| Download | Main Page |
|
MICROZONING OF THE EARTHQUAKE HAZARD
IN ISRAEL
PROJECT 5 EMPIRICAL DETERMINATION OF SITE EFFECTS FOR THE ASSESSMENT OF EARTHQUAKE HAZARD AND RISK TO BEIT SHEAN & AFULA November, 2005 Job No 569/175/05 Principal Investigator: Dr. Y. Zaslavsky Collaborators: Galina Ataev, Marina Gorstein, Dr. Avraham Hofstetter, Michael Kalmanovich, Dagmara Giller, Ilana Dan, Nahum Perelman, Tatyana Aksinenko, Vadim Giller, Ion Livshits, and Alexander Shvartsburg Submitted to: Earth Sciences Research Administration Nation Ministry of Infrastuctures and The Ministry of Absorption |
 |
| ABSTRACT |
|
The long documented history of destructive earthquakes in Israel shows that the whole area is subject to strong earthquakes, which have, in the past, caused considerable damage and many casualties. Similar earthquake to that occurred in January 749 and destroyed structures of the Roman-Byzantine period in Bet Shean will shake this region with its residents, buildings and facilities. In the present study forming part of a project "Microzoning of the seismic hazard in Israel" we once again used for evaluating site effects and estimating their influence on seismic ground motion a three-step process. At the first step, microtremor measurements with a dense grid were carried out and interpreted in order to map the predominant frequency and maximum relative amplification of ground motion. At the second step, all available geological information and well data were collected and incorporated as an aid to construct subsurface models for different sites within the investigated area. Finally, one-dimensional analytical models were used to predict site-specific acceleration response spectra from future earthquakes. The horizontal-to-vertical spectral ratios of ambient noise were used to approximate the fundamental resonance frequencies of the subsurface and their associated amplitudes. About 210 and 300 sites were instrumented in the towns of Bet Shean and Afula, respectively. The soil sites exhibit H/V peak amplitudes ranging from 2 to 7 in the frequency range 0.9 Hz to 13 Hz for Bet Shean. H/V spectral ratios in the Afula area reveal two resonance peaks corresponding shallow and deep reflectors. The first resonance frequency varies from 0.35 Hz up to 12 Hz with amplitudes of 2-8 units. The second resonance peak shows amplitude of units 2-8 in the frequency range 1 Hz to 10 Hz. These results imply significant variations in the shear-wave velocities across the area and considerable variations of sediments thickness. S-wave velocity profiles derived from limited geophysical and borehole data enabled the calibration of the H/V spectral ratios at corresponding locations by analytical site response functions calculated for 1D subsurface model. Velocity models further were used for estimating subsurface structure at sites where information on soil column was scarce. Certain sharp differences in the H/V ratios have been interpreted as being associated with a subsurface discontinuity, i.e., an unmapped fault. The evaluated subsurface models are introduced using the SEEH procedure of Shapira and van Eck (1993) to assess Uniform Hazard Site-Specific Acceleration spectra (probability of exceedence of 10% during an exposure time of 50 years and a damping ratio of 5%) for different zones within the town of Bet Shean and Afula. The shape of the linear spectra obtained for all zones differ significantly from those prescribed by Israeli Building Code (IS-413). The Code requirements essentially underestimate the accelerations in the period range from 2 to 0.1 sec. We have estimated nonlinear response spectra for all zones and not found significant difference between the linear and nonlinear approaches for site classes B and C at peak input motions of 0.1g to 0.2g. Based on these comparisons, a nonlinear approach is recommended for site classes C, D and E for input motion greater than a few tenths of the acceleration of gravity. |
| 1. INTRODUCTION |
|
Most examples from several destructive earthquakes during the two past decades, for example, in Mexico-City, 1985 (Singh et al., 1988; Reinoso and Ordaz, 1999), Spitak, Armenia, 1988 (Borcherdt et al., 1989), California, Loma Prieta, 1989 (Hough et al., 1990) and Northridge, 1994 (Hartzell et al., 1996), Kobe, Japan, 1995 (Iwata, et al., 1996), Kocaeli (Izmit), Turkey, 1999 (Ozel et al., 2002) Algeria, 2003 (Hamdache et al., 2004) have clearly shown that local site conditions can greatly increase ground shaking during an earthquake. Therefore an accurate estimation of the seismic ground motion across the cities is of prime importance for urban developments and mitigation of seismic risk. In 2001 a special team was formed in the Geophysical Institute of Israel to map site effects in different areas across in Israel. Up to now, seismic mirozonation studies in the towns of Lod and Ramle, Qiryat Shemona, Kefar Sava, Dimona, Arad, the Coastal Plain, Hashefela and Qrayot (in the final stage) regions have been carried out. Detailed results are presented on the webpage www.relemr-merc.org. Choice of next object for microzonation study was dictated by the following concurrence of circumstances: The long history of earthquakes does not leave place for doubts that sooner or later an earthquakes as large as, or larger than the earthquake that occurred in January 749 and destroyed structures of the Roman-Byzantine period in Bet Shean will shake this region together with its residents, buildings and facilities. We know that effect of local site conditions on ground motion plays a major role in the shaking levels and hence should be seriously incorporated into seismic hazard estimations at a specific site. This is particularly important for both Bet Shean and Afula where we have strong impedance contrast between the soft sediments and the underlying bedrock represented by the basalts of Pliocene and Miocene ages. In addition, over the area of study the soil conditions change significantly from place to place. We are fully confident that shaking levels of future earthquakes can be predicated with sufficient details in order to consider in the seismic design of buildings and other structures. We believe that some of destructions might have been avoided if more information regarding resonant frequencies of the ground had been available. In the present study forming part of a project "Microzoning of the seismic hazard in Israel" we again used a three-step process for evaluating site effects and estimating their influence on seismic ground motion (Zaslavsky et al., 2005). At the first step, microtremor measurements with a dense grid were carried out and interpreted in order to map the predominant frequency and maximum relative amplification of ground motion. In the second step, all available geological information and well data were collected and incorporated as an aid to construct subsurface models for different sites within the investigated area. Finally, one-dimensional analytical models were 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. This report presents the results of main tasks performed within the framework of the project on evaluation of ground shaking characteristics in Bet Shean and Afula. |
| 2. BRIEF REVIEW OF SEVERAL EXPERIMENTAL METHODS FOR SITE EFFECT EVALUATION |
|
Among the empirical technique for site response estimating were summarized and discussed (Field and Jacob, 1995; Kudo, 1995, Lachet et al., 1996; Satoh et al., 2001 and others). Areas of high seismicity present opportunities for determining the site response functions through analysis of recorded ground motion during an actual strong event by comparison with recordings at a nearby reference site located on rock (Jarpe et al., 1988, 1989; Darragh and Shakal, 1991; Satoh et al., 1995; Hartzell, 1998; Reinoso and Ordaz, 1999 and others). The method for assessment site response function by ratio between the spectrum observed on a site of interest and the spectrum from the same source recorded at reference site have involved since Borcherdt (1970) and usually refers as the Standard Spectral Ratio (SSR). In Israel, where the seismic activity is relatively low, using strong motion data for analysis of site effect is impractical (Zaslavsky et al., 2003c; in Hebrew). Many investigators (McGarr et al., 1991; Field et al., 1992; Liu et al., 1992; Carver and Hartzell, 1996; Steidl et al 1996; Zaslavsky et al., 2000b; Lebrun et al., 2004 and many others) use SSR that to evaluate site response functions from moderate or weak motion recording of earthquakes.
The basic assumption of SSR method is that signal recorded in reference site (usually a bedrock outcrop site) presents the input motion of the base of the soil layers site. In many cases (Steidl et al., 1966; Zaslavsky et al., 2002) the weathered and cracked bedrock can have a site response of their own. It my be note that use these surface rocks as a reference sites often lead to underestimating of the amplification by factor 2-4 in frequency range which is within the range of engineering interest. Computation of spectral ratios relative to a bedrock reference station widely used to analyze microtremor records with purpose to study site effects. This approach can be valid when on reference and investigated sites microtremor originated from the same source. Kagami et al. (1982, 1986) concluded that microtremor generated by distant oceanic disturbances can be used as a measure of ground motion amplification. Really, law frequency microtremor over relatively short distances should have similar source and path effects. The applicability high frequency microtremor in site effect studies has been investigated by several studies (Field et al., 1990; Rovelli et al., 1991; Lermo and Chávez-García, 1994; Gaull et al., 1995; Zaslavsky et al., 1995; Ojeda and Escallon, 2000; Horike et al., 2001 and other). They consider it possible to estimate resonance frequency and amplification of sediments from high frequency microtremor by SSR method. However, in urban and suburban areas, microtremor is generated by human activities and intensity of microtremor source may essential change from place to place. Therefore, this method should be used within limited areas (a distance from reference station some hundred meters). Nakamura (1989) hypothesized that site response could be estimated from the spectral ratio of the horizontal versus vertical component of microtremor (HVSRN) observed at same site. The technique HVSRN widely using for estimating site amplification factors of S wave due to earthquake. Lermo and Chávez García (1994), Chávez García and Cuenca (1998) show that result obtained from H/V ratios agree with SSR method of the S-wave of weak and strong motion. Seekins et al., (1996) reported comparison of earthquake and microtremor using traditional station-par SSR method in order to clarify the applicability of microtremor data to ground motion prediction. They pointed out that microtremor station pair method may be use only for identifying the frequency of the fundamental resonance of soil site while Nakamura's method resulting are similar to those that derived from S-wave station-pair ratio. Zhao et al. (2000) also concluded that the H/V ratios of microtremor almost coincide with those of seismic motion. The experiment of Enomoto et al. (2000) based on simultaneous recording microtremor at basement and surface using borehole at two sites. From these results H/V ratio is much more coincided to the theoretical transfer function due to SH wave propagation in the surface soil layer than the observed transfer function used simultaneous microtremor measurements between the basement and surface. Toshinawa et al. (1997) show that weak motion amplification factors for peak ground acceleration with respect to rock site correlated well with peak values of HVSRN. Konno and Ohmachi (1998), support HVSRN ratios because for high frequency microtremor their results agree with theoretical site amplification factors. Teves Costa and Senos (2000) presented predominant frequencies and respective amplification factor for the Lower Tagus Valley obtained with microtremor analysis. They concluded that the distribution of predominant frequency is in good correlation with the surface geology; however authors do not make comments on distribution of amplification factor. To the contrary, Horike et al. (2001) showed based on array measurements of microtremor, that amplification factor can be inferred from horizontal component ratio to a reference site, but HVSRN do not agree with site amplification factors. During recent years, more field studies suggest that the amplification determined from HVSRN exhibit clear peak that is well correlated with the resonance frequency but amplitude of this peak different from S-wave amplification determined by SSR (Teves Costa et al., 1996; Lachet et al., 1996; Bonilla et al., 1997; Satoh et al., 2001). Nevertheless, Malagnini et al. (1996) pointed out that H/V spectral ratio failed not only in identification amplification level but also of fundamental resonance frequency. It is necessary to remind that many parameters of recording influence data quality and its reliability: instruments, weather and ground conditions, soil-structure interaction, traffic, different almost-harmonic motion generated by the several of machinery operating in urban areas and others. Mucciarelli (1998) showed that under proper measurements conditions Nakamura's technique provides stable and reliable results, but great care has to be devoted to avoid factors that may significantly and adversely affect the results. Recently a European project SESAME (see detailed results on the webpage http://sesame-fp5.obs.ujf-grenoble.fr/) was initiated aiming to study the site effects assessment techniques using microtremor. Guidelines prepared within the framework of this project included the experimental factors affecting the H/V ratio results, the standardized processing software and the recommendations with regard to the experimental evaluation (Atakan et al., 2004), will be helpful for site effect assessment. In a resent comprehensive study of Nakamura's method (Bard and SESAME participants, 2004) it has been concluded that the microtremor HVSRN does allow to identify the site fundamental frequency. However, they pointed that the microtremor H/V peak amplitude is smaller than actual spectral amplification factor measured by standard spectral ratio. It is farther concluded that the H/V amplitude at that fundamental frequency may serve as a lower bound of the expected amplification level. Lermo and Gháves-Garsía (1994) attempted to apply the Nakamura's method to study the intense shear wave part of strong or weak earthquake recording in different cities in Mexico. They found that in all cases the results give a robust estimate of the frequency and amplitude of the first resonant mode. Early, this technique has been applied to studies of the Earth's interior from teleseismic P-waves (Langston, 1979) and usually refers as receiver function. Many studies s report that the frequency dependence of site response can be obtained from measurements made at only one station at the analyzed site. For example, Yamazaki and Ansary (1997) used horizontal-to-vertical Fourier spectrum ratio (receiver function) of accelerogram consist 2166 three component sets from 387 events. The stability of the spectrum ratio they explained by the transfer function between the transfer function between the ground surface and soft-soil outcrop due to S-wave propagation. Zaslavsky et al. (2000) and Moya et al. (2000) demonstrated that sediment-to-bedrock spectral ratio and receiver function are remarkable similar for fundamental frequency and amplification factors of a site. Mucciarelli et al. (2004) compared stability horizontal-to-vertical spectral ratios, composed of 674 triggered noise record and 132 earthquakes and showed that resonance peaks obtained with two different data sets are very similar as in frequency as amplitude. The present report shows that ambient vibration measurements can be used to obtain reliable information related to seismic behavior of sediment layers (in linear conditions) with thickness from 20 to 600m; and appropriate ensembles of carefully selected windows of ambient vibration time segments and careful analysis of the horizontal and the vertical amplitude spectra and their ratios, provide estimates of the site response similar to those obtained from the H/V spectral ratio of seismic events. |
| 3. H/V SPECTRAL RATIO TECHNIQUE |
|
The Nakamura's technique has been applied to the microtremor measurements carried out in the presented study. Very simplified geological section for the town of Bet Shean is presented in Figure 1,. On this factual example, we shall give simple explanations of the Nakamura's technique. Main assumptions of Nakamura's approach are: If we follow these assumptions and for a geological situation shown in Figure 1, site amplification can be defined as expression Se(ω)=Hs(ω) / Hb(ω) where Hs(ω) and Hb(ω) are the horizontal amplitude spectrum and the ground surface and at the bedrock. In the absence of amplification in the vertical component, microtremor source spectrum can be expressed as ratio of vertical Fourier spectrum at the surface and bedrock, i.e. As(ω)=Vs(ω) / Vb(ω) After normalization by spectrum allowing removing the unknown source effects from the soil amplification we obtain transfer function of the soil layer Sm(ω) = Se(ω)/ Ae(ω) = [ Hs(ω)/ Vs(ω)] / [ Hb(ω)/ Vb(ω)] Since according to Nakamura's assumption Hb(ω) / Vb(ω)=1 Transfer function may be expressed as following: Sm(ω) = Hs(ω)/ Vs(ω) Or, by the other words, the vertical component of ambient microtremor on the surface retains the characteristics of horizontal component of the bedrock. |
| 4. GEOLOGICAL SETTING | ||||||||||||||||||||||||||||||||||||||||||||||||
|
Figure 2 presents the geological map of Israel to a scale of 1:200,000 (Sneh et. al., 1998) in the Lower Galilee, where the studied areas of Afula and Bet Shean are located. The Lower Galilee is characterized by tilting blocks bounded by listric faults. There is no clear structural and morphological expression of the Syrian Arc fold belt in this region. The principal fault structures were established in the Miocene, in relation to the tectonic activity of the Dead Sea Fault, and were active mainly in Plio-Pleistocene times, when the Dead Sea Rift was formed. The Miocene faults are mainly right lateral strike-slips, whereas the later are characterized more by vertical displacements of up to a few hundred meters (Garfunkel, 1981). The NW-SE trending tectonic depression of the Yizreel and Herod valleys meets the Jordan Rift in the Bet-Shean Valley area. The region is underlies by a volcano-sedimentary sequence up to 800 m thick of Miocene to Pleistocene age. The Bet Shean valley is separated from the Jordan Valley by a morphological escarpment 30-50 mw high, which is the surface expression of the western marginal fault of the Dead Sea Rift. The town of Afula is situated in the central part of the Yizreel Valley, while the Bet Shean town is built above the western marginal fault of the Dead Sea Rift which is an active fault (Zilberman et al., 2004). The investigated areas include threemorphotectonic units: lower Afula (Yizreel Valley) unit (+60m); the Givat Hamore unit (+517m); the town of Bet Shean (-120m in the west and -200m in the east). | ||||||||||||||||||||||||||||||||||||||||||||||||
|
4.1. BET SHEAN
Bet Shean is located around the junction of the aligned Harod-Bet Shean valleys with the Dead Sea Rift. It is bounded in the north by Ramot Yissakhar. The area in underlies by Neogene-Quaternary sequence, which consists of a fluviatile- lacustrine sediments interbedded with and volcanic and pyroclastic rocks, included in the Tiberias Group. This sequence, which is 500-800 m thick (Gardosh and Bruner, 1998) unconformably overlies Cretaceous- Tertiary rocks. The hard basalt units, which are interbedded in the Pliocene sequence, can be assumed as potential reflectors and therefore they were the subject of our geological analysis is this area. The geological map presented in Figure 2 is based on studies of Hatzor (2000), Rozenbaum et al., (2004) and also our own field investigations. Tiberias Group. The Tiberias Group includes the Hordos Fm. and the Lower Basalt unit. The typical section of the Hordos Fm. consists of a thick fluvial-lacustrine sequence, which interfingers with several basalt flows which are the northernmost extensions of the Lower Basalt (Schulman and Rosenthal, 1968). According to Shaliv et al. (1991) the Hordos Fm. is subdivided into two members: Lower Conglomerate up to 20m thick consisting of poorly sorted and poorly rounded pebbles cemented by carbonate; and Clastic-Carbonate Member sequence thickening eastwards from 30m up to 160m. The second member consists of alternations of thinly bedded red calcilutite, sand, dolomite and chalk. The age of the Hordos Fm. ranges from Early Miocene to Lower Late Miocene. The Lower Basalt reaches the thickness of 630 m in Ramat Yissakhar and extends as far as the Gilboa in the south (Schulman, 1962). The Lower basalt is of the alkali-olivine type, finely crystalline, with partially altered olivine, pyroxene, and plagioclase, olivine, pyroxene, and ore minerals, in a trachytic texture. The age of the Lower Basalt in the Bet Shean and Harod valleys ranges from 17.5 to 10.3 Ma (Shaliv et al., 1991). The lower basalt is penetrated by Sede Nahum-2 well at a depth of 217 m 2 km away from the study area near Ramat Yissakhar. Lithological section of Sede-Nahum-2 borehole is shown in Figure 4 . Dead Sea Group The Dead Sea Group, Zak (1967) overlies the Tiberias Group. In the study area it consists of the Umm Sabune Conglomerate, Bira Fm., Intermediate Basalt, Gesher Fm., Cover Basalt, Wadi Malih Conglomerate, Lisan Fm., and Bet Shean Travertine (tufa). The Cover Basalt builds the top of the south-west tilting block of Ramot Yissakhar and it is also exposed in the northeastern part of Nahal Harod stream-channel (See Figure 2). The Cover Basalt is well known from boreholes in Harod Valley, where it reaches a thickness of a few tens of meters. "The basalt flowing from Ramot Yissakhar towards the northern Gilboa area, was blocked by the already-existing Gilboa escarpment" (Rosenthal, 1972). The radiogenic age of this basalt in the Nurit area is 5.2 Ma and in Nahal Avinadav is 4.9 Ma, Pliocene (Shaliv et al., 1991). The outcrops of the Basalt units and the tufa were mapped more accurately in the study area (see Figure 3 ). The Bet Shean travertine (tufa), is rather local unit extending along a belt about 10 km long in north to south direction and 4-6 km in wide extending from the Gilboa margins eastwards. It was first described by Picard (1943). Shaliv et al. (1991) subdivided it into the following two members: the travertine Member of 60m thick found, for example, in Ain Ashtori T/6 well (see Figure 4) and the marl Member, which overlies the travertine Member and is about 30m thick. Rozenbaum, A. G. et al., (2004) divide tufa into two main facies: first is Phytoherm framestone and phytoclast tufa facies, and second is intraclast tufa and Cyanolith "oncoidal" tufa. "The travertine units are the stratigraphic markers permitting an identification of young tectonic displacement in the Bet Shean area." The Bet Shean travertine is exposed in the study area along the creek of the Harod River and along the NS oriented morphological escarpments, which separates the Bet Shean valley from the Jordan Valley. The contours of the travertine shown in the Geological map ( Figure 3 ) are copied mainly from the Preliminary map of tufa outcrops in the area of Bet Shean Valley (Rozenbaum et al., 2004). Holocene sediments are represented by colluvium, alluvium and anthropogenic sediments. These sediments are exposed in the southwestern part of the study area. They consist of few meters of topsoil and in the eastern part contain scattered tufa fragments in a gray silty matrix up to 4-5m thick. These sediments overlay tufa. In the north Industrial Zone colluvium consists of clay with debris basalt of 0-15m thick overlying Cover Basalt, according to the data from well R, Mifal “Rada”. All available wells located within the study area are given in Table 1. Tectonics features Picard (1943) suggested that "the subsidence of the Dead Sea Rift (DSR) in the Jordan Valley area and the uplift of the surrounding mountains occurred in the Neogene and that these differential movements continue, with less intensity, through the Quaternary". This tectonic history framework is still accepted today. The Bet Shean Valley is a triangular depression (Belitzky, 2002) bounded to the west by the Mount Gilboa block and to the east by the western marginal fault of DSR. According to Rozenbaum et al., (2004), two main groups of morphological linear elements were observed in the area (brown line in Fig. 3). The first one is a set of N-S oriented morphological escarpments, which separates the Bet Shean Valley.
(* - information on the wells was kindly provided by the company “Eng. David David Ltd.”) These escarpments represent the trace of the western marginal fault of the DSR. The second one is a set of NW-SE oriented lineaments that cross the Bet Shean and Jordan valleys assumed to represent deformation related to another fault system. A paleoseismic study of Bet Shean Valley (Zilberman, 2004) carried out along a segment of the marginal fault of the DSR north of Tel Rehov provided evidence of continuous tectonic activity since the latest Pleistocene. The total vertical displacement along the marginal fault in the last 20,000-30,000 years is estimated by Zilberman (2004) as 40-50 m. | ||||||||||||||||||||||||||||||||||||||||||||||||
|
4.2. AFULA
Geological analysis of the Afula area was based on the researches of Dicker (1969), Shaliv (1991) as well as the geological map to a scale of 1: 50000 by Dicker (1969) shown in Figure 5 . The study area was divided into three structural zones: The Yizreel basin of Neogen age including the Lower Afula and the settlements of Merhaviya and Sulam. This zone is a flat plain overlain by the Quaternary deposits of 70 meters thick maximum. In the Merhaviya kibbutz crop out clay with conglomerate of the Bira Fm. (Pliocene). Stratigraphy and lithology Sediments of Jurassic-Cretaceous period Jurassic sediments penetrated in Sarid-1well at a depth of 1640m (see Figure 6 ) are represented by continuous sequence of limestone, dolomite and minor shale (Politi, 1983). Figure 6. Lithostratigraphyc section for Gan Tapukhim, Gidon-5 and Sarid-1 wells The Cretaceous sediments, consisting mainly of dolomite with limestone, some marls, is also penetrated in Sarid-1 well at a depth of 700m and crop out in the SE part of the map in the Khirbet Qara tilted blocks. At this place is also exposed the Chalk complex of the Mt Scopus Gr. (Senonian age) of 120 m thick. Eocene Limestone Complex This complex is exposed at the Givat Hamore Mount block and discordantly beds on the Chalk complex of the Mt. Scopus Gr. The thickness of Eocene sediments in the western part of Givat Hamore is approximately 300m (Dicker, 1969). This complex consists of Limestone with chalk and flint of the Timrat Fm. (Early Eocene) of 210m thick (unit E1 in the Geological map) and fine-grained limestone of the Bar Kokhba Fm. (Middle Eocene) of 95m thick maximum (unit E2). Miocene igneous intrusive bodies are exposed at Givat Hamore. The petrography of these rocks is micromeltegeite and alkaline dolerites (Oppenheim, 1962). The country rock belongs mostly to the Eocene Limestone complex. The intrusive bodies are dykes, stocks and single volcanic vent (unit "G"). Lower Basalt (unit ßL) exposes along Balfouriyya – Afula Illit basalt ridge in the western direction as a continuation of the Givat Hamore block. The rock is olivine basalt. The basalt is intensely calcite jointed. Contact basalt-limestone on the western slopes of Givat Hamore is often altered to a soft, calcareous material, on which a secondary nari crust develops (Coordinates: 230800/725870). In Afula Illit (Coordinates: 230100/725100) basalt sometimes is weathered with sporadic patches of clay-alteration products of basalt. The basalt flowed from Givat Hamore, where it is discordantly resting on the erosion surface of the Eocene Limestone complex. The two basalt bodies located in Givat Hamore (Coordinates: 233370/725100 and 234500/725870) asserted that their provenance was from a central eruption centre in the Givat Hamore. In wells drilled in the Yizreel basin, Lower Basalt was found at elevations that comply with the general southwestern plunge observed on the surface. Within the study area following wells penetrated the Lower Basalt: Merhaviya well at a depth of 44m; Merhaviya-1 well at a depth of 84m; Afula Gan Tapukhim well at 276m and Gidon-5 well at 468m (see Figure 6). Dead Sea Group. Sediments of the Dead Sea Group overly the Lower Basalt. Their thickness increases from Givat Hamore to the southwest reaching 500 m (well Sarid-1). On the basis of the borehole data the following six lithology-stratigraphic sequences can be selected, from bottom to top: 1.”Clay series” deposited unconformably on the Lower Basalt and consists of the calcareous brown soft clay. Its maximum thickness penetrated in the well Gidon-5 is 253m and minimum thickness in the well Merhaviya 1 is 19m. 2. Bira Fm. deposited conformably on the Clay series and consists of the calcareous clay to marl white to grey, interbedded chalk and silts. Its maximum thickness is 96 m (Afula Gan Tapukhim well) and it is outcropping in northeastern part of the area (kibbutz Merhaviya). 3. Gesher Fm. Consists of a series of freshwater limnic, mainly oolitic limestones unconformably overlying the Bira Fm. Its facies change gradually into calcareous sandstone, green marl, chalky limestone, occasional hard limestone beds and oncolitic gray limestone (Shaliv, 1991). Thickness of the formation according to the borehole data ranges between 20 and 70m. 4. Fragments of Cover Basalt layer (Pliocene age) is well known from boreholes in Yizreel Valley, where it reaches a thickness of 16m maximum (Gidon-5 well). Its basalt flowing from Gilboa area towards the NW exists in the Yizreel Valley and unconformable overlay Gesher Fm. According to the well data thickness of layer basalt unsteady, sometimes exist debris of basalt (wells Afula Alef, Afula Mifalei Sukar). The basalt is mostly hard, barely weathered, sometimes interbedding with tuff (well Afula Gan Tapukhim). 5. Conglomerate (Pleistocene age) has thickness varying from 70m to few meters with average of 20 meters. This conglomerate may be compared with the Wadi Malih Conglomerate. It is resting on the Cover Basalt and consists of pebbles limestone ranging from the L. Cretaceous to the Eocene and basalt, the matrix is silt. Nari- covered conglomerates and slope-breccia (unit cm) are found at the northern and southern margins of the Givat Hamore block, resting on basalt and abuts against a fault scarp in the Eocene Limestone complex. 6.Silty clay with conglomerate (Quaternary age). It is overlain by conglomerate in Yizreel basin, has thickness from 75m to few meters (average 25m). Northern Afula Illit recent sediments overlying the Lower Basalt have thicknesses up to ten meters. |
| 5. THE PLANNING OF MICROTREMOR MEASUREMENTS AND DATA ACQUISITION | ||||||||||||||||||||||||||||||||||||||||||||||||||
|
Microtremor measurements were carried out during the period from May to September 2005 in the town of Bet Shean (W-245850; E-249510; S-709840 and N-713900) and from September to December, 2005, in the town of Afula including Merhavia, Balforyya and Sulam settlements (W-223450; E-235400; S-720080 and N-727455). The work areas are approximately 20 km2 and 39 km2 for Bet Shean and Afula, respectively. The distributions of measurement points over Bet Shean and Afula are shown in Figures 3 and 5. In the beginning we designed a large spacing between measurement points (500 m grid) but through lateral variation of the results, density the grid point spacing, down to 250 m and for specific sites spacing was 150 m. An important issue that was raised before and during the investigations is the question of how dense should the grid of measured points be? In retrospective, after many projects in order to estimate site response functions in different towns of Israel, we may state that we gained reliability to the obtained results only because we had a dense grid of measured sites. Again, it has to do with the application of the Nakamura technique. Furthermore, in most of Israel there is very limited availability of densely distributed geotechnical information such as S-wave velocities and densities of the materials, especially at depth. We could compensate for the need of a dense grid of measured points of microtremor by drilling new borehole, conduct many geophysical surveys and monitor strong enough earthquakes at points across the area. These alternatives are by far more expensive, time consuming and may not always provide the necessary information.
Reliability and applicability of fundamental frequency and its amplitude obtained from Nakamura's technique may be influenced from different factors during microtremor records. So far as this approaches to microzonation usually is used in urban areas such factors as anthropic noise, underground piping and construction, soil-structure interaction may seriously to change results. Therefore, during data collection we not only visual checking of quality of time series of microtremor at each site, but twice (in the beginning and at the end set of record) computed Fourier spectra and horizontal-to-vertical spectral ratios for two-three time windows. Ground motion (velocity time history) was recorded using the multi-channel digital seismic data acquisition system designed for site response field investigations (see Shapira and Avirav, 1995). The system includes: a multi-channel amplifier with band pass filters 0.2-25 Hz, GPS (for timing) and a laptop computer with analog-to-digital (A/D) conversion card. The seismometers (L4C) used were sensitive velocity transducers with a natural frequency of 1.0 Hz and damping at 70% of critical. The microtremor motions were digitized at the ratio of 100 samples per second by a 16-bit A/D converter. The duration of each microtremor recording also is very important parameters. For many authors (Rovelli et al., 1991; Mucciarelli and Monachesi, 1998; Lebrun et al., 2004) duration of each noise recording was 10-20 min. Our experiment saw that for high stability and high fidelity of H/V ratio estimation preferable to perform at 50-70 min observation at each point. Therefore, at each site, the microtremor was recorded continuously for 60 minutes, creating data files of 3 minutes each of microtremor data. In Figure 7 and Figure 8 we present examples of the locations of the seismic stations during the site investigation in the towns of Bet Shean and Afula. Prior to performing measurements we checked and determined the transfer function of the instrumentations in order to facilitate transformation of the record signals into true particle velocity system. The individual seismometer constant (free-frequency, damping and motor constant) were determined from sinus and step calibration signal. The instrument characteristics of the stations are given in Table 2.
In addition, all seismometers were placed at the same locations and in the same orientation to record the same waves ( Figure 9). These measurements allow assessing identity different channels of the entire monitoring system i. e. transducer, amplifier, filter, and analog-to-digital conversion. Figure 10 a,b present, as example, the seismograms (volts) and corresponding Fourier spectra of microtremor of 6 horizontal and 6 vertical seismometers. We can see that traces and its Fourier amplitude spectra show very good identity. The procedure "instrument response removal" used in seismology de-convolves output of seismometer channel to estimate real ground motion. Figures 11a, b shows horizontal components of ground velocity of microtremor (micron/sec) and Fourier velocity spectra after remove responses instruments. From visual inspection we can see that both time series and spectra are not identical for different channels and there are essential distinctions in identity of channels in frequency range 0.2-0.8 Hz. (Figure 11b ). Therefore, if the fundamental frequency of site effect is less than 1.0 Hz we do not recommend "remove instrument" in data processing. Figures 12 and Figure 13 demonstrate influence of "remove instrument" on the vertical seismometers channels with analogous conclusions. However, influence of the "remove instrument" procedure to the vertical seismometers channels is somewhat less. |
| 6. DATA ANALYSIS |
|
The reliability and applicability range of the Nakamura's technique is strongly depended on the different stages of data processing and requires special research, knowledge, experience and intuition. In the spectral analysis of microtremor data we have to answers several questions, for example: To study the spectral character of the microtremor within the Bet Shean and Afula towns, we computed spectra and spectral ratios using two different time windows, consisting of 30 sec records for sites with resonance frequencies above 1 Hz and with resonance frequencies of 60 sec records for sites with resonance frequencies less than 1 Hz. The selected time windows were Fourier transformed, using cosine-tapering (1 sec at each end) before transformation and then smoothed with a triangular moving Hanning window. The H/V spectral ratio was obtained by dividing the individual spectrum of each of the horizontal components [SNS(f) and SEW(f)] by the spectrum of the vertical component [SV(f)]. To obtain systematic and reliable results from the spectra of microtremor, we used several time windows (60-70) that yielded a number of spectral ratios that, in turn, were averaged. The horizontal-to-vertical spectral ratio AH/V(f) is obtained by dividing the individual spectrum of each of the horizontal components SNS(f) and SEW(f) by the spectrum of the vertical component SV(f). If the shapes of SNS/V and SEW/V are similar then the average of the two horizontal-to-vertical ratios is defined. A(f)=1/2n[Σi=1(Sns(f)i / Sv(f)i)+Σi=1(Sew(f)i / Sv(f)i)] (3) We have consistently observed that averaging the spectral ratio arithmetically or geometrically does not significantly change the results. It is worth noting the importance of this averaging procedure, the main problem related to the selection of time windows. Routine analysis and processing of data were carried out using the software SEISPECT developed in the Seismology Division of GII (Perelman and Zaskavsky, 2001). SEISPECT is a MATLAB application for spectral analysis and processing of ground motion, including seismograms recorded by short-period and broad-band seismic stations, as well as strong motion accelerometers. The main modules realized in the program are visualizing and editing of the input data; selecting time window and computing FFT and H/V spectral ratios; saving and displaying results. SEISPECT was distributed and successfully used in the countries-participants of the RELEMR program. |
| 7. DISTRIBUTION OF THE FUNDAMENTAL FREQUENCY AND ITS ASSOCIATED H/V SPECTRAL RATIO LEVEL |
|
The fundamental resonance frequencies and maximum values of H/V spectral ratios maps were constructed to integrate all results from the microtremor measurements and puts constraints on the 1-D subsurface model to be developed using geological and geophysical information. Also assuming that the increased intensity of the damage during earthquakes is, to a great extent, correlated with resonance effects, mapping the predominant frequency and maximum relative amplification of ground motion become key elements for seismic hazard scenarios (Shapira et al., 2001). 7.1. BET SHEAN The maps for the town of Bet Shean shown in Figures 14 a,b were constructed on the basis of 210 H/V spectral ratios from microtremor. One glance at these maps tells that the prominent feature of the both maps is sub-meridian strike of isolines that coincide with young tectonic displacements in the Bet Shean area. As seen from the fundamental frequency map presented in Figure 14a, resonance frequency in the town of Bet Shean varies within the wide range 0.9-14 Hz. The highest frequency values are attained in the areas surrounding the outcrops of basalts and near the north edge of the town. Moving from the north to the south and southeastern directions we observe general decrease of the resonance frequency that correlates with the depth of the basalt occurrence. Two zones, where we did not detect site effects, are distinguished in the map. Since the first zone is located at the exposed basalt, it was not surprising that H/V ratios in the first zone contain no evidence of ground motion amplification. The lack of resonance frequency at points distributed in adjacent the outcrop areas may be explained by too thin sediments overlying reflector. The absence of site effects in the second, southeastern zone, attached to the western marginal fault system of the DSR may be connected to the ruptured zone along the faults. Distribution of the maximum H/V amplitude within Bet Shean is depicted in Figure 14b. Reflecting the variation of the impedance contrast between the bedrock and overlying sediments, the highest amplification values (up to factor 6) are attained at sites in the north and eastern edge of the study area where alluvial deposits are lying directly over the Cover basalt Pliocene age. Moderate site response (range from 2.5 to 4) is observed in areas where the lithological section consists of alluvium of various thickness and travertine overlying basalt. With the exception of the areas with no site effects, around the N-S faults of the DSR, whose lithological section is represented by the relatively high velocity travertine overlying basalt, yield low amplifications (less than factor 2.5). Within this field of low amplifications there we observe an islet of higher values connecting probably with variation of S-velocity for the upper layer. 7.2. AFULA Typical examples of H/V ratios from microtremor obtained in the town of Afula are shown in Figure 15. In these examples two clear resonance peaks appear at different frequencies. Therefore, to characterize distribution of the site effect parameters in Afula we considered frequency and amplitude for two peaks. Map of the resonance frequency and amplitude of the first H/V ratio peak H/V spectral ratio exhibits the first resonance peak at frequencies 0.3 -12 Hz with amplitude changing from 2 up to 8 (see Figures 16 and Figure17). According to the geological data, this peak is associated with the Lower Basalt of Miocene age and its morphology is reflected in the measurement results. The general trend from 0.3 up to 3 Hz toward the east and northeast is observed. In the Lower Afula are distinguished four areas divided by deep fault systems. The northwestern area is the deepest and is characterized by the frequencies of 0.3-0.6 Hz. According to the Gidon-5 well data, the Lower Basalt is found at a depth of 476 meters. In the northeastern area the resonance frequency is distributed in the range 0.6-0.8 Hz and disappears on the contact with the Lower Basalt Ridge that allows suggesting continuation of the southern marginal fault system to the west, exposed only in the Giv'at Hamore (Dicker, 1969). This area is limited in the south by fault, which is identified by the sharp shift of the resonance frequency from 0.7 Hz up to 1 Hz. To the southeast from this fault the increase of the resonance frequency up to 3 Hz is detected, confirmed by the data from Kfar Merhavia well, according to which the lower basalt is found at a depth of 40 meters. In the south area of the Lower Afula we obtained the resonance frequencies from 0.4 Hz up to 0.8 Hz. In the Lower Afula the impedance contrast determining amplitude of the main H/V peak is formed by Pliocene-Pleistocene sediments overlying the lower basalt. Therefore, the amplitude values that we observed are 2-4. Only in the southeastern part of the Lower Afula the amplitudes reach value of 5. There, according to the Kfar Merhavia well data, the Pleistocene conglomerates overly directly the Lower Basalt. Flat H/V spectral ratios with no resonance frequency are attained at sites located at the Balfouriyya-Afula Illit ridge (outcrop of the Lower Basalt) and Giv'at Hamore block (exposure of the Eocene deposits). Within the limits of area with no site effect, some anomalies characterized by the frequencies 4-12 Hz and amplitudes of 3-9 are detected that is probably connected with weathered basalt, alluvium and marginal conglomerates of Quaternary age covering the Lower Basalt. Resonance frequency and amplitude level of the second H/V ratio peak The resonance frequency and its amplitude level for the second H/V ratio peaks are depicted in Figures 18 and Figure 19. This peak is related to the shallower reflector represented in the greater part of the study area by rocks of Gesher Fm. together with the Cover Basalt (Pliocene age) and only in the southeastern part of the Afula depression, where the Gesher Fm. and Cover Basalt are wedging out, the Quaternary conglomerate is reflector. Like the first resonance frequency the changes in the second frequency reflect the relief of the shallow reflector. The highest resonance frequencies as well as the highest amplitude value are revealed in the eastern part of the Afula depression. It is explained by the small thickness of alluvial sediments (silty clay) over the uplifted Cover basalt. The area without site effects is extended in comparison with that for the first frequency at the expense of sites in the southeastern part of the Afula depression, where alluvium layer is too small to produce the resonance frequency. |
| 8. VALIDATION OF H/V RATIO USING SEISMIC REFRACTION MEASUREMENTS | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Many different techniques exist for mapping site effects based on various kinds of experimental data (strong or week earthquake recordings, explosion recording, and microtremor measurements), most frequent empirical method (e.g. sediment-to-bedrock spectral ratio, horizontal-to-vertical Fourier spectrum of microtremor – Nakamura's method or using this method for S-wave from earthquake ground motion - receiver function) or many different empirical correlation between some geological and geotechnical information and some ground motion parameters. Their cost varies a lot from one to another. The Nakamura's method is much cheaper than other empirical techniques and, in addition, devoid of limitations hampering the site investigations in urban environments. But it is crucial to use low cost tools for site response investigation, from an economical as well as qualitative point of view. We must be fully confident that this low cost method is reliable. We should emphasize that during more than ten years of intensive site effect investigations in the different areas of Israel we have met quite a few cases when the Nakamura technique failed to yield conclusive results (Zaslavsky et al., 2001 (in Hebrew), 2004b, 2005). The key problem is that this method has been developed empirically and now we have some controversial theoretical investigations performed to clarify its underground physics (Lachet and Bard, 1994; Nakamura, 2000; Zhao at al., 2000; Fäh at al., 2001 and others). Thus, in order to evaluate the ability on Nakamura's method to predict resonance frequency and amplification level, we must have an independent estimate of site effects. The best validity of Nakamura's method is comparison with site response inferred directly from horizontal-to-horizontal spectral ratio for S-wave of weak or strong motion records obtained on the soft-soil site with respect to nearby hard-rock reference site. Unfortunately, to record earthquake with sufficient signal to noise ratio we sometimes have to wait very long time (some years). Instead, in the town of Bet Shean and Afula we compared assessment based on H/V spectral ratios from microtremor made at locations where a geophysical survey (seismic refraction) was conducted with theoretical one-dimensional transfer functions for a single or multi layers over a half-space subject to a vertically incident shear wave. Absolute value of the function, assuming a vertically incident shear wave, is computed using SHAKE program (Shnabel, 1972). The sediments and bedrock density and damping values needed for the 1-D model calculation were assigned in accordance with determined for similar rocks from the different places of Israel. 8.1. BET SHEAN Data collected from the seismic refraction lines carried out in the area investigated provided us with S-wave velocities in the bedrock and shallow sediments. Locations of the refraction lines are shown in Figure 3. Velocity depth section along the line RL-1 is presented in Figure 20. Based on the analysis of the seismic survey results, geological and borehole information in the study area we could correlate the first layer with S-velocity 280 m/sec to alluvial sediments; the second layer corresponds to travertine, whose S-wave velocity is 1050 m/sec. The third layer (Vs=2030 m/sec) is associated with the Cover Basalt being the most likely reflector. Four microtremor measurements at points 162, 201, 81 and 202 were carried out along RL-1 line. H/V spectral ratios obtained at these points are presented in Figure 21. Both fundamental frequency and its corresponding amplitude for all four points are similar and that is in accordance with the general picture of velocity-depth section of RL-1 line, where we observe only slight alterations in relief of the reflector. The typical feature of H/V spectral ratios curves is presence of the second peak, which broadens very considerably the frequency range of motion amplification. The absolute values of one-dimensional theoretical transfer functions, named further for simplicity the theoretical transfer functions, were computed with direct use of thicknesses and S-velocities from the seismic survey and compared with H/V ratios in Figure 21. One can see a fair agreement between them. Two frequencies f0 and f1 are indicated in the figure. The geophysical data and theoretical models for points along RL-1 line are given in Table 3.
Refraction line RL-2 is located in the northern part of Bet Shean in the vicinity of the outcropped Cover Basalt. In the velocity depth section along line RL-2 shown in Figure 22 three layers are differentiated: low velocity upper layer of 10-15 m thick (290 m/sec) and two high velocity layers (1590 m/sec and 2050 m/sec) both correlating with the Cover basalt. Figure 23 shows the theoretical transfer functions superimposed with the H/V spectral ratios for points 199 and 200. These functions are calculated for model comprising two layers over the bedrock with Vs=2100 m/sec, as determined in the seismic survey. As is evident from this figure, direct use of S-velocities and thicknesses from refraction survey for calculating transfer function yields a good fit. Theoretical functions for both sites calculated assuming one-layer structure (in this case the upper basalt layer with Vs=1650 m/sec is the reflector) are shown in Figure 23 as well. As seen, difference between two theoretical functions is negligible. We suppose that this layer provides the resonance effect and the deeper basalt layer differentiated by geophysical data does not really influence the fundamental frequency and its amplitude. Therefore, the H/V spectral ratios have only one peak. In such a case, thickness of sediments overlying reflector and, consequently, drilling depth can be almost four times less than that predicted by geophysical data. In spite of aforesaid, we give in Table 4 the soil column model calculated for deep reflector in order to keep the general geological structure and velocity model in this area.
8.2 AFULA Transfer functions calculated on the basis of refraction and boreholes data (see Figure 5 for locations), compiled with information on velocities from the previous investigations, were used to validate H/V ratios at corresponding locations and determine velocity structure. Lithological sections of Gidon 5, Afula Gan Tapukhim and Merhavia wells located at different parts of the study area and penetrating the Lower basalt at different depths (Figure 6), characterize generally the lithological structure of the Afula depression and provide us information on layer thicknesses. Velocity models of the upper layers of the section we derived from the results of the refraction survey along profile Af-1 carried out close to Afula Gan Tapukchim well. The H/V ratio and analytical transfer function for point 115 located at Gan Tapukhim well are shown in Figure 24. The velocity depth section along profile Af-1 is shown in Figure 25. All available geotechnical data for point 115 are concentrated in the left part of Table 5. Since this refraction survey provides Vs for 30 upper meters only, rest of the required information on Vs (down to a depth of 275 meters) was supplemented from other sources. Vs of the deep reflector, the Lower Basalt, we assumed equal to 2200 m/sec by extrapolation of the available velocities down to a depth of 300-500 meters. Results of refraction survey along profile Af-2 near the Afula Hospital site (Ezersky, 2003), presented together with the H/V ratio and analytical function for point 295 in Figures 26a,b, identify alluvium layer (Vs=320 m/sec) overlying the weathered Lower Basalt layer (Vs=600 m/sec) and underlain by the hard Lower Basalt with Vs=1750 m/sec at depth of 13-25 m. These velocities are used for modeling at Balfouriyya-Afula Illit area. We utilized S-velocities for marl and clay layers derived from the investigations in the Coastal Plain and Hashefela regions (Zaslavsky et al, 2003b). Analytical transfer function, showing the pretty good match with the H/V spectral ratio at point 25 located at Gidon 5 well (see Figure 28a), was calculated by directly use of thicknesses from the borehole data and S-velocities described above. Soil column model for this point is given in Table 6. S-velocity for the conglomerate layer was inferred through modeling Merhaviya well penetrating the Lower Basalt at a depth of 44 meters, Lithological section of this well includes 30-meters conglomerate layer.. For the upper alluvium layer we used Vs=160 m/sec, presence of which in the study area is indicated by the refraction data along line Af-3 (see Figure 27). The H/V ratio and analytical function for point 209 (Merhaviya well) is shown in Figure 28b Figure 28b.
H/V spectral ratios validated at borehole and refraction survey locations were utilized, by velocities extrapolation, to study other sites without geophysical and borehole information. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 9. ESTIMATION OF SUBSURFACE STRUCTURE USING H/V SPECTRAL RATIOS FROM MICROTREMOR | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
The program based on the stochastic optimization algorithm (Storn, 1995) was applied in order to obtain a better fit of theoretical transfer function to spectral ratio, considering the dominant frequency, its level and the shape of the H/V curve. Within the chosen frequency interval [ ω1, ω2 ] we look for thickness ( Hi2 ) and S-velocity ( vi ) that minimize the misfit function , F=Σk=1(g(ωk)-f(ωk))2 where ωk are points from the frequency interval [ ω1, ω2 ] , g(ω) is 1-D theoretical transform function calculated by SHAKE program; and f(ω) is H / V spectral ratio. Velocity and thickness are limited: and V1i < Vi < V2i,i=1,M+1 and H1i < Hi < H2i,i=1,M where M is number of layers in 1-D model. Since we apply the stochastic optimization method practically not depending on number of parameters in question, an exhaustive search of the model is computationally quite reasonable. 9.1. BET SHEAN Reconstruction of the subsurface structure in the town of Bet Shean is demonstrated on two cross sections A-A' and B-B' directed east west and north south correspondingly (locations in Figures 3). S-velocity characteristics for litho-stratigraphycal units in the study are summarized in Table 7. These velocities are used to find the thicknesses of layers providing the best fit of the calculated transfer functions and H/V ratios. Ranges of S-wave velocities for litho-stratigraphycal units represented in the study area taken from refraction survey
9.1.1. Profile A-A' A simplified sketch of the geological cross section beneath the profile A-A-A' constructed using microtremor measurements is depicted in Figure 29. Microtremor measurement at points 162, 202, 81 and 201 indicated on the cross section by the blue triangles owing to their location at refraction line Rl-1, were used for calibration of H/V spectral ratios (see Figure 21 and Table 3). H/V spectral ratios for these points are characterized by two close resonance frequencies about 4 Hz and 6-7 Hz and our soil column model explains this by influence of two reflectors: travertine and basalt. Figure 29. Characteristic 2-D cross section along profile A -A' (Bet Shean) This feature is distinctive also for H/V ratios obtained at points 79, 82, 80 situated to the west from RL-1 line (see Figure 30). Therefore, for calculation of the transfer functions at these points we applied analogous velocity models. One can see in Figure 30 that the theoretical transfer functions match fairly well to the corresponding H/V spectral ratios. Figure 31 presents H/V spectral ratios and theoretical transfer functions for points 1 and 2 located at the eastern edge of the profile. Comparison with previous point shows that while shape and H/V amplitudes are remained practically the same, sharp decrease of the fundamental frequency from 4.5 Hz down to 2.1 Hz at point 2 is observed. Such a variation of the fundamental frequency may be the most likely explained by the presence of fault in the basalts. After fitting the theoretical transfer function to H/V ratio we estimated vertical shift as about 60 meters. Point 1 next to point 2 exhibits the similar fundamental frequency and, therefore, reflector depth was not changed. H/V ratios for points 4, 7, 27 and 9, shown in Figure 32, exhibit one clear resonance peak with amplitude of about factor 2-2.5 and they are completely different from all points observed earlier. For all numerated points and also for point 6 located hereabout, H/V ratio for which is shown separately in Figure 33, we failed in attempts to fit the theoretical transfer function to H/V spectral ratio keeping S-wave velocity model accepted for the western part of the profile A-A'. The S-velocity model needed to calculate theoretical transfer functions was obtained from seismic refraction survey and downhole measurements on BH-1 well (Ezersky and Shtivelman, 1999). The velocity models are in Table 8. BH-1 well is located 10 meters from the refraction line was drilled to a depth of 30 meters, penetrated alluvium layer, upper and lower travertine layers and did not reach the basalt. S-velocity for the basalt was taken from another refraction survey along RL-1 line. Once S-wave velocities are determined, it become possible to adjust thickness of the lower travertine layer to get the best fit between analytical transfer function and H/V ratio at point 6 located at refraction profile (see Figure 33). The S-wave velocity model from refraction survey and downhole measurements are given in Table 6 together with the optimal soil column model. The theoretical transfer function for point 6 is shown together with H/V spectral ratio. The theoretical transfer functions for points 4, 7 and 27 shown by the black dashed lines in Figure 32, are calculated using velocity model for point 6. Variations of velocity for the upper travertine layers obtained in the fitting process (see Table 9) exceed significantly the accuracy of frequency determination and clearly indicate the tendency Vs to be increased in the east direction. Such tendency is reflected in the model distribution in the study area. Velocities for Lower travertine layer turned out practically the same. Different thickness of the travertine layers for points 6 and 7 was interpreted as a fault. Figure 34 demonstrates individual H/V ratios at point 164 with no resonance frequency. It should be noted that we observed the whole area with analogous pictures of the. This area is contoured in the frequency map (Figure 14) and we mark it on our cross section as well.
The H/V spectral ratios for points 10, 11 107, 108 and 191 are shown in Figure 35. All this ratios are characterized by very low amplitude level for the reason that impedance contrast in this geological situation is formed by thick travertine layer outcropped in this part of the study area and basalt. We observe an increase of resonance frequency at point 108 in comparison with point 107 (1.8 Hz vs. 1.0 Hz), which we connected with the fault. The modeling procedure allowed estimating the relative vertical displacement roughly as 50-70 meters; however, absolute values of reflector depth may be hardly determined due to low impedance contrast between travertine and basalt. Therefore, in this part of the profile the top of basalt is shown by dashed line, underlining uncertainty of the depth estimation. 9.1.2. Profile B-B' The geological cross section along profile B-B' passed through the town of Bet Shean in the south-north direction is presented in Figure 36. The spectral ratios obtained at points distributed along the profile demonstrate very broad range of resonance frequency. There are sites at basalt outcropped where no resonance frequency was detected; and there are sites where resonance frequency reaches 13 Hz. H/V amplitude level varies in the range from factor 2 up to 7. Figure 37 shows H/V spectral ratios for points 120, 112 and 109 located in the southern part of the profile. They are characterized by main clear peak at frequency near 1.7 Hz and minor peak at 4 Hz. Bearing in mind distribution of the fundamental frequency and its corresponding H/V level within the study area (maps in Figures 14 and 15), we applied for calculation of the theoretical transfer functions the velocity model used for points 6, 4, 7, 8 and 9 at the east-west profile A-A' (for details see Figures 32, 33 and Table 8). Modeling showed that difference between soil column models for points on profiles A-A-A' and B-B' is redistribution thicknesses of layers. In particular, the thicknesses in the soil column model characterizing the southern part of the B-B' profile (points 120, 119, 112 and 109) are distributed in the following way: alluvium layer is of 10-15 m thick, upper travertine layer is of about 50-60 m thick and lower travertine of 60-70 m thick overlay basement represented by the Cover Basalt. The theoretical transfer functions are shown together with spectral ratios at Figure 37. In the middle part, profile B-B' intersects with east-west profile A-A' near points 3, 162, 82 and 79. These points were already analyzed (see Figures 21 and 30). Surrounding points in this section (91, 87, 83 and 76) are depicted in Figure 38 and show shape of the curves very similar to mentioned above with a typical broad resonance frequency range formed by influence of two reflectors. At a border between southern and central sections (points 109 in Figure 35 and 91 in Figure 38) the fault is detected. It is fixed, first of all, by sharp shift in the fundamental frequency (from 1.5 Hz up to 3. 5), that corresponds to the vertical displacement of about 100 meters. What is no less important that the H/V amplitude level and shape of the curves changed and show the same broad resonance frequency range typical for two reflectors. We have already seen that such variations are a sign of model change. The soil column model for the middle part of the profile comprises alluvium and lower travertine layers overlying the cover basalt. As we are moving to the north from point 91 to point 76 the gradual increase from 3.5 Hz up to 7 Hz is detected. Correspondingly, the total sediment thickness decreases from 70 meters down to 40 meters. This reduce of total thickness is distributed evenly between all three layers in agreement with H/V amplitude level for points in this section, which is about factor 4 for all points. The part of the cross section with points 161, 51 and 44 cuts through a sector where the basement (the Cover basalt) crops out. The H/V method presents flat spectral ratios for this zone (see Figure 39). The fault is mapped near this outcrop and the microtremor measurements yield total change of the subsurface model that is confirmed by the refraction survey data along line RL-2 located not exactly but close to our profile. As seen from Figure 40, the measurements at point 46, 176, and 177 situated in the northern part of profile B-B' yield the high amplitudes peak (up to factor 10) at extremely high frequencies (8-14 Hz). On the basis of the geological and geophysical observations we constructed soil column models consistent with our measurements. They are shown together with H/V ratios. The lithological section for this zone may be represented by a few meters of alluvium overlying basalt. 9.2. AFULA 9.2.1. Profile A-A-A' West-east and further northeast direction of profile A-A-A' was chosen to demonstrate representative points in the different geological conditions and suggest an interpretation of contact between the Balfouriyya-Afula Illit ridge and Afula depression. The ranges of frequency 0.35-10 Hz with amplitude from 2 up to 8 units for the fist resonance peak; and 1-8 Hz with amplitude from 2 up to 8 units for the second peak imply very significant variations of the subsurface structure expressed in both sediment thickness and velocity profile (see maps in Figures 16-19). Profile A-A' is depicted in Figure 41. Among a number of H/V ratio curves derived from the records along the selected profile we will show in Figure 42 only points which are distinguished by the uncommon features, such as sharp changes in frequency or amplitude of peaks at a short distance. Point 25 located directly at Gidon 5 well was described in details previously (see Figure 28 and Table 6). H/V ratios obtained at points from 25 up to 88 show the stable first (0.35 Hz with amplitude up to 3) and second peaks (1.1 Hz with amplitude of 4). The thinning of the conglomerate layer is compensated by small increase in velocity of the upper layer at the expense of some gravel within the alluvium (data from Afula-A well). At point 89 we observe increase of the second resonance frequency up to 2 Hz. Such change according to our model corresponds to decrease of the sediment thickness above upper reflector (the Cover Basalt and Gesher Fm.) from 70 meters down to 40 meters. 40-meters displacement of the deep reflector (the Lower basalt) is reflected in the increase in the first frequency from 0.35 Hz up to 0.4 Hz. Further gradually increase of the first resonance frequency up to 0.65 Hz at point 115 (Afula Gan Tapukchim well) matching to variation of the reflector depth from 480 m to 280 m, is broken only once by fault located between points 104 and 115, the vertical shift of which is about 50 meters. Our observation at points located to the northeast from point 115 yield very close first resonance frequencies, which indicates only slight slope of the Lower Basalt. The second resonance frequency increases gradually up to 9 Hz (the alluvium layer 3m thick). Point 295 exhibits H/V ratio with two peaks at 6 Hz with amplitude 3.5 and 14 Hz with amplitude 5. This point is situated at the Balfouriyya-Afula Illit ridge in the completely different geological conditions. Shift in the frequencies of the first peak for neighboring points 116 and 295 (0.6 Hz Vs. 6 Hz) corresponds to the 230-meter shift in the depths of the Lower Basalt and must be followed by fault. This fault is traced by the geological data in the southeast, at the contact of the Afula depression and Givat Hamore. The shallow reflector associated with the second peak is marginal conglomerate, which is identified in the velocity depth section along refraction profile Af-2 (see Figure 26). Beginning from point 15 and to the end of the profile the H/V ratios is characterized by high frequency and high amplitude single peak (see point 19 in Figure 42 as an example. The second H/V peak disappears due to small thickness of conglomerates. 9.2.2. Profile B-B’ Profile B-B’, reconstructed on the basis of the microtremor measurement, is depicted in Figure 43. Characteristic examples of the H/V ratios and corresponding analytical transfer functions for representative points along profile are shown in Figure 44. Common feature for all the points located at the southwestern edge of the profile B-B' (46, 48 and 3) as well as surrounding points is H/V ratios with the low amplitude second peak that implies relative high velocity layers above the shallow reflector represented by the Cover Basalt and Gesher Fm. This layer is possibly Pleistocene conglomerate. After the analytical transfer function was adjusted to the corresponding H/V spectral ratio we derived thickness of this layer of about 100 meters. The first resonance frequency changes between points 46 and 3 from 0.5 Hz up to 0.8 Hz. Depth of the deep reflector (the Lower Basalt) varies from 390 m up to 250 m correspondingly. Since in the southwestern part of Profile B-B' there is no borehole data on thicknesses of Marl-clay of the Bira Fm. and "clay series" and, mainly, owing to close S-velocities, we united these two layers into Pliocene marl-clay layer. H/V ratios of next some points along the profile show common feature significantly different from previous points, specifically the second peak has amplitude almost three times higher than at point 3, for example and higher than the their own first peak. This situation is already examined on profile A-A' and the model is similar to that for point 115 at Gan Tapukhim well (see Figure 24). Points 3 and 54 are divided by fault with vertical displacement of 70 meters. H/V spectral ratio with the second dominant peak is retained up to point 145(well Mifalei Sukar). We observed gradual changes in the depth of the Lower Basalt from 230 m down to 285 m. Considerable variations in thicknesses of the Cover Basalt & Gesher Fm. and united marl-clay layer on different sides of the fault, inferred from the models, may be connected with period of lifting and erosion, which preceded the forming of the Gesher Fm., as it is indicated by Shaliv, 1991. At point 140 we again came back to the model including the thick conglomerate layer owing to changing balance in the H/V amplitudes in favor of the first, main peak, like as in case of points 46-3. Fault between points 145 and 140 is traced by the geological data. The fault located below point 270 is mapped by the microtremor measurements only. It is interesting that H/V ratio for point 270 (see Figure 44) has more complicated shape of the second peak than the other ratios, which is probable connected with presence of fault. From point 271, located at the Bira Fm. outcropped and showing single H/V peak, up to points 180 and 198, located at exposure of the Timrat Fm., we observe general increase of the fundamental frequency with local shifts up and down (point 144). Points 309 and 209 reveal the second H/V peak related to alluvium layer a few meters thick overlying the conglomerate layer. The depth of the Lower Basalt at point 209 (Merhaviya well) is 45 meters. According to the borehole data the Lower Basalt layer is underlain by limestone and chalk of the Timrat Fm. Points 180 and 298 yield H/V ratios with no resonance frequency. 9.2.3. Profile C-C' Profile C-C', directed NW-SE is shown in Figure 45. Characteristic H/V spectral ratios for measuring points along this profile one can see in Figure 46. Points from 313 to 211 show identical two-peak shape of H/V ratios; the second peak is higher than the first one. Similar pictures we observed at both profiles A-A' and B-B' and according to the refraction survey and borehole data columnar section of this part of the profile C-C' from the top to the bottom consists of alluvium, the Cover Basalt and Gesher Formation as shallow reflector, "Clay series" and Lower Basalt as deep reflector. The frequency of the first peak equal to 0.65 Hz for all these points indicates practically constant sediment thickness above the Lower Basalt. Moreover, slight changes in the second frequency and its amplitude are connected with variations in the velocity of the upper layer, while its thickness varies in the limits of the ten meters. Sharp shift in the first frequency from 0.75 Hz for point 299 and 1.1 Hz for point 262 supposes fault with the vertical displacement of 80 meters. Comparing the H/V ratios for point 262 with next point 263 we see significant decrease in the second amplitude from 8 units down to 5 units. According to the geological data further along the profile is found exposure of the marl-clay of the Bira Fm., S-velocity of which is 750-800 m/sec vs. 1700 m/sec for the Cover Basalt and the Gesher Fm. This fact may explain decrease in the second amplitude. Point 288 yields further increase in the frequency of the first peak and low amplitude of the second peak and, at last, point 312 shows H/V ratio with no resonance frequency being located at the Bira Fm. with very small thickness not far from the outcropped Basalt. Figure 46. Characteristic H/V spectral ratios obtained at measuring points along profile C-C' | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 10. IDENTIFICATION OF FAULTS USING H/V SPECTRAL RATIO FROM MICROTREMOR |
|
Despite the issue of fault identification was already arisen in the previous sections, here we once again analyze and classify all available in Bet Shean and Afula cases of sharp changes in the H/V ratio parameters, which may be interpreted as vertical displacement in the basement. Two basic faults systems were mapped in the Bet Shean area by geological data: NW-SE oriented lineaments that cross the Bet Shean and Jordan valleys and the western faults of the DSR. There are a lot of investigations in which the authors by different way trace the faults (Shulman, 1962; Shaliv, 1991; Hazor, 2000; Gardosh and Bruner, 1998; Zilberman, 2004, etc.). In Figure 47 we suggest our interpretation of fault identification and location based on H/V spectral ratio measurements. Some of our results correlate well with the one or another geological interpretation, but in some cases our interpretation is different. It is known that complicated geological conditions in the vicinity of faults influence very significantly the original time domain recordings and, consequently, results of H/V curves from microtremor. Therefore, not every fault was mapped with Figure 47. Map showing different interpretations of faults location in the Bet Shean area equal confidence and reliability. In some cases only distribution of H/V frequency and amplitude level in the extensive areas allowed to make conclusion about faults presence. Below we demonstrate the examples illustrating the main criteria, on which we based in detecting faults. Type I. While the H/V fundamental frequencies for points located on the both sides of the fault are different, their amplitude and shape of curve are identical. In terms of subsurface models it means change of reflector depth and unaffected impedance contrast between sediments and reflector. For points 2 and 80 shown in Figure 48 the vertical displacement is more than 60 meters. It is important that we trace the fault until this feature is true. The theoretical transfer functions are shown in the same figure. Type II. For this type is typically that all three characteristics of H/V spectral ratio, i.e. frequency, amplitude and shape are different. The first example ( Figure 49) showing comparison of points 81 and 4 illustrates the situation which was already explained in the comments to the cross section A-A'. It was said that we interpreted variation in H/V characteristics as a change of velocity model. Soil column models for points 81 and 4 are represented by following sequences: alluvium-travertine over cover basalt and alluvium-upper travertine- lower travertine- over cover basalt respectively. Variation of the reflector depth is about 50 m. Next example in Figure 50a,b shows by pairs points located in the beginning and end of the fault in the southwest of the area. One can see that while the difference in the amplitude level for opposite points (109 vs. 91 and 85 vs. 87) is not great, H/V curves are completely dissimilar. This dissimilarity is revealed in the location of the first and second peaks of H/V curves and their balance. This example characterizes transition between soil column consisting of alluvium-upper travertine-lower travertine over cover basalt into alluvium-lower travertine over cover basalt. One more example of fault detected by the criteria of Type 2 is shown in Figure 51. Again, the combination of the frequency shift and different H/V ratio shapes was reason to change velocity model and suppose fault between points 50 and 74. This fault divides two blocks with different models. And an example of reverse transition is from alluvium-cover basalt (point 21) to alluvium-travertine-cover basalt (point 42) is shown in Figure 52. In Figure 53 one can see additional examples of H/V ratios for opposite points (204 vs. 13 and 72 vs. 36) in the different places of the study area, which indicate presence of faults between them. Type 3 is a pretty rare situation in the study area, in which one of two neighboring points is located on the outcrop of basalt and we observe flat H/V curve in contrast to the peak with amplitude of about 4 (point 10 vs. point 164 in Figure 54). The transition to the zone without site effects, which accompanies the western branch of the DSR (see Figures 13, 14), is reflected in the similar picture, as one can see in Figure 55 where point 99 is opposed to point 147. The same principles were assumed as a basis for fault identification and mapping in the Afula area. The geological map in Figure 56 depicts faults detected by microtremor measurements compared with those mapped by geological data. Summarizing all aforesaid about identification and tracing faults in the study area using H/V ratio method we could conclude as follows: |
| 11. PRELIMINARY SEISMIC ZONATION AND PREDICTION OF ACCELERATION RESPONSE SPECTRA FOR LINEAR AND NON-LINEAR BEHAVIOR OF SOIL SEDIMENTS |
|
Theoretical subsurface models constructed on the basis of H/V measurements together with available geological and geophysical information, in turn, we used for estimating the expected site effects during earthquakes. In the engineering practice, the a-seismic building design and assessments of the earthquake risk refer to the site-specific acceleration (or displacement) spectrum. The design acceleration spectrum is essentially a representation of the maximum acceleration amplitudes for a prescribed probability of occurrence developed on a set of one degree of freedom oscillators with a given damping ratio. Since seismic activity in areas such as Israel is low, local acceleration data from strong earthquakes is insufficient to estimate directly the design acceleration spectrum; therefore, in areas covered by soft sediments, we must resort to the use of synthetic data. For this purpose Shapira and van Eck (1993) developed the SEEH method (Stochastic Estimation of the Earthquake Hazard). In brief; SEEH produces a number of synthetic earthquake catalogues that represent the possible future seismic activity within 200 km of the investigated site. These catalogues adhere to the available information about the seismogenic zones in the area and their associated seismicity. The Monte Carlo statistics are used to generate different catalogues which reflect the uncertainty associated with the spatial and temporal parameters of the seismicity. For each of the earthquakes in a catalogue, SEEH implements the stochastic simulation method (see e.g. Boore, 2000) to generate synthetic S waves accelerogram for the surface of the bedrock which then propagate through the soil column of the site (Shnabel, 1972) to the surface. The synthetic free surface accelerogram is used to calculate the acceleration response spectrum for a predefined damping. Here again, the Monte Carlo statistics are used to select the values of the parameters used in the ground motion simulations. For example; we assume a unified distribution for locating the hypocenter within a defined seismogenic zone and within a 5-20 km depth, we assume that the estimated seismic moment of the event (and thus the energy at the source) are log-normally distributed around the expected value with an uncertainty factor of 3 and so forth. The parameters used are based on studies done in the area and reflect our current knowledge (and uncertainty) about seismic activity and the main parameters that control the SPECTRA of expected ground motions at a given site. The ensemble of these hundreds (sometimes thousands) of synthetic acceleration response spectra are statistically analyzed in order to assess the spectral amplitude level to be exceeded at least once in a certain exposure time (usually 50 years) and a certain probability (usually 10%) (for more details see Shapira and van Eck, 1993). Seismic zonation comprised two stages. At the first stage, H/V spectral ratios for 210 sites in Bet Shean and 100 sites in Afula were categorized into several characters of their shape, H/V frequency and amplitude values. In Afula, two H/V ratio peaks were considered. At the next stage, for 70 representatives of selected groups for Bet Shean and 60 for Afula the Uniform Hazard Acceleration Spectra were computed and again compared. In dependence on characteristics of the acceleration spectra and also spatial distribution of sites within the study area the selected groups were or united into greater zones or, vice versa, subdivided if obvious outsiders were revealed. 11.1. BET SHEAN The final version of zones division for Bet Shean is presented in Figure 57. The generalized theoretical transfer functions and spectral accelerations together with soil column models are given in Table 12. One can see that maximum of the spectral acceleration varies from 0.6 g up to 1.6 g. We also plotted on the same graphs in Table 12 acceleration spectra required by the current building code IS-413 for ground types S1, S2 or S3 corresponding to the model with the design horizontal Peak Ground Acceleration (PGA) value of 0.247. With the exception of Zone I, the computed spectral acceleration exceeds to the different degree the accelerations prescribed by IS-413 in the period range from 0.1 sec to 0.4 sec.The calculated theoretical model can than be easily used to modify synthetic seismogram computed for rock and predict also nonlinear site specific ground motion during large earthquakes at sites where the ground motions have not been recorded (Shapira and van Eck, 1993). For many years, the geotechnical engineering and seismology communities have had different approaches to significance of non-linear soil behavior during ground shaking. The central question of the discussion is when soil amplification is amplitude dependent. For example, Gutierrez and Singh (1992) described character of strong motion on soft sites and assert that they did not found clear evidence of nonlinear behavior even then when peak horizontal acceleration exceeding 0.3 g. In accordance with Su et al (1998) the difference between weak-and strong-motion site responses becomes significant at stations where peak acceleration was above 0.3 g. At lakebed sites in Mexico city, the ground motions were amplified as much as several tens times relative to hill zone during the 1985 Michoacan earthquake, however only at Station Central de Abastos (Reinoso and Ordas, 1999) little non-linear behavior where observed. The identification of nonlinearity in site response is challenging because classical spectral technique with reference station only partly relieves this problem. Perhaps, the reality of nonlinear soil response could be completely hypothetical a decade ago. The field observation of recent earthquakes by modern surface and downhole vertical array of seismometers indicated that soil non-linearity influenced on ground motion (Field et al., 1997; Trifunac and Todoravska, 1998; Pavlenko and Irikura, 2002; Frankel et al., 2002 and other) and its correct prediction have major im |