Journal of Engineering Geology

---

./files/site1/files/1Extended_Abstract.pdfExtended Abstract
(Paper pages1-28)
Introduction
The earthquake is one of the most devastating natural disasters that always threats human societies in terms of health and financial issues. Iran is one of the most seismic prone countries of the world due to locating on Alpine- Himalayan Orogenic belt. On the other hand, growing population and increased construction of tall buildings, increases the damages caused by large earthquakes, especially in large cities. Karaj is one of the most populous cities in Iran which there has been considerable industrial and economic development in recent years. When an earthquake occurs, seismic waves radiate away from source and travel rapidly through the earth crust. When these waves reach the ground surface, they produce shaking that may last from several seconds to a few minutes. During earthquakes, different alluviums with different structures show various reactions. It is well-accepted that, besides the earthquake magnitude and fault distance, local geologic conditions, known as site effects, can also exert significant influences on characteristics of the seismic waves such as amplitude, frequency content and duration of strong ground motion at a given location. The seismic ground motion at any site is influenced by the type of soil in that region. Younger and softer soils usually amplify ground motion more than older soils or bedrocks .
There are theoretical and experimental methods to evaluate the site response. In the present study, the Nakamura's H/V spectral ratio method has been used to evaluate the resonance frequency in 37 locations at Karaj site. In addition, a preliminary 1-D site response modelling has been conducted using Deepsoil program according to downhole, array and geology data. Site frequencies obtained from modelling are presented and compared with site frequencies obtained through microtremor measurements.
Materials and Methods
Single station microtremor measurements at the Karaj site were carried out by the International Institute of Earthquake Engineering and Seismology (IIEES) in 2012 with a three-component broadband seismometer (Guralp CMG-6TD). In the present study, we have used 37 microtremor data along the north-southwest profile because at this profile, geological section was available and these stations contained geotechnical boreholes data. Dynamic range of sensor changes between 0.033 -50 Hz and has a natural period of 1 second. 24-bit analog-to-digital (A/D) converter digitized the recorded data. The recording system was operated continuously for about 30 minutes with sampling frequency of 100 Hz. The use of ambient vibrations for analysis of the local site effects has been studied in detail in the framework of the European research project SESAME (Site Effects Assessment Using Ambient Excitations). The recommended guidelines on the H/V spectral ratio technique are the result of the comprehensive and detailed analysis performed by the SESAME participants during three years of investigations (2001-2004).
H/V spectral ratio was carried out by the Geopsy software. The process starts by converting data from binary format to ASCII format. After DC offset removal, eighth order Butterworth band pass filter used within the range of 0.1 Hz to 50 Hz. The Anti-triggering algorithm STA/LTA has been selected to reject energetic transients from ambient vibration recordings, so STA and LTA were considered respectively 1 and 30 second. Minimum and maximum STA/LTA thresholds were selected between 0.2 and 2.5. For each station, the time-series of the record is divided into windows of 40 to 100 seconds in three components with an overlap of 50%. Also, a cosine taper with the length of 5% of the total window length was used at each end.
The amplitude spectra of each selected window is computed with a fast Fourier transform (FFT) and smoothed using the Konno-Ohmachi function (Bandwidth=40). Then, two horizontal components are merged by squared average. Finally, the H/V spectral ratio of Nakamura is applied for each individual window, and the final predominant frequency is obtained by averaging the H/V spectral ratio of all window. The presence of clear peak on H/V spectral ratio curve is indicative of the impedance contrast between the uppermost surface soil and the underlying hard rock, where large peak values are generally associated with sharp velocity contrasts, and is likely to amplify the ground motion. The H/V spectral ratio in some stations shows a clear peak and at the others might show two or multiple peaks which represents the geologically complex areas. Calculated dominant frequency changes between 0.4 and 2 Hz. These low values indicate the existence of basement at greater depths and large thickness of sediments on basement (Parolai et al., 2002).
Site modelling
The results of H/V spectral ratio are affected by the local geologic structure. Based on this assumption, we can produce theoretical H/V curve with knowledge of the geologic structure in the area. One-dimensional modelling is a suitable method to evaluation of the site response due to the local geology which requires geotechnical and geophysical data. In the one-dimensional modelling, it is assumed that all boundaries are horizontal in the infinite media and the response of a soil deposit is predominantly caused by SH-wave propagating vertically from the underlying bedrock. In this present study, one-dimensional modelling was carried out using Deepsoil software. Due to the very small deformations in soils by microtremor and producing a low levels of strain, we applied the linear method to evaluate the ground seismic response during mild earthquake shakes. In this software, homogeneous and isotropic soil profile is considered as N horizontal layers. The site response (transfer function) is evaluated by parameters such as layer thickness (m), density (ρ), shear modulus (G), and damping factor of layers (β), which are obtained from available geotechnical boreholes.
Usually, engineering bedrock is considered for the purpose of numerical modelling. According to TC4 (1994), the seismic bedrock was defined as a layer with a shear wave velocity of more than 600 m/s. Shima (1978) recommended that the upper crust with a shear wave velocity of about 3000 m/s, is adopted as bedrock when large scale structures with longer vibration period are being considered. International building code (ICC2000) has defined the seismic bedrock by a shear wave velocity of more than 760 m/s. According to Unified Building Code (UBC97), bedrock is defined into two groups: A (very hard rock with a speed of more than 1500 m/s) and B (rock with a speed of 760 to 1500 m/s). Therefore, the proposed values of the shear wave velocity are different for considering seismic bedrock. In order to consider the uncertainty of the shear wave velocity in the present one-dimensional modelling, three scenarios for the bedrock, were performed with three speeds of 760 m/s (based on engineering bedrock), 1300 m/s (bedrock geology), and 2500 m/s (corresponding to tuff-andesite of the Karaj basement) at different depths, according to the regional geological map. Then, three scenarios of the numerical modelling were compared with microtremor transfer function.
1. One-dimensional modelling at the Karaj site using downhole data for engineering bedrock (> 760 m/s)
In order to access the shear wave velocity profile for 1-D modelling, downhole data from 21 boreholes were used in nine sites which were available up to the maximum depth of 50 meters at 20 boreholes and 96 meters at A09 borehole. Low thickness of alluvium (about 17-85 meters) was considered with engineering bedrock (>760 m/s) for numerical modelling. The results represent higher frequency range compared with the microtremor data. In some previous studies where engineering bedrock had been defined by shear wave velocity values between 700 to 800 m/s in 1-D modelling, the results of the theoretical model is incompatible with experimental results. Thus, it seems that it is not suitable to consider the engineering bedrock in 1-D modelling.
2. One-dimensional modelling at Karaj site using microtremor array data for geology bedrock (> 1300 m/s)
By considering the seismic bedrock (>760 m/s) at depths of 17 to 85 meters and calculating the one-dimensional transfer function, the peaks in higher frequency compared with the experimental method is observed. According to reliability of experimental H/V results which has been proved by researchers around the world (Haghshenas et al., 2008), the difference between the transfer function results in experimental and theoretical methods indicates that two variables of shear wave velocity or depth of bedrock and alluvium thickness have not been properly modeled. It seems that in order to get better results, it’s necessary to analysis by considering the geology bedrock at greater depth. Tchalenko, et al., (1974) considered lower part of Plio-Quaternary sediments of Hezardareh Formation and Miocene marl-limestone of Upper Red Formation as the bedrock in the Karaj plain. Shafiee and Azadi (2006) computed shear wave velocity characteristics of these geological units throughout Tehran city. Therefore, a mean velocity of 1300 m/s was considered for the geology bedrock during the modelling.
In order to access the shear wave velocity profiles at greater depths, microtremor array stations were designed by seven seismometer with 100 m radius at A09 (site 8) borehole. As it can bee seen, a clear contrast at a depth of about 230 m is observed. Therefore, the modelling was carried out by taking 230 m alluvial thickness on geology bedrock according to lithology of the region. The result of this modelling has shown a peak at frequency range of 0.87 Hz that is compatible with the microtremor peaks at this site. In other site this modelling was performed using array and downhole data. The results indicated that the first effective contrast occurs at depth of 200 to 300 meters.
3. One-dimensional modelling at the Karaj site for basement (> 2500 m/s)
Transfer functions obtained from the previous model, did not cover low frequency peaks in the experimental methods. Therefore, the presence of other low-frequency peaks is either due to the geometry of the sedimentary basin or deep contrast. It seems that due to the geology of the region, tuff- andesite of the Karaj Formation as basement plays an important role in the creation of low-frequency peaks. Therefore, to obtain a better model, deep contrast was considered about 2 kilometers due to differences in the type of bedrock with a shear wave velocity of 2500 m/s. For this purpose, according to the properties of the Upper Red Formation, an average constant speed of 1400 (m/s) was considered in modelling and by changing the thickness of this layer, the modelling was continued in a trial and error manner until the numerical model is consistent with microtremor peaks. The modelling results in nine site indicate that there is basement at the depth of 2000 to 2250 meters.
Two-dimensional model of the Karaj site
Using the one-dimensional analysis and evaluation of the geological map of the area, two dimensional geological structure was rebuilt in studied profiles. Green and gray tuffs and igneous rocks of Karaj Formation outcrops in north of Karaj and constitute the Alborz Mountains. This Mountains eroded by the action of rivers and were deposited in the form of large alluvial fans. Coarse sandy sediments were deposited near mountains wherein energies of rivers and streams were extremely high (site 1 to 4). Furthermore, fine-grained sediments were deposited at far distances by decreasing in the energy of streams (site 5 to 9). Berberian et al (1985) divided B Formation in two parts: heterogeneous deposits of sand, gravel, rock and clay in north of Tehran (Q_bn) and silts and clays of Kahrizak (Q_bs) in south of Tehran. According to 1-D modelling, thickness of this layer is about 200 to 300 m which has been deposited on geology bedrock. As mentioned before, lower parts of Hezardareh Formation at the north of Karaj and Upper red Formation in the south west of Karaj are considered as geology bedrock. Upper Red Formation was deposited with unconformity on tuff-andesite of the Karaj basement at depths of 2000 to 2250 meters.
Conclusions
The use of empirical methods based on microtremor is an efficient way to estimate the site effects in Karaj city, although the use of earthquake records could provide better evidence of the depth and geometry of basement. One-dimensional modelling of shear wave velocity profiles obtained from downhole data and considering the engineering bedrock (> 760 m/s) at depths of 17 to 85 meters, is not a good way to estimate the dominant frequency of alluvium. By considering the greater depth of alluvium and using shear wave velocity profiles obtained from microtremor array, 1-D modelling was carried out for geology bedrock (1300 m/s). Therefore, peak frequency in transfer function at the range of 0.87 Hz has been associated with effective contrast at depths of 200 to 300 meters. It seems that Karaj basement (> 2500 m/s) with about 2 kilometers depth plays an important role in the production of low-frequency peaks in transfer function.
 

---

Introduction
Local site conditions considerably influence all characteristics of the ground strong motion including the domain, frequency content, and duration. The level of such an effect could be considered as a function of geometry, properties of the materials embedded in the underlying layers, the site topography, and properties of excitement. Site effect fall into two categories: a) the effect of the surface soft layers triggered by the shear velocity differences between the soil layers and b) the surface and subsurface topography effects that lead to the wave reflection and refraction based on the site geometry, and subsequently enhance the level of amplification.
Since most cities have been constructed in the vicinity of or on sedimentary basins, geotechnical earthquake engineering devotes particular attention to effects of the sedimentary basins. Basin edge curvature deposited with soft soils are capable to trap the body waves and generated surface waves within the deposit layers. Such waves could create stronger and lengthier vibrations than those estimated in a 1D analysis that assumes the shear waves to be vertically propagated.
Although critically important, the 2D effect of the site has not been included in seismic codes and standards of the world. This might be due to the fact that the site effect depends on a number of parameters such as the site geometry, the type of wave excitement, properties of the materials, etc. that in return make it almost out of the question to make predictions about the effect. This study was an effort to compare the responses of four sedimentary basins with hypothetical geometries of rectangular, trapezoidal, elliptical, and triangular shapes in order to examine the effect of the geometrical shape of the basin on its responses and the extent of the response sensitivity to the excitation frequency of the wave. The study assumed the edge to depth proportion to be both constant and equal in all four basins so that the effect of the geometrical shape could be equally examined and compared in all four basins.      
Material and methods
In order to validate the results of the sedimentary basin modeling, firstly, ABAQUS finite element software was used to create a free field motion of a semi-circular alluvium valley in accordance with Kamalian et al. (2006) and Moassesian and Darvinsky (1987).  Then, the results from the model were compared with those from the above mentioned studies. The following descriptions are to present the model in details.
To evaluate the geometrical effect of the sedimentary basin on its response, the authors relied on the software to examine four sedimentary basins with the fundamental frequency (2.04 Hz). The basins enjoyed rectangular, trapezoidal, elliptical, and triangular geometrical shapes with a constant edge to depth proportion (49m to 300m respectively). The implicit method was also applied to perform the dynamic analysis. The materials were all viscoelastic and homogeneous. The soil behavior/treatment model was considered to be of a linear nature.  The Rayleigh damping model was used to specify the damping level. The soil element was a plane strain and SV waves (the Ricker wavelet) were used for seismic loadings in a vertical dispersion. The side boundaries (right and left) of the model were of a combinational type (viscous and free field boundaries); the down side boundary was composed of viscous. To achieve higher levels of wave absorptions, heavy columns were used as the free filed columns.
Next, it was the time to conduct the 1D analysis of the site. Three waves were in use in order to examine the effect of the frequency content of the excitation load on the basin response: 1) a wave with the dominant frequency of 1Hz that was out of the frequency range of all basins (2.04 Hz), a second wave with the dominant frequency of 2Hz that was close to the fundamental frequency of all basins, and a third wave with the dominant frequency of 4Hz. The waves were applied to a 2Dmodel. The results were compared with those obtained from a 1Dmodel in terms of the timing.
Then, the basin responses to all three waves (1, 2, and 4 Hz) were subjected to an individual analysis in order to examine the sensitivity of each basin response to its geometrical shape. Results indicated that while the responses of the rectangular and trapezoidal basins were significantly more sensitive to the excitation frequencies, the elliptical and triangular basins showed more stable behaviors to such frequencies. The final stage of the study was dedicated to examine the site 2D effect during the ground motion.
Results and Conclusions
According to the results of the present study, it could be suggested that the geometrical shape of the sedimentary basin has a significant effect on the responses of the field of seismic waves and that it could result in so different responses from the ones attained after a 1D analysis of the site. In addition, the pattern of the seismic waves’ responses is highly dependent on the geometrical shape and the frequency content of the seismic load. Also, the location where the maximum horizontal acceleration occurs along with the sedimentary basin depends on the excitation wave and varies accordingly. Further, it could be suggested that the site 2D effect results in both considerable amplification and an increase in the length of ground motion.
The results of the 2D analysis showed remarkable differences with their 1D counterparts: a 1.45 larger response for the rectangular basin, a 1.28 larger response for the trapezoidal basin, a 1.22 larger response for the elliptical basin, and a 1.19 larger response for the triangular basin.
With the frequency of 1 Hz where the excitation frequency is out of the basin range (i.e. the excitation frequency is below the lowest frequency of basin), the sedimentary basin did not show any signs of amplification and chaos (unlike two other frequencies); instead, it was a cause for de-amplification.
The frequency of 2 Hz that is subject to resonance resulted in amplifications (absent in 1D analysis) and there are traces of a reduction in the acceleration responses near to the edges of the basins. The proportion of the amplification (in the center of the basins) in 2D to 1D analysis was 1.4 for the rectangular basin, 1.28 for the trapezoidal basin, 1.22 for the elliptical basin, and 1.15 for the triangular basin.