Resent destructive earthquakes have clearly shown that near-surface geological 
conditions play a mayor role in the level of ground shaking. Mapping predominant 
frequency of soil resonance and amplification permits identification of zones at risk
in seismically-prone areas. The motivation and objectives of this study were to 
divide the Hashefela area into geographical zones, each one characterized by a 
fundamental frequency and amplification factor. In final version of this report we 
will adjust the overall soil-column model, which facilitates forecasting ground 
motion from future earthquakes, for each zone.
    Owing to the relatively low seismicity of the region, we concentrated our 
efforts on estimating site response by implementing the horizontal-to-vertical 
spectral ratio (H/V) of ambient vibration. Ambient vibration surveys were carried 
out at more than 300 sites; 160 of which were located either at, or very close to 
wells. The selection of an appropriate ensemble of windows of ambient noise for H/V 
spectral ratios facilitates successful removal of time variant source effects. We 
obtained a good correlation between the empirical and analytical evaluations of the 
fundamental frequency as well as for the amplifications level for shallow and deep 
soils with a multi layer distribution. In the studied areas the ground motion 
amplification is a factor 2-8 over the frequency range of 0.3 to 6 Hz. The 
amplification obtained by H/V ratios may be explained not only by peak in the spectra
of the horizontal components but also by a trough in the spectra of the vertical 
components. We present maps that reflect the fundamental characteristics of site 
effects in the study area: dominant frequency and maximum relative amplifications. 
    The extensive database of ambient vibration measurements carried out at drilling 
sites facilitated estimation of deep shear-wave velocity profiles by trial-and-error 
fitting of the calculated to the empirical transfer function. 

     Many methods have been used to characterize site amplification. The best
approach is through direct observation of seismic ground motion, although 
such observations are limited to high seismicity areas and, also, by theirhigh cost. 
An alternative method, recording only one station, consists of dividing the spectrum
of the horizontal component by that of the vertical component of ambient vibrations 
(Nakamura, 1989, 2000).
     During the last decade, many sites in Israel have been investigated in an 
attempt to estimate the possible amplification of the seismic ground motion 
[Zaslavsky et al., 1995; Gitterman et al., 1996; Zaslavsky and Shapira,2000; 
Zaslavsky et al., 2000; Shapira et al., 2001; Zaslavsky et al., 2002b,d,e and 
Zaslavsky et al., 2003]. All these studies are based on analysing ambient vibration 
and weak motion measurements incorporated with geological and geophysical information
on the subsurface. We used various empirical methods to determine the site response 
functions including reference and non-reference techniques and referring to different
sources of excitation - earthquakes, explosions and ambient vibration. Appropriate 
ensembles of carefully selected windows of ambient vibration provide estimations of 
site response that are similar to those obtained from H/V spectral ratio of seismic 
events. However, there were cases where the Nakamura technique failed to yield 
conclusive results. This often happens when the ratio of the shear-wave velocity of 
the soil to the shear wave velocity of the underlying half space (bedrock) is higher
than 0.5-0.6 (amplification up to a factor of ~2) or when we are dealing with a 
complicated 3D structure of the underlying geology. Other examples are associated 
with poor excitation of the soil column due to weakness or remoteness of the 
microtremor sources. Thus, in many cases this poor behaviour of the Nakamura method 
could be foreseen and other methods should have been used. In other cases, where 
situation is better suited to the feasibility of the method, the results showed great
similarity to the results obtained by other techniques and, thus, provide useful 
feedback to improve the reliability of the experimental results. In rare cases, the 
Nakamura technique even provided estimations of higher harmonics of the resonating 
soil column. It is interesting to note that large differences in the site response to
seismic motion were observed over very short distances (several tens of meters). 
Site effects were also observed on rocky sites such as in the case where weathered 
and cracked granite bedrock showed amplifications by factor 4 in the frequency range 
of 6 to 7 Hz, well within the range of engineering interest for low-rise building 
[Zaslavsky et al., 2002a]. 
     In order to tackle the task of mapping site effects using ambient noise 
measurements with a variable grid density in different areas across Israel, a team of
new immigrant scientists with expertise in many different (seismology, geology, 
geophysics, data processing) was formed in the year 2000. So far the ambient noise 
survey has been carried out at more than 800 sites in the towns of Lod-Ramla (360 
sites), Qiryat Shmona (290 sites) and, at Coastal Plain along a ~10 km wide strip 
between Ashqelon and Haifa (190 sites). The results of the investigations have already
been published (see Zaslavsky et al., 2001; Zaslavsky et al., 2002c). As a 
continuation of the Coastal Plain mapping project, site effects were measured in the 
Hashefela and Hasharon areas (further named as Hashefela), which owing to its high 
population density, may be considered as a high seismic risk zone.
      The motivation and objectives of this study were to divide the Hashefela area 
into several geographical zones, each one characterized by fundamental frequency and 
amplification factor. In the final version of the report we shall adjust the overall 
soil-column model, which facilitates forecasting ground motion from future 
earthquakes, for each zone. The computation of the theoretical seismic response of 
the Hashefela area is essential for the estimation of specific seismic hazard and 
provides essential information for realistic earthquake damage scenarios. 
The applied methodologies take on added importance in regions with low seismicity but
with high seismic risk, as the case of the Hashefela region.

     The investigated area, the Hashefela region, extends from Qirayt-Gat 
in the south to Binaymina in the north for about 120 km and is almost 15-40 
km wide (see Fig.1). The geological data of the region were collected from 
Gvirtzman (1969, 1970, 1984), Fleischer et al. (1993), Fleisher and Gafson 
(2000), and the geological map of Israel to a scale of 1:200,000 
(Sneh et al., 1998).
     During the initial phase of compilation and determination of 
geothechnical data we gathered the information from 650 structural, water 
and oil wells from the database of the Geophysical Institute of Israel. The 
160 wells were chosen for ambient noise measurements during the work 
planning stage.
     QUATERNARY ROCKS
     The Quaternary sediments outcropping in the Hashefela area are 
represented by alluvium, sand dunes of the Kurkar Group (see Table 1).
    The Kurkar Group of Pleistocene age consists  of alternating  
marine and eolian calcareous sandstones named "kurkar", some reddish 
silty-clayey "hamra", silts, clays, loose sands, loam and conglomerates. 
Occasionally the Kurkar Group unconformably overlies the Judea, 
Mt. Scopus, Avedat and Saqiye Group rocks. Two different representative 
provinces the west and east can be recognized in the Kurkar Group. The 
thickness of the Kurkar Group decreases from about 200m near the 
shoreline to 0m in the eastern part of the Hashefela area. 
TERTIARY ROCKS
     The Saqiye Group overlies unconformably the Avedat and 
Mt.Scopus groups and consists of deep marine chalky marls of the Bet 
Guvrin Fm. and limestone of the Lakhish Fm. of Oligocene age, marls 
of the Ziqim and evaporates of the Mavqi’im formations of Middle- Late 
Miocene age, and some volcanics such basalt flows of the National Park 
Volcanics (310-m thick at Site177 close to the Rishon Le Zion-1 well).
      The sedimentary rocks of Pliocene age are represented by 
homogenous clay and marly clay of the transgressional Yafo formation. 
The sedimentation of the Yafo Fm. was dominated by a westward tilting 
of about two degree, which produced an increase in thickness from a few
meters in the central Hashefela area to more than 2,000 m in the 
offshore region.  The Yafo Fm. is overlaid diachronously by the Kurkar 
Group.
     The Avedat Group of Eocene age consists of massive soft 
silicified chalks and marl 50-300m thick related to the Adulam Fm., 
whereas its upper section is composed of massive soft   chalks of the 
Maresha Fm. In the Coastal Plain area, the Avedat group occasionally 
overlies the Albian Talme Yafe Fm., the Judea Group and Senonian 
Mt. Scopus rocks. It is unconformably overlain by the Saqiye and/or the
Kurkar Gr. In the Hashefela region, the Eocene sediments were deposited
in a buried synclinorium situated between the Judea and Samaria heights 
and the Helez structure and they outcrops in the southestern part of the 
Hashefela area.
      Mount Scopus Group is representing by a marl-chalky facies 
of the ‘En Zetim Fm., the silicified chalky limestone of the Ghareb Fm. 
and the limonitic shale of the Taqiye Fm. The upper boundary of the 
Mt. Scopus Gr. is overlain by flinty limestone of the Avedat Gr. The 
thickness of the Mt. Scopus Gr. varies from 0 to 250m. It outcrops in 
the northeastern part of the Hashefela area 
(upper reaches of the Hadera river) and in the Modi’in area.
       CRETACEOUS ROCKS
The Judea Group sequence has been divided into three parts consisting 
of massive dolomites and limestones of the Albian-Cenoman, Yagur Fm.; 
dolomites with interbedded marls of the Cenomanian Negba Fm., and 
limestone, dolomites and marls of the Turonian (Bina and Dalya) Fm.
      The Judea Group, represented generally by the Bina Fm., 
consists of hard white to gray limestone and dolomite; containing 
rudist and coral fragments with a thickness varying from 100 to 160 m. 
The Bina formation. outcrops along the foothills. The upper contact of 
the Bina Fm. is unconformably overlain almost everywhere by the 
'En Zetim Fm.




Figure 1.Geological map and location of the observation points 
 

     During the period July 2002 to April 2003 about 300 microtremor 
measurements were carried out in the Hashefela region (W-155000; 
E – 205000; S –1600000; N – 1715000). The work area is approximately 
2300 km2. Measurement sites were distributed along the lines, which are, 
in fact, extensions of those on the Coastal Plain from coastline to 
foothills. Measurement points were planned in close co-operation with 
Dr. Zohar Gvirtzman (the Geological Survey of Israel). The distribution 
of measurement points is shown in  Figure1). In order to obtain more reliable
microtremor data the main part of the measurement sites comprised two 
observation points separated by tens of meters. 
     Ground motions were recorded using the multi-channel, PC-based, digital 
seismic data acquisition system GII-SDA (see Shapira and Avirav, 1995) 
designed for site response field investigations. The seismometers used are 
sensitive velocity transducers with a natural frequency of 1.0 Hz. Digital 
recordings use 0.2-25 Hz band-pass filter and a sampling rate of 100 samples 
per second. Prior to and during the measurements we checked and determined the
 transfer function of the round motion data, i.e., particle velocity. One 
vertical and two horizontal seismometers (oriented north-south and east-west) 
are installed at each site. All seismometers are wired directly to the 
recording site. 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)]. To obtain systematic and reliable results from the spectra of 
microtremors, we used many time windows that yielded many spectral rations 
that, in turn, were averaged. The average of the two horizontal-to-vertical 
ratios is defined as the site amplification function:

A(f)=1/2n[(i=1..n)(SNS(f)i/SV(f)i)+(i=1..n)(SEW(f)i/SV(f)i)] 
The critical assumption in all experimental site response estimations is 
that the "reference" motion represents the true input to the soil site. 
Even when the surface-rock sites are used as the input (reference site) to 
the basin, in the case of earthquake excitation, amplification factor may be 
underestimated at the basin site by a factor of 2 to 4 depending on frequency 
and site (Steidl at al., 1996; Zaslavsky et al., 2002a). In Nakamura's method 
the "reference" site is the vertical motion of the ambient vibration and 
therefore we cannot separate ambient noise from the "true" input of their 
wavefield. Moreover, in our opinion, excitation of resonance vibration from 
ambient noise in multi-layers medium is stochastic process. Therefore, it is 
very important to select the appropriate ensemble of windows of the ambient 
vibration in the spectral estimation procedure. The estimates of the spectral 
ratio for Points 240 and 236 obtained from pilot analysis and refined analysis 
are plotted in Figure2. We should point out again here that the transfer 
function may be obtained from the spectral ratio of input and output of linear 
system, but not every spectral ratio is really a transfer function.

Figure 2. Pilot and refined spectral ratios for two sites 

      A number of measurements were made either at or close to boreholes, 
where the thickness of layers and geotechnical description of the local 
sediments are available. Clearly, the main geological structure is reflected 
in the measurements results. In this section we give some examples.
      In Figure3 we present average spectra for three components (two 
horizontal and vertical) and individual and average H/V spectral ratios from 
ambient vibration recorded at Points 324-3 and 331. Lithological sections at 
both sites are represented basically by alternating high-velocity chalk, marl 
and limestone. Depth to the basement is 600 m and 450 m, correspondingly. An 
increase in the spectral levels of the horizontal components is clear at a 
frequency of about 0.45 Hz for Point 324-3  (Fig.3a) and close to 0.55 Hz for 
Point 331 (Fig.3b), while the spectra of vertical components are flat. Hence, 
the H/V spectral ratios curves show amplification with a factor 2.0 at 
frequencies of about 0.45 Hz (Point 324-3) and 0.55 Hz (Point 331).
     A comparison of average horizontal (NS and EW) and vertical components 
spectra from ambient vibration recorded at Points 270 and 281 and its H/V 
spectral ratios are shown in Figure 4. At these points, as in the previous 
example, a rock basement is overlain by chalk, marl and limestone. However, 
the thickness of the sedimentary layers is 200 m and 130 m. The general 
character of these spectra is that the spectral levels for vertical components 
exceed the levels for the horizontal components within the frequency range 0.2 
to 10 Hz. It is worth pointing out that there is a narrow-bandwidth trough at 
a frequency close to 1.9 Hz at 1.2 Hz at Points 281 and Point 270, respectively 
in the spectral levels of the vertical components. Spectra ratios, therefore, 
show a prominent peak at about 1.9 Hz with amplification up to 3 for Point 281 
 (Fig.4a) and near 1.2 Hz with amplification about 2 for Point 270  (Fig.4b). 

Figure 3. Examples of average Fourier spectra and H/V spectral
 ratios obtained for two sites with the weak impedance contrast 
of thick (up to 600m) sediments 

Figure 4. Examples of average Fourier spectra and H/V spectral
 ratios obtained for two sites with the weak impedance contrast 
and intermediate thickness (from 120m to 220m) of sediments 

The H/V ratios at Points 178-3 and 234 are shown in Figure5. For two 
points we can see a similarity among the individual functions not only in 
terms of the peak position and intensity, but also in the whole shape. The 
dominant feature of all spectral ratios at Point 178-3 is a well defined peak 
at about 1.7 Hz with amplification factor up to 4. As shown, the average 
spectral ratio at Point 234 has a predominant peak near 1 Hz with amplification 
up to 6. In these sites the soft sediments and chalk over dolomite show a strong 
contrast and cause main peak.

Figure 5. Influence of soft sediments and chalk with different 
thickness over dolomite on site response function:
 Site 178-3 – soft sediments - 40m, chalk 40m; Site 234 – soft sediments - 65m,
 chalk 70m.

    Examples of the average Fourier spectra and individual and average H/V 
spectral ratios at Points 211-A and 231 are plotted in  Figure 6. According 
to the borehole data these points of measurements are located on sand and loam 
with a thickness of 35–40m overlaying the carbonates of the Judea Gr. The 
impedance contrast is very strong and the spectral ratios at these sites present 
significant peaks in the frequencies 2.0 Hz and 2.5 Hz with amplification above 6.0.

Figure 6. Examples of average Fourier spectra and H/V spectral 
ratios obtained for two sites with thin soft sediments (30- 40m) 
and the strong impedance contrast 

     In  Figure7 we plot spectral ratios at points 329 and 359. These points are
situated on kurkar and clay with a depth to basement of 470m and 630m, 
respectively. We can see that different combinations of sediments produce a 
predominant peak near 0.3 Hz with amplification factor up to 3.  Figure8a 
displays a simplified sketch of the geological section along line “A-A”, 
showing the locations of the points where ambient vibration measurements were 
carried out (see  Fig.1 for location of this line on the geological map). 
 Figure8b reveals amplification factor of site response functions 
(fundamental mode) for these sites. From examination of this figure we can see, 
that spectral amplification varies from one station to another in both maximum 
amplification and the position of the predominant peaks. Further the west, 
resonance frequency decreases from 5 Hz to 0.35 Hz, corresponding to the 
increasing thickness of the sedimentary cover.
      

Figure 7. Examples of average Fourier spectra and H/V 
spectral ratios obtained for two sites with kurkar and clay and 
different depth to basement: a- 470m, b-630m.

Figure 8. (a) Simplified 
geological section in E-W direction (line “A-A” in Fig. 1) and (b) average 
H/V ratios obtained for 6 points.

     A data set of the ambient vibration measurements was used to construct 
the distribution maps of the fundamental frequency and maximum relative 
amplification in the Hashefela region. During our campaign of systematic 
ambient vibration measurements in the Hashefela, we took the opportunity to 
supplement with new data and revise available results from the Coastal Plain. 
Frequency and amplification maps, shown in  Figs.9, 10 integrate all experimental 
data obtained in both the Hashefela and Coastal Plain areas.
     The Hashefela region shows a large variation in the fundamental frequency
(from 0.3 up to 7 Hz) and amplification (from factor 2 up to 7). Comparison of 
the results with the geological structure confirms the general trend of the 
correlation between the dominant frequency and the depth of the reflector 
represented by carbonates of the Judea Group in the central and eastern part 
of the Hashefela region. Low frequency band responses (0.3-0.6 Hz) occur in the 
central part gradually increasing up to 7-8 Hz near the eastern edge of the area,
in the foothills.  In the south and southeast of Hashefela region, where chalks 
and marls of Eocene age outcrop, we observe frequencies in the same band of 
0.3-0.6 Hz. In the west, in the Coastal Plain strip, calcareous sandstone of 
the Kurkar Gr. dominate site response, substituting at depths of 600-700 m for
the carbonates of the Judea Gr. Significant steps in the frequency domain  
(from 0.2-0.3 Hz to 1-2 Hz) caused by reflector substitution, are observed 
along the Coastal Plain. Toward the coastline the frequency, correlating with 
the thickness of the sedimentary deposits above the calcareous sandstone of the
Kurkar Gr., increases up to 4 Hz everywhere, except in the small area located 
on the upstanding block near the northeastern boundary of the investigated 
region. This small area is an object of interest because the reflector changes 
twice within a distance of 4-5 km. According to our measurements, a reflector 
represented by carbonates of the Judea Gr. is substituted by calcareous 
sandstone of the Kurkar Gr. Then, the top of the Judea Gr. rising up to depths
of 700m and higher was again interpreted as a reflector.
      The map of the maximum relative amplification shown in  Fig.10 reflects 
the impedance contrast between the bedrock and the overlying sediments. 
Amplification of factor 2-3 prevails in both the Hashefela and Coastal Plain 
areas, but for different reasons. In the Coastal Plain (see Zaslavsky, 2002c) 
amplification factor 2-3 matches loose deposits lying on calcareous sandstone 
of the Kurkar group, while in the Hashefela, the velocity contrast is defined 
by chalk and marl of Campanian and Eocene ages overlying dolomites of the 
Judea Gr. Moderate amplification values from factor 3 to 4 we observe in 
limited areas at Coastal Plain (Tel-Aviv, Ashdod - up to factor 6) where 
the bedrock is represented by hard calcareous sandstone, as well as in the 
central part of the Hashefela region where the impedance contrast is created 
by combination of the rocks of the Kurkar Gr., clay of Saqie Gr., and marl 
and chalk of the Avedat and Mt. Scopus Groups. Quaternary sediments, directly 
overlaying carbonates of the Judea Gr., cause the higher amplification values 
(from 5 up to 7) we observe in the eastern part of the study area, within the 
belt between the towns of Ramla and Kefar Sava, near the Shomron hills. 
       The main outcome of analysis of the 500 ambient vibration measurements 
(Coastal Plain area included) was to produce a map of zones, each characterized 
by a fundamental frequency and amplification factor. Generalized lithological 
structure and ranges of layer thickness for each selected zone, efficiently 
recognized in the ambient vibration measurements, are stated in Table 2.
      In the future we plan to adjust the overall model for each zone and to 
use this for the purpose of earthquake hazard assessment in the Hashefela region.
 
Figure 9. Distribution of the 
fundamental frequency in the Coastal Plain and Hashefela areas 


 


Figure 10. Distribution of maximum amplification 
level in the Coastal Plain and Hashefela areas 


 


Figure 11. Map of zones division 
in the Coastal Plain and Hashefela areas
  
        In the future we plan to adjust the overall model for each zone and to
use this for the purpose of earthquake hazard assessment in the Hashefela region. 

     One of the major requisites for numerical prediction of site effects is 
to feed the models with reliable parameter values concerning geotechnical 
properties of the subsurface structure. 
     The SHAKE (1971) and Joyner’s (1977) programs, which we use for site 
response modeling, suppose sufficiently detailed knowledge of the thickness, 
density and shear-wave velocity of each layer of the soil column. Data on 
underground structure can be obtained from borehole data. At that point we 
collect all available information on S-wave velocities and suggest rough 
starting values. The S-wave velocity values, which after iterative procedure 
of trial-and-error fitting yield good agreement between calculated and 
observed response functions, are accepted as optimal and used further. 
As initial velocity values for Hashefela we used the results obtained in the 
previous investigations in the Lod-Ramla and Coastal Plain areas (Zaslavsky 
at al., 2001; Zaslavsky et al., 2002c).  These results are summarized in Table 3.
In order to verify the applicability of the velocity models derived in the 
Lod-Ramla and Coastal Plain areas for the Hashefela region and estimate 
shear-wave velocity distribution with depth (for sediment thickness of more 
than 300m) 140 wells with different lithological structures and layer 
thickness were involved in the modeling process. We give below we some 
examples of comparison of the calculated transfer functions and experimental 
spectral ratios.
      Measurement point 210 is located at borehole Ld-19 and its soil column 
consists of sandy loam of the Kurkar Gr. overlaying dolomites of the Judea Gr. 
Using corresponding S-wave velocity values from Table 3 gives good agreement 
in both the fundamental frequency and amplification factor, as shown in Fig.12.
 
Figure 12. Comparison of analytical 
response function and experimental spectral ratio for Point 210 (LD-19 well)
 
     Measurement point 188 is situated at borehole Tunis 5 and represented by 
sandy loam, calcareous sandstone and clay overlaying Judea Gr. carbonates. 
Also in this case we can see in Figure 13, that Vs=650 at a depth of more than 
100 m is assumed for clay of the Yafo Fm. and provides good coincidence between 
the observed and calculated response functions.

Figure 13. Comparison of analytical response 
function and experimental spectral ratio for Point 188 (Tunis-5 well)
 
     We calculated analytical response functions for sites 319, 320, 363, 359, 
395, and many others, located at wells with a soil column including clay layers. 
The thickness of the clay layers varies from 30 m up to 700-800 m. In the models
 with significant misfit between analytical and experimental response function 
we corrected Vs value depending on depth. Our conclusions regarding optimal Vs 
velocity model in-depth are shown in  Fig.15.
 
Figure 14. Selection of the optimal model 
for Lachish 14 well (Point 346).
    
 In order to estimate vertical distribution of Vs for chalk layers, the 
analytical response functions were calculated at measurement points 201, 406, 
281, 270, 266, 267, 276, 309 etc. closed to boreholes. The depth of chalk bedding
 at these drilling sites changes from tens of meters to about 600 meters. 
According to our investigations in Lod-Ramla shear-wave velocity of united layer 
of marl and chalk facies varies from 700 m/s at depths of 0-50m to 950 m/sec at 
depths of more than 200 m. Figure 14 illustrates our failed attempt to use 
Vs=950 m/sec for a chalk layer with 590m thick bedding at a depth of 10m (point 346,
Lachish 14 well). The difference between the fundamental frequency of the 
experimental and theoretical transfer functions is obvious.  By trial-and-error
fitting we found that the Vs value giving satisfactory agreement is 1200 m/sec.
Consecutive analysis of the ambient noise measurements at boreholes allowed us 
to suggest a velocity model for the chalk (see  Fig.15).  
     Velocity-depth distribution for clay and chalk-marl derived by means 
iterative fitting procedures are depicted on  Fig.15. 


Figure 15. Shear-wave profile estimated from 
ambient vibration measurements
 

     After recent earthquakes, a priori estimations of site effects became 
a major challenge for efficient mitigation of seismic risk, because, in the 
case of moderate earthquakes, significant damage and loss of life has been 
directly related to local geotechnical conditions. As illustrated by a test 
performed in the Turkey Flat, located near the Parkfield section of the San 
Andreas Fault (Field and Jacob, 1993), the numerical approach requires a 
very good knowledge of the local structure responsible for site effects. 
Only in the ideal case is it possible to perform a geophysical and 
geotechnical campaign with, for instance, cross-hole tests in order to obtain 
a reliable S-wave velocity and density of different lithological units and 
thickness of each layer. On the other hand, the known experimental techniques 
obtain reliable estimates of site effects, when the data are several tens of 
good quality earthquake recordings at the sites. These techniques, however, 
are costly, particularly in regions such as Israel with relatively low 
seismicity.The H/V ratio, i.e. the ratio between the Fourier spectra of the 
horizontal and vertical components of ambient vibrations (Nakamura’s method) 
has proved to be a valuable tool in determining first modes of transfer 
functions of site effects in the Hashefela and Hasharon areas, if applied 
“with care” and “appropriately”. 
     Most of all we require analytical models to improve the reliability and 
relevancy of the results obtained. Where reliable data on subsurface velocity 
structure are lacking, we derived them via iterative procedure of adjusting 
theoretical response function to experimental spectral ratio at measurement 
points close to drilling sites. In this way in the Coastal Plain investigations 
revealed that the reflector (half-space) in the area of Ashkelon to Binyamina 
is calcareous sandstone of the Kurkar group, while at Carmel coast, according 
to our observations, the reflector is carbonates of the Judea group. Velocity 
models obtained on the basis of the extensive database of microtremor 
recordings in the Lod-Ramla area, considering our conclusions in the Coastal 
Plain, were distributed over the Hashefela region. We give some interesting 
examples for discussion.
     Measurement point 218-1 is located at well Yarkon 1 on the upcast side of 
the fault marked by borehole data (see  Figs.16). The depth of the 
Judea Gr. limestones is 285 m. Modeled fundamental frequency and maximum 
amplification factor are 0.6 Hz and factor 3, while the measurements give us 
a frequency of 0.4 Hz, which corresponds to the depth of more than 400 m.

Figure 16. Fragment of the structural 
map of top Judea Gr. and location of Yarkon 1 well (Point 218-1)
 
     An analogous situation where the depth of the Top Judea Gr. determined by 
well data and the depth of reflector inferred from our measurements are 
considerably different, is observed at Nir’am-4 well (measurement point 364). 
In the geological interpretation of the well, the Top Judea Gr. is at a depth 
of 237 m that corresponds to a frequency of 0.8 Hz. The experimental spectral 
ratio shows a very clear peak at 0.65 Hz. From this coincidence of experimental 
and calculated response functions may be reached only when the reflector depth 
is 300 m.
      The issue of kurkar velocity must be solved separately at each site, 
because the Kurkar group represented by alternating calcareous sandstones, 
“hamra” and sand is usually marked as a united layer in well descriptions. 
In the previous investigations in the Lod-Ramla and the Coastal Plain areas 
we showed that ambient noise measurements might be a way to obtain a general 
idea of the subsurface structure. Here we give an example of a “kurkar issue” 
solution in the Hashefela region. The upper 100 m of the lithological column of 
Brur-3 well are described as calcareous sandstone of the Kurkar group. 
Calculating the analytical response function with appropriate S-wave velocity 
for sandstone yields amplification factor 2.5 at frequency 0.65 Hz while the
 result of the measurements at point 350 close to the well are: amplification 
factor of 4.8 at frequency 0.6 Hz. We repeated the procedure assuming Vs equal 
to 350 m/s (sand) and reached the desired result (see  Figs.17). A fairly thick 
layer of sediments above reflector (about 300m) explains the small alteration in 
fundamental frequency for two models. 

Figure 17. Comparison between theoretical response 
function and experimental spectral ratio for Point 350 (Brur-3 well). 
Experimental ratio - red line; theoretical function, calculated 
assuming for kurkar layer: Vs =700 m/c - dashed line; Vs=350 m/sec –black line
 
       Using the database of ambient vibration measurements in the Hashefela 
region, it is possible to show correlation between fundamental frequency and 
thickness of sediments overlaying dolomites of the Judea Gr. for a wide range 
of sediment thickness from ten meters up to 700 m. The graph of this dependence 
is plotted in  Fig.18. 
 
Figure 18. Observed fundamental frequency 
plotted versus sediment thickness of deposits overlaying dolomites 
of Judea Gr. in the Hashefela region 
   
    The significant deviations of the data points in both frequency and 
thickness values from strict linear relationships, despite a correlation 
coefficient of 0.85, are probably a consequence of lithological inhomogeneties. 
Drawing a straight line through the data points assumes an averaging of 
lithology and shear-wave velocity depth dependence. A thickness derived from 
this correlation cannot be valid for a particular site. Hence, the practical 
relevance of this correlation for calculating of models at locations, where no 
borehole information is available, is questionable.  In the final version of 
this report we plan to extend the database of measurements and take into 
consideration lithological composition of the sediments above reflector.
        Some important conclusions from the experiment discussed in the 
present study were reached:
          Maps of fundamental frequency and maximum amplification factor 
constructed on the basis of 500 ambient vibration measurements in the 
Hashefela and Coastal Plain areas exhibit amplification of factor 2-7 
over the frequency range from 0.3 up to 7 Hz. The key parameter of ground 
motion amplifications is impedance contrast between soil deposits and 
underlying bedrock, therefore we observe the same amplification level 
of factor 2-3 in areas with a very different geological structure: loose 
deposits lying on calcareous sandstone of the Kurkar group at Coastal Plain 
and chalks of the Avedat group overlying dolomites of the Judea Group.
          The fundamental frequencies determined in the present study in 
the Hashefela region were found to correlate generally with the dip of the 
reflector represented by dolomites of the Judea Gr. from the east to the 
west. Contrarily, in the Coastal Plain, the increase in the fundamental 
frequency eastward matches the dip of the reflector represented by calcareous 
sandstone of the Kurkar Gr. from the west to the east that was confirmed by 
additional measurements in the Coastal Plain. A change in reflector occurs at 
a depth of top Judea Gr. of 600-700m. The boundary is marked on the maps by red 
line in the central part of the study area, as well as near the northwest corner.
          The uncertainties associated with the proposed subsurface models
 yield a too high a variability between the analytical site response 
functions. Hence, we found it useful to compare the possible analytical 
functions with those obtained empirically. After a trial and error process, 
we obtained 1D models that yield response functions consistent with our H/V 
observations. These models were used to define and constrain the average 
shear wave velocities of the unconsolidated materials overlying the bedrock. 
In particular we inferred velocity-depth model for clay and chalk, which 
show variation of Vs from 500 m/s up to 700 m/s within a depth interval of 
50-450 m for clay; and from 700 m/s up to 1300 m/s in the depth range of 0-650 m. 
          Owing to the fact that the H/V spectral ratio techniques are 
relatively simple and inexpensive, we would strongly recommend that they be 
performed in site response investigations to support and verify theoretical 
calculations. Predictions based on models inferred only from geological and 
geophysical information may differ significantly from empirical estimates 
owing to the geological complexity of the site and the significant uncertainty 
associated with evaluating model parameters.
          Based on subsurface information from borehole data and the 500 site 
response measurements, we divided the study area into twelve zones, 
each characterized by resonance frequency and amplification of fundamental 
mode of site response function. These models will be used in earthquake 
scenarios for damage and losses assessments. The ground acceleration and 
elastic response spectra in the each zone will be presented as a result 
convolution analysis of rock ground motion with analytical transfer function 
of zone. It must be remembered that our results for microzonation studies 
were obtained using a very sparse grid (7 km2) and can be used in earthquake 
scenarios for damage and losses assessments only on a large scale. 
Site effects assessment at small scales in urban areas requires much more 
dense grid (0.2*0.2 km2).
         We suggest that future research efforts in investigated areas should 
have a two-fold focus:
          to identify sediment sites where known or suspected strong 
          impedance contrast; 
          to investigate ground motion characteristics for these sites 
          using very dense grid.

     Our cordial thanks for the financial support of the Steering Committee for the
National Earthquake Preparedness and Mitigation.
     We wish to thank Dr. Zohar Gvirtzman for his willingness to share knowledge and
information on geology of the region and, especially, for his contribution in the 
planning of measurement points.
     We are most thankful to Dr. A. Hofstetter and L. Fleisher for fruitful 
discussion. 
     Thank are also due to I. Chelinski, D. Artzi and Y. Menahem for their assistance
in preparing this report.

Field, E.H., and Jacob, K.H., 1993. Monte-Carlo simulation of the theoretical 
        site response variability at Turkey Flat, California, given the 
uncertainty in the geotechnically derived input parameters, Earthquake Spectra,
 V 5, No. 9, 669-701.
Fleischer, L., Gelberman, E. and Wolff, O., 1993. A geological-geophysical 
        reassessment of the Judea Group, IPRG Report No. 244/147/92.
Fleischer, L., Gafson, R.,2000. Top Judea Group- digital structural map of 
        Israel. Phase 3, part 2, IPRG Report No. 873/55/99.
Gitterman, Y., Zaslavsky, Y., Shapira, A., and Shtivelman, V., 1996. 
        Empirical site response evaluations: case studies in Israel, Soil 
Dynamics and Earthquake Engineering, 15, 447-463.
Ezersky M., 1999. Seismic borehole studies at the Qiryat Shemona, Beit Shean and 
     Hagoshrim sites, GII Report No.803/35/99.
Gvirtzman, G., Shachnai, E., Bakler, N., and Ilani, S., 1984. Stratigraphy 
    of the Kurkar group (Quaternary) of the Coastal Plain of Israel, GSI 
    Current Research, 1983-84, pg. 70-82. 
Gvirtzman, G., 1969. Subsurface data on the Saqiye Group in the Coastal 
    Plain and Hashelphela regions, Israel, Geological Survey of Israel, OD/1/69.
Gvirtzman, G., 1970. The Saqiye Group (Late Eocene to Early Pleistocene) in 
    the Coastal Plain and Hashephela regions, Israel, Ph D. Thesis, The Hebrew
    University, Jerusalem.
Nakamura, Y., 1989. A method for dynamic characteristics estimation of 
    subsurface using microtremor on the ground surface. QR of RTRI, Vol. 30, 
    No.1, 25-33.  
Nakamura, Y., 2000. Clear identification of fundamental idea of the 
    Nakamura's technique and its applications, 12th World Conference of 
    Earthquake Engineering, Auckland, New Zeeland, Proceedings, No.0084, pp8.
Shapira, A., and Avirav, V., 1995. PS-SDA Operation Manual, IPRG Report 
     Z1/567/79, p.24. 
Shapira, A., Feldman, L., Zaslavsky, Y., and Malitzky, A., 2001. Application 
    of a stochastic method for the development of earthquake damage scenarios: 
    Eilat, Israel test case. Computational Seism., 32, 58-73. 
Shtivelman, V., 1999. Using surface waves for estimating shear wave velocities 
    in the shallow subsurface onshore and offshore Israel, European J. of 
    Environmental and Engineering Geophysics, 4, 15-35.
Sneh, A., Bartov, Y., and Rosensaft, M., 1998. Geological Map of Israel 
    1: 200,000, Geological Survey of Israel.
Zaslavsky, Y., Gitterman, Y., and Shapira, A., 1995. “Site response 
    estimations in Israel using weak motion measurements,” Proceedings of 
    5th International Conference on Seismic Zonation, Nice, France, pp.1713-1722.
Zaslavsky, Y., Shapira, A., and Arzi, A.A., 2000. Amplification effects 
    from earthquakes and ambient noise in Dead Sea Rift (Israel), Soil 
    Dynamics and Earthquake Engineering, 20(1-4), 187-207.
Zaslavsky, Y., and Shapira, A., 2000. “Questioning nonlinear effects in 
    Eilat during MW=7.1 Gulf of Aqaba Earthquake,” Proceedings of XXVII 
    General Assembly of the European Seismological Commission, Lisbon, 
    Portugal, pp.343-347.
Zaslavsky, Y., Shapira, A., Gorshtein, M., Kalmanovich, M., Giller, V., 
    Livshits, L., Giller, D., Dan I., Leonov, J., and Peled U.2001. 
    Microzoning of the earthquake hazard in Israel. Seismic microzoning of 
    Lod and Ramla, GII Report N 569/143/01.
Zaslavsky, Y., Shapira, A. and Arzi, A.A., 2002a. Earthquake site response 
    on hard rock – empirical study. Proceedings of 5th International Conference
    on Analysis of Discontinuous Deformation, ICADD-5, Beer-Sheva, Israel 
    pp 133-144.
Zaslavsky, Y., Shapira, A. and Kenigsberg, M., 2002b. Earthquake site response
    study for designed bridges in Israel, Proceedings of 12th Conference on 
    Earthquake Engineering, London, UK, Paper Reference 059.
Zaslavsky Y., Shapira, A., Gorshtein, M., Kalmanovich, M., Perelman, N., 
    Giller, V., Livshits, L., Giller, D., Dan I., Aksienko, T., and Ataev, G., 
    2002c. Generalization of site effects for earthquake scenario applications: 
    the Coastal Plain area, GII Report, N 595/274/02.
Zaslavsky Y., Shapira, A., Gorshtein, M., Kalmanovich, M., Giller, V., Livshits,
    L., Giller, D, and Dan I., 2002d. Detailed distribution of ground motion 
    characteristics detected by ambient noise measurements – a case study in 
    Lod and Ramla, Israel. The XXVIII General Assembly of the European Seism. 
    Comm., Genoa, Italy, 1-6 September.
Zaslavsky, Y., Begin Z. B., Gorstein, M., Kalmanovich, M., and Shapira, A., 
    2002e. On the correlation between surface geology and site response along 
    the coastal plane of Israel, GII Report No. 595/222/02, GSI Report 
    No. GSI/24/02.
Zaslavsky, Y., Shapira, A., and Leonov, J., 2003. Empirical evaluation of site
    effects by means of H/V spectral ratios at the locations of strong motion 
accelerometers in Israel, Journal Earthquake Engineering, in review.