.

18 8 Figure 4. Examples of seismometers locations during various sets of the site investigations in different geological conditions.

19 Figure. Examples of seismic station locations in Tiberias. 9

20 Spectral ratio Spectral amplitude, µm/s*s 0 RESULTS Site response in Tiberias estimated by H/V spectral ratio from microtremor Empirical estimation of site effects in Tiberias is carried out by implementing H/V spectral ratio from microtremor method. Figure 6 displays examples of the average amplitude spectra of one of the horizontal, the vertical components of motion and spectral ratios obtained at several sites in Tiberias (for location see Figure 3). A common feature of the presented examples is the appearance of a single peak in the H/V spectral function which also coincides with a peak in the amplitude spectrum of the horizontal motion. The vertical spectral component is almost flat. Figure 7 presents cases where the Fourier spectra show two frequency bands of site effect, manifested on the H/V curves by two resonance peaks. The second peak is most likely caused by an intermediate hard layer in the subsurface. While the first resonance frequency is related to the hard rock at depth, the position of the second resonance peak depends mainly on the thickness of the intermediate hard layer. Amplitude level of both peaks is determined by the S wave velocity in the soft sediments T T T T49 0 T34 T33 T79 T49 0. f 0 0. f 0 0. f 0 0. f Figure 6. Examples of average Fourier spectra (top) and H/V spectral ratios (bottom), which yield a single peak. The black line indicates a vertical spectral component; the grey line indicates the average of NS and EW horizontal components of motion.

21 Spectral ratio Spectral amplitude, µm/s*s T T T 0.03 T T T T66 T66 0 f 0 f f 0 f 0. f 0 f f 0 f Figure 7. Examples of average Fourier spectra (top) and H/V spectral ratios (bottom), which yield two resonance peaks. The black line indicates a vertical spectral component; the grey line indicates the average of NS and EW horizontal components of motion. Spectral analysis of microtremor measurements revealed another common feature characterizing both Fourier spectrum and spectral ratio obtained at a great part of measuring sites throughout the study area. This is a trough in amplitude of the vertical component of Fourier spectra in the frequency range Hz, whose origin is not clear. When the fundamental frequency of site is significantly higher than Hz, it looks like as a single trough in the vertical component (site 46 in Figure 8) and clearly visible looking peak is distinguished in the spectral ratio curve at frequency 0.3 Hz, while the fundamental frequency is Hz. Some spikes in the frequency range -8 Hz are generated by the various types of machinery operating nearby. However, we obtained at many sites the fundamental frequency in the range Hz. In this case, the typical picture is a wide-bottomed common trough like at site T3 or T6 in Figure 8. We note that if at site 3 the fundamental frequency albeit not plainly but can be seen at the vertical spectral component, at site T6 it is impossible divide two frequencies in the Fourier spectra. Only reprocessing with careful selection of microtremor time windows allowed separating these peaks. It is of interest to understand a possible nature of deviation of the horizontal and vertical spectral component in the low frequency range and appearance of peak at part of the

22 Spectral ratio Spectral amplitude, µm/s*s measurement sites. First of all, we tried to correlate presence of this peak with the surface geology. Three sites from example in Figure 8 are located at the outcropped Cover Basalt. However, there are a lot of sites in the similar geological conditions whose spectral ratios yield no peak at frequencies Hz. Together with this, we do revealed this peak at sites located on expose of the Bira Fm. in the central part of the study area (site 70 in Figure 9) and do not reveal it at Bira outcropped in the northeastern part (site in Figure 9). Similarly, we observe the peak in question at only one of sites T0 and T0 located on alluvium (Figure 9). Generally speaking, distribution of H/V spectral ratios yielding low frequency peak superimposed on the geological map does not show apparent correlation with lithological units in the study area T46 T3 T T46 T3 T6 f f 0 f f 0 f f Figure 8. Fourier spectra (top) and H/V spectral ratios (bottom) obtained at sites located on the exposure of the Cover Basalt. f 0 indicates the fundamental frequency of the measurement site; f is an artificial frequency. H/V ratios at site T6 obtained by processing and reprocessing of the measurement data are shown by dashed and solid lines respectively. It was interesting to look at H/V spectral ratios obtained from microtremor measurements carried out near the Town Hall in February 007 (site TVR) and during the current measurement campaign in August 008 (site T6). H/V spectral ratios shown in Figure 0 demonstrate similarity in both frequency and amplitude of the fundamental and second peaks. However, a

23 3 peak at frequency 0.4 Hz is clearly seen on the H/V curve at site T6 while it is absent at site TVR. Another example shows the results of two microtremor recordings performed in June 008 and July 008 at the same site T3. Spectral ratios (T3 and T3a in Figure 0) again demonstrate resemblance and all the difference is the presence of peak at frequency 0. Hz for one of measurements. T70 T f f 0 f T0 T0 f f 0 f Figure 9. H/V spectral ratio obtained at sites located on the outcropped Bira Fm. (T70 and T) and alluvium (T0 and T0). T TVR T T3a Figure 0. Comparison between the H/V spectral ratios obtained near the Tiberias Town Hall (T6 and TVR) and at site T3 at different times. This brief research aimed to find out whether the peak controlled by trough in the vertical spectra at low frequencies relates to the geological structure. Since we failed to find such a

24 4 correlation, we concluded that this peak should not be considered while developing analytical model of the subsurface. Comparison of H/V spectral ratios from microtremor and seismic events The site response estimated from microtremor measurements we compared with that obtained from two local earthquakes occurred in 004 and recorded by accelerometers. Locations of the strong motion stations used in this study are shown in Figure 3. Figure shows two horizontal and vertical components of accelerograms from two seismic events given in Table 3 and recorded at site Tiberias Hotel (TVR). The accelerograms demonstrate the considerable differences in amplitude and duration that characterize horizontal and vertical components. In terms of peak acceleration, amplitudes recorded at horizontal components are more than twice larger than at vertical components. The quasi-monochromatic nature of the motion of horizontal components strongly suggests sediment resonance. Horizontalto-vertical spectral ratios (NS and EW components) for earthquake occurred on.0.04 indicate amplification about near 0.8 Hz significantly and higher effect in the frequency range - Hz. Analysis of the earthquake occurred in the North Dead Sea reveals significant difference in amplification ground motions between EW and NS components in the frequency range - Hz and the low frequency peak. We note that the low frequency peak, which is clearly seen on H/V ratio from February earthquake is missing in the H/V ratio from July Earthquake. Figure depicts acceleragram recorded at site Tiberias (TVR) from the Dead Sea earthquake (February 004). Horizontal motions at this site are about three times higher than the vertical ones. While receiver function of NS component shows a clear peak characterized by amplitude of 3 at frequency 0.8 Hz, EW component is different and reveals two peaks at frequencies 0. Hz and 0.8 Hz. Figure 3 displays the three components of accelerograms from two earthquakes recorded at site Poriyya Hospital. One can see that in both cases the shear waves on the horizontal components exhibit larger amplitude than the vertical components. However, the difference is significant noticeable on EW components. The amplification in the time domain is comparable to that seen in the frequency domain and the common feature of H/V curves from both seismic events is a clear peak at frequency 0. Hz.

25 Table 3. Parameters of earthquakes recorded by accelerometer stations used in this study. Distance is to the surface projection of the rupture. No. Recording site Date Time Geographic coordinates Distance M L Lat.(N) Long.(E) (km) Poriya Hospital 7 Tiberias 04/0/ 08: Tiberias Hotel 0 Poriya Hospital 04/07/07 4: Tiberias Hotel 89 Epicentre region Dead Sea 8 North Dead Sea a b Figure. Accelerograms of the earthquakes recorded at site Tiberias Hotel (TVR): (a) earthquake occurred in the Dead Sea (004 0-, 08:, M L =., R=0 km) and (b) earthquake occurred in the Dead Sea fault ( , 4:3, M L =4.7, R=89 km);

26 6 Figure. Accelerograms of the earthquake occurred in the Dead Sea (004 0-, 08:, M L =., R=0 km) and recorded at site Tiberias Town Hall (TVR). a b Figure 3. Accelerograms of the earthquakes recorded at site Poriya Hospital: (a) earthquake that occurred in the Dead Sea basin (004 0-, 08:, M L =., R=0 km) and (b) earthquake that occurred in the Dead Sea fault ( , 4:3, M L =4.7, R=89 km).

27 7 Figure 4 presents a comparison between the average H/V spectral ratios from accelerograms recorded at strong motion stations (Receiver Functions) and spectral ratios obtained from microtremor measurements recorded at the same sites in different years. Figure 4a shows the spectral ratio of EW component of accelerogram from the earthquake occurred in February, 004, and recorded at strong motion station Tiberias Hospital (TVR). The fundamental peaks of the earthquake and microtremor spectral ratios at frequency 0.8 Hz look surprisingly similar. However, the amplification range observed in the receiver function from Hz up to 4 Hz has shifted toward higher frequencies in the spectral ratio of microtremor. It is noteworthy that the peak at Hz is not a resonance frequency but a result of soil-structure interaction. It is confirmed by conducting of ambient vibration test on the roof and at the basement of building where the strong motion station was installed. Peak at frequency of Hz is interpreted as the fundamental frequency of the building so it is practically not visible in the spectral ratio from microtremor obtained at site situated 00 meters from the accelerometer location. In the case of Tiberias (TVR) strong motion station shown in Figure 4b both the fundamental frequency and amplification factors determined from the February earthquake and microtremor recorded in 007 and 008 concur well. The curves of spectral ratio obtained from microtremor have an additional peak at. Hz with amplitude about. This peak is rather relates to topography. The horizontal-to-vertical spectral ratio from microtremor recorded at site Poriya Hospital shows fundamental peak identical to that identified by the receiver function considering both the resonance frequencies and amplitude (Figure 4c).

28 8 (a) (b) (c) Figure 4. Comparison of different estimates of site amplification based on H/V spectral ratio techniques applied to earthquakes and microtremor recordings at sites TVR (Hotel) (a); TVR (Town Hall) (b) and Poriya Hospital (c). DISTRIBUTION OF THE RESONANCE FREQUENCY AND ITS ASSOCIATED H/V AMPLITUDE OVER THE STUDY AREA The increased intensity of the damage during earthquakes is, to a great extent, correlated with resonance effects, therefore mapping of resonance frequencies and their associated H/V amplitudes is very useful for at least a qualitative assessment of the seismic hazard. Figure presents maps of the contoured fundamental resonance frequency (f 0 ) and the associated H/V amplitude. The data exhibit peaks changing from to 8, occurring at frequencies Hz. The western part of the study area is mostly characterized by H/V spectral ratios with a single peak at the fundamental frequencies.0-.3 Hz and amplitude less than 3 and is associated probably with dolomite of the Judea Gr. Irregularly appearing second resonance peak is related to the soft

29 Profile 9 alluvial layer and has relatively high amplitude. Such high amplitude values (-7) but associated with the fundamental peaks at frequencies 4-7 Hz are observed in the northwestern part of the study area. We suppose a change of the fundamental reflector. It may be the limestone of the Gesher Fm., whose local outcrop is marked on the geological map, while the Cover Basalt is eroded in this part. These two areas are separated by a sublatitudinal fault mapped also by the geological data (transversal fault of Mizpa according to Shulman (966)) Fuliya f, Hz Rock Tiberias Rabbi Aqiva Mizpa 3 Nasr ed Din Profile Ein et Tina Herodes Poriya Fault detected by shift in H/V frequency Fault according to Sneh, Fault according to Schulman, Figure. Distribution of the fundamental resonance frequency over Tiberias.

30 Profile Amplification Rock Profile Fault detected by shift in H/V amplitude Figure 6. Distribution of amplitude associated with the fundamental frequency. Decrease in the fundamental frequency is observed in the central part of the study area, which is in turn subdivided by the Mizpa fault into northern and southern parts with characteristic resonance frequency 0.7 Hz and Hz, respectively. This down-dip block is located between the NW trending faults with the downthrow to the northeast (Shulman, 966). The eastern one (the Nasr ed Din fault) is mapped also by Sneh (008). The fault of Rabbi Aqiva, however, is not identified by the microtremor measurements. A wedge-shaped structural block is distinguished in the study area owing to higher fundamental frequency values (.-4 Hz) within the field of Hz. This block is limited by the faults. The eastern one, NW-SE directed, coincides with the southeastern segment of the Tiberias fault. Its continuation to the northwest is

31 not detected by shift in the H/V fundamental frequency. The fault delineating this uplifted structural block from the southwest is the most likely extension to the northeast of the Poriya fault. This interpretation is supported in the geological map of Tiberias edited by A. Sneh (see Figure ). The fault of Poriya which is traced by N. Shulman (966), Sneh (008) and B. Medvedev (008) is also clearly detected by the H/V analysis in the segment along the Judea outcrop; however near the southeastern edge of the study area we could not trace it accurately due to the sparse measurement network. We obtained almost flat H/V ratios with no resonance frequency on the Upper Cretaceous rocks exposed in the structural high of the Poriya block. The fundamental frequency in the range of - Hz characterizes sites adjacent to the Judea outcrop from the south. An area located at the lakeshore eastern of the Poriya fault is characterized by the fundamental frequency in the range of.-.7 Hz and the close second resonance frequency (.-. Hz) produced by the talus. Two faults of SW-NE direction are identified by shift in the H/V fundamental frequency. One of them is a segment of the fault of Herods (Schulman, 966) and limits the Judea outcrop from the northwest. Another fault is detected at a distance of 00 m to the north by changing frequency from.-.7 Hz to.-3 Hz. The northeastern part of the study area is characterized by gradual increase of the fundamental frequency from 0.8 Hz up to -. Hz. This increase in the frequency is explained by the thinning sediment layers above the Judea Gr. due to the sharp topography of the erosion surface of the lakeshore. The main feature of the H/V ratios obtained at sites located on the slope is two inseparable resonance peaks. The first one is associated with the Judea Gr. and the second one is caused by impedance contrast between clayey marl of the Bira Fm. and underlying layers. The second resonance frequency varies from. Hz up to Hz. Distribution of maximum amplitude associated with fundamental H/V peak (Figure 6) retains the general trends characterizing the frequency map, however a correlation with only part of the faults is revealed. These faults were taken into account in the map constructing. The amplitude varies from to 3 at a great part of sites in the study area. The higher (up to 7) values are attained in those areas, where the higher fundamental frequency values are observed. In these areas there is the thick alluvial layer and there is no the Cover Basalt. The amplitudes up to are also attained at sites located in the northeastern part of the study area, where the Bira marl outcrops at the lakeshore. The variations of amplitude associated with the second resonance peak are connected with local variations of Vs in the upper soft part of the geological section.

32 ESTIMATION OF SHEAR-WAVE VELOCITY MODELS AND RECONSTRUCTION OF SUBSURFACE STRUCTURE A prerequisite of a reliable analytical model for site response estimation by using computer codes such as SHAKE (Schnabel et al., 97) is the knowledge of the local geology, including spatial distribution of softer materials above the hard bedrock with corresponding S- wave velocity of each layer. Densities and specific attenuation in different lithological units were selected from of literature sources (Borcherdt et al., 989; McGarr et al., 99; Theodulidis et al., 996; Reinozo and Ordas, 999; Pergalani et al., 000 and many others). Recently, Pratt and Brocher (006) used spectral decay in the shear-wave spectral ratio with respect to reference site amplification curves and estimated Q-values for shallow sedimentary deposits. They concluded that the range of Q values is These values agree well with those used in our studies. Data collected from a few seismic refraction profiles provide information on the S-wave velocities and thickness of shallow sediments within the accuracy and resolution of the geophysical technique. Refraction profiles are normally designed to obtain maximum information on Vs of the lithological units represented in the study area and in the vicinity of boreholes. However, in the area of special interest in terms of both the geological conditions capable of producing site effects and the location of historic Tiberias archaeological remains, considering that part of ancient town is still un-excavated, we found only two locations appropriate for deploying the refraction survey equipment (see TB and TB3 profiles in Figure 3). Refraction profiles TB- and TB-3 provide us P- and S-wave velocities on upper 30 meters represented by the alluvial deposits (Ezersky, 008). According to the geological data, in this part of the study area the Quaternary sediments are underlain by the Miocene Hordos Fm. consisting of conglomerates and limestone over the dolomite of the Judea Gr., which is the fundamental reflector. Two upper layers determined by the refraction survey and characterized by S-velocities 80 m/se and 340 m/sec are alluvium. The third layer with Vs=470 m/sec is probably talus. Lacking direct Vs measurements of the dolomite in the well, we adhere to Vs assigned to Vs= m/sec for dolomite of the Judea Gr. suggested by a refraction survey in the Parsa area located in the Dead Sea area (Zaslavsky et al., 000) and in the town of Dimona (Zaslavsky et al., 008) and used everywhere in the previous studies. Thickness and velocity of the Hordos Fm. are fitted. Figure 7 shows the analytical transfer functions matching the experimental

33 3 spectral ratio for two sites located along TB-3 profile. Strictly speaking, site 8 is located less than one hundred meters south of the refraction profile, therefore we slightly adopted also thickness of the upper layers known from the refraction survey. Table 4 presents the geophysical and optimal models for sites and 8. T T Figure 7. Comparison between the analytical transfer function (grey line) and experimental H/V spectral ratio (black line) obtained at two sites along TB-3 refraction profile. Table 4. Geophysical and analytical models for calculating transfer functions at points located along TB-3 refraction profile Site 8 Geophysical data TB-3 Analytical model Layer Soil Thickness, Vs, Thickness, Vs, Density, Damping, No. m m/sec m m/sec g/cm 3 % Alluvium Talus?? Hordos Dolomite (Judea Gr.) Alluvium Talus?? Hordos Dolomite (Judea Gr.) Similarly to refraction profile TB-3, TB- provides us very scarce information on velocity and thickness of layers, characterizing the geological section in the area of historic Tiberias. Two layers, presumably alluvium and talus or movement material, are detected in the S-wave velocitydepth section (Ezersky, 008). In this case, the procedure of adjustment of the analytical model to