Introduction

Pakistan is susceptible to various natural disasters. It has encountered earthquakes, floods, droughts, and landslides as a consequence of its unstable geological setting. Few recently recorded seismic events of magnitude 7.0 or more have jolted the entire country resulting in infrastructural devastation and heavy death tolls. October 2015 Hindu Kush earthquake (7.5Mw) is an example of one of such events which have posed severe threat to people and infrastructure residing in the northern and north western areas of Pakistan. This event triggered more than 100 aftershocks some of which were more than 6.5 Mw.

Local site conditions are generally considered to be a substantial factor in case of severity of damage that occurs in association with seismic events. Such conditions are critical in understanding the phenomenon of sediment-induced amplification; those are characteristics of the most hazardous zones (Bard 1999; Şafak 2001). Bedrock depth, bedrock variability, geometry and the characteristics of different soil layers, and irregularities in topography are some of the most significant aspects of local site effects that can play a vital role in highlighting the outcome of soil amplification (Athanasopoulus et al. 1999; Şafak 2001). Geological configurations, such as basins and valleys, are considered to be topographic and lithological irregularities. Seismic waves can be amplified because of acoustic impedance contrast among the geological boundaries as well as lateral discontinuities. This phenomenon can lead to seismic hazards in sites with soft soil or topographic undulations. Various studies have explained that individual sites can be defined by particular resonance frequencies at which the amplification of ground motion can be observed (Bard 1999; Mukhopadhyay and Bormann 2004). There are numerous approaches for undertaking site response analysis of particular sites. One empirical approach is to make use of microtremors or ambient noise measurements. H/V spectral ratio technique has been used by very few researchers in Pakistan to characterize the site response. No previous study has been conducted to characterize the site response of Attock Basin, which is not only densely populated but also comprises of some important and valuable infrastructures. The present work highlights site response of Attock Basin using different parameters, e.g., fundamental frequency (f 0), amplitude (A 0) of corresponding H/V spectral ratios, alluvium thickness (H), and soil vulnerability index (K g), using microtremor recordings. Evaluated fundamental frequencies were then converted into bedrock depths using regression equation developed by Parolai et al. (2002). The results were then correlated with available borehole data to enhance the credibility of the study.

Geology of the Attock Basin

The Attock Basin (previously called the Campbellpore Basin) borders on the south of Peshawar Basin, dissected by the Attock-Cherat ranges, as shown in Fig. 1. Precambrian rocks belonging to the Lesser Himalayas are overlain by Paleozoic to Neogene sequences in the Attock Basin. Attock slates are the oldest exposed rocks of Precambrian age in the study area, and they are overlain by an alluvial and fluvial basin fill of Quaternary deposits (Cotter 1933) as shown in Fig. 2. Attock is the main city in the basin, which also hosts numerous other urban and semi-urban settlements including Ghorghushti, Kamra, Hazro, Hattian, Sanjwal, Burhan, and Taxila (Jabeen 2013). The total population of the Attock Basin was 1.15 million according to the census of 1998. In the north, it is surrounded by the Indus River, to the south by the Kala Chitta range, and in the east by the Haro River. The total area of Attock Basin is approximately 64 km long and 40 km wide (Jabeen 2013).

Fig. 1
figure 1

a Modified seismic zonation map after geological survey of Pakistan. b General map encircling the study area after Hussain and Yeats (2002)

Full size image
Fig. 2
figure 2

Geological map of the Attock basin (Campbell pore Basin) and surrounding areas modified after Hussain et al. (1990)

Full size image

A complex pattern of active faulting reflects north-south shortening with reverse and strike-slip faulting characteristically indicative of seismicity close to the study area (Armbruster et al. 1978). The study area is folded into a number of various synclines and anticlines which commonly verge southward. The Khairabad, Hisartang, and Cherat thrusts are three major faults (Fig. 3) dominating the structure of the study area (Yeats and Hussain 1987). Four left stepping pressure ridges, which involve late quaternary deposits, are located near the study area (Yeats and Hussain 1989). Some high-angle en-echelon faults trending ENE show instrumental seismicity and indicate sinistral slip (Yeats and Hussain 1989).

Fig. 3
figure 3

Structural description of complex structural geology of the study area and probable configuration of hinterland dipping duplex of Kala Chitta and Attock-Cherat Range after Yeats and Hussain (1987)

Full size image

Fluvial and alluvial sedimentation in the Attock Basin started some 1.8 million years ago (Ma) in a southwestward direction (Burbank 1982), as a consequence of the uplift of the Kala Chitta Range along with the Main Boundary thrust, and continued up to approximately 0.6 Ma (Burbank and Tahirkheli 1985; Pivinik and Johnson 1995). This resulted in the thickness of recent sediment showed variation from 0 m at the basement complex exposed near Kala-Chitta Range to few hundreds of meters in the center of the basin. The Attock Basin contains young sedimentary fill in the form of lacustrine, deltaic, fluvial, loess, flood plain, alluvial fan, and stream channel deposits. The lacustrine, deltaic, and fluvial deposits are mainly composed of siltstone, claystone, and loosely cemented sandstone. Loess deposits of sub recent to recent age are composed of reworked silt and clay with thin gravel cover (Burbank and Tahirkheli 1985). Similarly, the sub recent to recent age is allocated to the flood plain deposits that are dominantly composed of sand and gravel and subordinately composed of clay and silt. These deposits are part of the flood plain of the Indus and Kabul rivers. Alluvial fan deposits are also of sub recent to recent age comprising of silt, sand, gravel, and clay. Stream channel deposits are considered to be the most recent and are composed of gravel, sand, silt, and clay material (Fig. 2). The young basin fills lie over the Muree Formation of Early to Middle Miocene age in the Attock Basin (Fatmi 1973). The Muree Formation comprises interbedded sandstone, siltstone, and shale and acts as the first competent lithology underlying the soft sediments in the study area.

Material and methods

In contrast to active seismic methods, which require explosives or an impact source (e.g., a hammer) to generate seismic waves, the H/V method utilizes three component measurements of ambient seismic noise (microtremors induced by wind, ocean waves, anthropogenic activities, etc.) to evaluate the fundamental seismic resonance frequency of a site. The resonance frequency is determined by analyzing the spectral ratio of the horizontal and vertical components of the microtremor (Lane et al. 2008).

Microtremor recordings were made within urban settings in the Attock Basin using a sampling rate of 100 Hz. Various factors were analyzed prior to microtremor readings, e.g., population density, security issues, military installments, and accessibility, etc. Due to the presence of military installations and some security related issues, it was not easy to carry out dense microtremor recording throughout the basin. Moreover, limited availability of borehole data and obligations of the SESAME 2004 guidelines (SESAME 2004) lead to restriction of microtremor recordings at 50 sites. These 50 sites were selected using different geographical and geological maps. Our microtremor recordings covered almost the entire study area especially the sites of dense population and complex infrastructure. A Guralp CMG-40T seismometer recorded the data coupled with a field laptop computer. This seismometer was capable of recording the seismic wave field over a frequency range of 0.03 to 40 Hz. Microtremor recordings of minimum 10 min (10–25 min) duration were conducted at 50 different sites during night to avoid the impact of man-made noise on the signals and thus enhance the signal-to-noise ratio. Microtremor data were acquired as GCF (Guralp compressed format) files and displayed, viewed, edited, and replayed using SCREAM (4.5 version) software. The H/V spectral ratio technique was applied to analyze the fundamental seismic resonance frequency. This technique was applied to evaluate the relationship between Fourier amplitude spectra of the horizontal and vertical constituents of ambient noises. One of the major advantages of this technique was the simplicity of data collection.

The acquired microtremor data sets were analyzed using the Geopsy freeware software suite. H/V ratios were calculated over the range of 0.2 to 15 Hz using a 25-s time window and eliminating the time windows infected with transients—often related to urban sources or cultural noises (industrial and traffic noises). Processing techniques were applied to obtain stationary ambient vibrations. A conventional approach to detect the infected transients was based on a comparison between the short-term average, STA, and long-term average, LTA. The H/V computation was applied to only those windows with an STA/LTA ratio lying between 0.20 and 2.50. A Konno and Ohmachi (1998) filter and cosine taper with 0.25% width were applied to every time window to smooth the Fourier amplitude spectra along with a coefficient of 40 for the band width. DC suppress was applied to remove the effect of instrumental errors on microtremor data. Guidelines proposed by the SESAME syndicate (Bard and SESAME-Team 2005) were followed to check the clarity of H/V peaks in the Attock Basin, are as follows:

  1. 1)

    The time span for the time window (L w) must be 10 times greater than inverse of the frequency f 0, i.e., L w > 10/f 0

  2. 2)

    To evaluate the average H/V curve properly, the calculation of the H/V ratio over an adequate number of windows (n w), i.e., n c (f 0) > 200(where n c = n w × L w), is required.

  3. 3)

    One frequency (f −) exists lying between f 0/4 and f 0, such that Ahv (f −) < A 0/2, while Ahv is the H/V curve amplitude at f −.

  4. 4)

    One frequency (f +) exists lying between 4f 0 and f 0, such that Ahv (f +) < A 0/2, while Ahv (f +) is the H/V ratio amplitude at frequency f +.

  5. 5)

    The H/V peak amplitude must be greater than 2.

  6. 6)

    The standard deviation value must remain less than a frequency dependent threshold as mentioned by the guidelines of Bard and SESAME-Team 2005.

In the above-mentioned guidelines, f 0 is the fundamental frequency, while f + is the frequency which ranges between 4f 0 and f 0 and f − is the frequency between f 0/4 and f 0. When the H/V peak satisfies at least four of the six above-mentioned criteria, then there is an ample assurance that the site under examination represents a significant velocity contrast at a certain depth and is likely to amplify the ground motion during any seismic event (Bonnefoy-Claudet et al. 2009). During data analysis and processing, the above-mentioned conditions were very much focused, and therefore the H/V peaks satisfied at least four of the six reliability criteria.

Fundamental frequency f 0 and H/V amplitude A 0 were evaluated using Geopsy software. Alluvium thickness was calculated by using Parolai’s equation H = 108f 0 –1.551 (Parolai et al. 2002). Keeping in view the research conducted by Warnana et al. (2011), soil vulnerability index was also evaluated using the equation K g = A 0 2/f 0. The H/V method alone was not sufficient to analyze the response of a particular site. Therefore, an inclusive site response analysis could not depend only on results obtained through H/V curves (Bonnefoy-Claudet et al. 2009). The results needed to be correlated with any other geological or geophysical data set. In this case, borehole data was one of the vital information that could serve the purpose very efficiently. However, since boring and drilling are expensive approaches, such datasets were rarely available. Microtremor results are often complemented by other techniques such as seismic refraction, electrical resistivity, and gravity methods (Lane et al. 2008). Therefore, if they are available, borehole data can be used for correlating with microtremor data. In addition, the inverse distance weighted (IDW) method was applied on the evaluated results to present the spatial extent of various site response parameters using ARC GIS.

Results

Results obtained from the data acquired at 50 sites within the study area were highly variable. In general, three types of H/V curves were obtained: (i) clear H/V peaks as shown in Fig. 4a, (ii) broad flat peaks as shown in Fig. 4b with amplitudes greater than 2, and (iii) broad flat curves with amplitudes less than 2 as shown in Fig. 4c. Out of the 50 H/V curves, 36 (72%) of the acquired data showed clear peaks and nine (18%) waveforms showed flat broader H/V peaks with amplitudes greater than 2, whereas only five (10%) H/V curves had broad and flat peaks with amplitudes less than 2. Clear H/V peaks were obtained within the central parts of the study area. These clear peaks represented sharp velocity contrasts between soft sediments (Vs = 350 m/s) and sandstone bedrock (Vs > 2500 m/s) and as a consequence, amplification could be observed at these sites. SESAME guidelines also supported the above-mentioned fact that amplification could be observed at clear H/V peak frequency. Moreover, this argument was also agreed by the research conducted by Konno and Ohmachi 1998, Bard and SESAME team 2005 and Bonnefoy-Claudet et al. 2009. Although, the eastern and western parts of the study area showed haphazard trends in terms of the peak H/V curves, they did show broad peaks in general. These moderate-to-low amplification H/V curves were the consequence of weak velocity contrasts between overlying unconsolidated and underlying consolidated sediments. The flat H/V curves with moderate-to-low amplification were observed due to the presence of low acoustic impedance contrast among the local geological structures that did not create a strong impedance contrast. The above-mentioned observations suggest that weak impedance contrast were responsible for the absence of a clear H/V peak, and similar findings were supported by different researchers (e.g., Konno and Ohmachi 1998; Bonnefoy-Claudet et al. 2009). As far as SESAME guidelines were concerned, H/V flat curves with amplitudes less than 2 should not be entertained for quantitative interpretations, but other studies were more relaxed in this regard (Bard and SESAME team 2005). Since SESAME guidelines were followed in the present study, all data sets with amplitude less than 2, as shown in Fig. 4c, were not considered during interpretation of the H/V curves.

Fig. 4
figure 4

a H/V curves showing sharp single peak acquired from the selected sites in the study area. b H/V curves with broad gentle peaks acquired from the selected sites in the study area. c Flat H/V spectral ratio curves having amplitude A0 less than 2 and are not considered for interpretation by following SESAME (2004) guidelines

Full size image

H/V curves representing the fundamental frequency of sediments along the x-axis and H/V amplitude along the y-axis are shown in Fig. 4a–c. The gray bar in the H/V curves represented the average fundamental frequency f 0 and its corresponding H/V amplitude, denoted by A 0. Each H/V figure showed wiggles with variegated color, thus representing the existence of ambient noise at different frequency ranges. The upper (new high-noise model) and lower (new low-noise model) discontinuous lines were enclosed by middle continuous lines. The calculated fundamental frequency f 0, showed high variability in the range from 0.6 to 12.4 Hz. Similarly, the maximum amplitude A 0 of the H/V spectral ratio following the SESAME guidelines was very high (calculated to be 8.8). The thickness (H) of the soft sediments resting on the bedrock was evaluated by Parolai’s equation, i.e., H = 108f 0 –1.551 (Parolai et al. 2002) ranged from 2.1 to 238.5 m. The thickness was then correlated to the sparse borehole data obtained by various construction companies working at particular sites, e.g., AJK and NESPAK; the results showed a consistent correlation, as shown in Fig. 5. It is always highly recommended that geophysical data should be double checked with the geological data to ensure the reliability and competency of the study. Keeping the aforementioned idea in mind, borehole logs were obtained at the same sites where the ambient noises were recorded. At Hazroo, the thickness of soft sediments was estimated to be approximately 11.24 m using Parolai’s equation. When this thickness was correlated to the available borehole data, it was found that the bedrock was not observed up to a depth of 10 m. Similarly, at Kamra, the bedrock depth estimated using the H/V technique was approximately 19 m, and the borehole did not encounter bedrock within the upper 10 m. At Taxila, the thickness of unconsolidated sediment was expressed as 65 m. Borehole logs from the same site do not show bedrock, even at depths greater than 20 m. Since any other type of geophysical data set, e.g. gravity, reflection, refraction, and electromagnetic data was unavailable, therefore our dependency was largely on the borehole data. The correlation of estimated H/V depth with borehole data was inconclusive at the three locations shown in Fig. 5, but still it provides a consistent estimation with geological (borehole) data.

Fig. 5
figure 5

Correlation between H/V curves result and geological data (borehole log) for the selected sites within the study area. Thickness (H) of soft sediments evaluated from H/V curve was 65, 19, and 11.24 m approximately (from top to bottom) by using Parolai’s equation

Full size image

Discussion

In the present study, alluvium thickness residing over the bedrock was estimated keeping in view the work done by Parolai et al. (2002) in Cologne area (Germany). The study conducted by Parolai et al. (2002) considered the estimation of the sediment thickness by equation H = 108f 0 –1.551 instead of H = 96f 0 –1.388. The later equation was used for the estimation of alluvium thickness during his earlier work in 2001 on the same Cologne area (Parolai et al. 2001). Both the equations showed use of different coefficients for estimating alluvium thickness. Since these coefficients were dependent on geological conditions therefore the results could be different with different geological settings. In the present study, we found that Attock Basin showed some in situ geological similarities with Cologne area and on the basis of which we used the above-mentioned coefficients in Parolai’s equation, i.e., H = 108f 0 –1.551. Attock basin showed the presence of thick recent deposits in the center, some of which may range close to 250 m. These recent sediments were mainly gravel sand and clay residing on Muree formation of Eocene time. On the other hand, Cologne area represented a similar type of quaternary deposit mainly gravel sand and clay resided on the Devonian bedrock (Parolai et al. 2002). In addition, the thickness of Attock basin is variable, i.e., 238.5 m in the center to approximately 0 m in south of the basin where basement rocks are exposed. Cologne area also represents a variable sediment thickness, i.e., from 0 m in the northeast-east of Cologene near to site of basement rock exposure and reached up to a depth of 300 m in west-southwest part of the study area (Parolai et al. 2002). Another important feature of the coefficients in Parolai’s equation used in the present study, i.e., H = 108f 0 –1.551, was the precise estimation of the sediment thickness over the bedrock; whereas use of the different coefficients could lead to discrepancy in greater depth (Parolai et al. 2002).

The soil vulnerability index, K g, was also evaluated keeping in view the work done by Warnana et al. (2011) in the present study using the equation K g = (A 0)2/f 0. The highest recorded value for soil vulnerability index (K g) from the study area was 24. ARC GIS software was used to interpolate and analyze the spatial extent of different parameters studied in the present article. Different shades, e.g., light gray, intermediate gray, and dark gray, were used to reflect spatial variations in local site parameters. Dark gray shades in the interpolated map (Figs. 6, 7, 8, and 9) are shown as higher values and light gray shades are shown as lower values of the described parameter.

Fig. 6
figure 6

Interpolated map showing Alluvium thickness (H) in meters in the Attock Basin

Full size image
Fig. 7
figure 7

Interpolated map showing fundamental frequency (f 0) in Hz in the Attock Basin

Full size image
Fig. 8
figure 8

Interpolated map showing H/V spectral amplitude (A 0) in the Attock Basin

Full size image
Fig. 9
figure 9

Interpolated map showing soil vulnerability index (K g) in the Attock Basin

Full size image

Geologically, the center of the basin is called the depocentre. It corresponds to the maximum thickness of the older sediments overlain by younger basin fill. Results from the microtremor study also supported the above-mentioned geological phenomena (Fig. 6) as the central parts of the Attock Basin showed maximum thicknesses, which was estimated from the lowest values of the fundamental frequency f 0 (Fig. 7). The amplification factor was also moderate to high within the central parts of the basin (Fig. 8). Soil vulnerability index K g was also higher in central parts of the basin (Fig. 9) and showed complete agreement to the findings in the study. Moreover, the geological map of the study area (Fig. 2) also showed the presence of the greater thickness of soft sediments in the central parts of the Attock Basin. Therefore, the microtremor data sets showed an excellent correlation with the geological principles and consistent correlation with other data sets in the study area.

Conclusions

In the Attock Basin, a total of 50 site responses in the Attock Basin were analyzed using the H/V spectral ratio technique due to its simple and cost-efficient nature. The fundamental frequency, amplification factor, soft sediment thickness, and soil vulnerability index were evaluated during the study. Fundamental frequency values range between 0.6 and 12.4 Hz. Most of the data show sharp H/V peaks corresponding to strong impedance contrasts and confirms that the susceptibility of amplification is also moderate to high. The study area exhibits a complex structure and an irregular spatial distribution pattern of overlying soft sedimentary deposits. The central part of the Attock Basin showed the lowest fundamental frequency, i.e., maximum sediment thickness, higher amplification factors, and higher soil vulnerability indices. Results were consistent with the borehole data to enhance the credibility of the study and revealed that a microtremor study was reliable to estimate the sediment thickness of the study area.