Introduction

Site characterization is one of the main requirements to understand the behaviour of seismic waves during strong motion excitation. It provides a major input to ascertain basin characteristics for hazard zonation and risk assessment studies. Over the years, the damage distribution during an earthquake reveals the significance of local geology and thickness of soft sediments. The extent of damage in the frontal part of the Himalaya, during the 2005 Kashmir earthquake (Mw 7.6) and 2015 Nepal earthquake (Mw 7.8) also demonstrated the deleterious effects of loose alluvial soil (Mahajan et al. 2006; Chan et al. 2017; Gautam 2017). Thus, estimating the thickness of soft sediments above bedrock and its impedance contrasts are of major concern in the frontal part of the Himalaya. The Kangra Valley is one such region in the Northwest Himalaya that had suffered major effects from the 4th April 1905 (Ms 7.8) Kangra earthquake, and according to the Middlemiss (1910) report, a varied degree of damage within the same intensity zone (intensity X on Rossi–Forel scale) was reported from different parts of the valley, which indicates the influence of the varied thickness of the sediment. The seismic zonation map of India reveal the region falls in seismic zone IV and V with very high seismic hazard potential (BIS 2002). According to the seismic hazard analysis, the Kangra region can have a peak ground acceleration (PGA) of 0.50 g with 10% probability of exceedance in 50 years (Mahajan et al. 2010). Thus, to understand the effects of near surface material on strong ground motions in such a seismically active region with varied lithology, complex geology and varied thickness of sediments, a detailed mapping of bedrock depth and thickness of soft soil is required. The aim of this study is to estimate the thickness of soft sediments using horizontal to vertical spectral ratio (HVSR or H/V) analysis and multichannel simulation with one receiver (MSOR) data modelling and thereafter to establish new empirical relationship to derive thickness of sediment above bedrock. Since the thickness of sediment is linked with the frequency and stiffness of the soil column, efforts were made to measure the fundamental/resonance frequency (f0) and shear wave velocity (Vs) of each soil column.

The resonance frequency of the soil column was derived using ambient noise measurements. The concept of recording ambient noise measurements was first proposed by Kanai (1961), Kanai and Tanaka (1961) and Nogoshi and Igarashi (1971) for spectral analysis, and later improved by Nakamura (1989, 2000) by normalizing the source effect through horizontal to vertical spectral ratio (HVSR). Presently, the technique is being used worldwide as a low-cost and effective tool for estimation of resonance frequency (f0) and amplification (A) characteristics, although the theoretical background of HVSR technique is debatable, while some studies have even demonstrated the good consistency between the estimates derived from ambient noise and earthquake records (Bindi et al. 2000; Fah et al. 2001; Chan et al. 2017). Many researchers have provided critical reviews on the application of HVSR technique (Al Yuncha and Luzon 2000; Bindi et al. 2000; Mucciarelli and Gallipoli 2001; Bonnefoy-Claudet et al. 2006; Lunedei and Malischewsky 2014). Parolai et al. (2002) and Tun et al. (2016) confirmed that the HVSR technique provides a good estimate of the fundamental frequency if there is clear impedance contrast between soft soil and bedrock.

The shear wave velocity (Vs) is another input required for subsurface characterization that can be calculated using different geotechnical and geophysical methods. Traditionally, geotechnical investigations are generally carried out using standard penetration test (SPT) and borehole investigations, to provide information on the subsurface material properties. However, for the present study area, these techniques seem to provide erroneous results, because the valley is underlain by fan sediments, mainly consisting of gravels, pebbles and boulders, which are unsuitable for performing standard penetration test. Therefore, detailed site characterization was planned using seismic data collection in active and passive mode. The passive mode of data acquisition records ground vibrations which are present in the environment (e.g. microseisms and microtremors), whereas the active mode acquires the seismic data using multichannel simulation with one receiver (MSOR) mode. The MSOR is a modified version of the multichannel analysis of surface waves (MASW). It uses an active excitation source and requires a single sensor and trigger for data collection rather than a set of geophones like that of MASW data acquisition (Harutoonian et al. 2013). Thus, during seismic data acquisition, a velocimeter fixed at a location of interest allows us to estimate resonance frequency using HVSR technique and the same instrument can be used to gather Rayleigh wave dispersion characteristics through the MSOR technique at the same observation point. Since the Kangra Valley has rugged topography, finding flat open ground at every location was difficult, which limits the possibility of conducting shear wave velocity investigation using MASW. So, simultaneously, MSOR was used on small ground surfaces for deriving one-dimensional (1-D) shear wave velocity profile using joint-fit-inversion modelling of HVSR and dispersion curve. Moreover, at sites where even minimal linear profiles of MSOR survey could not be performed, only resonance frequencies were measured using ambient noise measurements (Fig. 1).

Fig. 1
figure 1

Location map of the study area with respect to a India and b Himachal Pradesh, showing, c the sites of ambient noise measurements (200) and MSOR (85) data collection on the geo-tectonic map of Kangra Valley. BrT Barser back thrust, JMT Jawalamukhi thrust, MBT main boundary thrust, MT Murree thrust, PT Panjal thrust

Full size image

To obtain a one-dimensional (1-D) shear wave velocity (Vs) profile, base information on the thickness of near surface top soil layer is required, which can be achieved by different techniques like seismic reflection, multichannel simulation with one receiver (MSOR) and multichannel analysis of surface waves (MASW). The MSOR as well MASW techniques are not capable of providing information down to depths greater than 40 m; however, passive mode of data acquisition using HVSR can provide deeper information about the subsurface material from the joint-fit-inversion modelling of the HVSR curve. Finally, new empirical relationships have been established for the Kangra Valley between resonance frequency and thickness of sediment. This newly established relationship will also be helpful in estimating the thickness of those areas which have only information about resonance frequency due to the presence of forest cover or high hill slopes.

Geology and tectonics

The Kangra Valley is a piggyback intermontane basin, which is approximately 75 km long and 10–25 km wide (Ori and Friend 1984; Thakur et al. 2014) formed in the Outer Himalayan region of Northwest Himalaya (Fig. 1c). The valley to the north is bounded by the steep range of Dhauladhar Mountain and Lesser Himalayan rocks, whereas the southern part is characterized by geomorphic features like ridges, valleys and small mountain peaks of the Outer Himalayan range. The ridges are underlain by tertiary rocks (Sandstone and Upper Siwalik Boulder Conglomerates) or glaciated moraine deposits and overlain by alluvial sediments, whereas valleys are underlain by fan sediments of varied thickness. Tectonically, the valley is marked by the presence of major thrust systems due to the continuous northward movement of the Indian landmass. The valley is sandwiched between two major tectonic features, i.e. the Jawalamukhi thrust (JMT) to the south and the main boundary thrust (MBT) to the north (Thakur et al. 2014; Jayangondaperumal et al. 2017; Fig. 1). Steeply dipping faults also traverse the area, which extend from the Outer Himalayan to Lesser Himalayan zone. Many of these faults are conjugate wrench faults striking approximately NNE–SSW (Mahajan and Kumar 1994). The valley also represents a unique example of evolutionary history of Quaternary Fan deposits that pass through an interval of several climatic oscillations (glacial and interglacial; Sah and Srivastava 1992). The presence of active tectonic features provides an indication of major contributions from tectonically driven erosional processes (Dey et al. 2016). The continued erosion and deposition of sediments in the frontal part of Himalaya lead to the formation of fan deposit in the lowland area. The vertical lithological profile of the Kangra Fan area shows the dominance of gravels in litho-facies, which is further characterized by the presence of glacial moraines and clay-rich debris-flow deposits in the upper fan area, sandy fill deposits in the middle fan and sand, silt and mud deposits in the lower fan areas (Sah and Srivastava 1992). The Quaternary sediments (clay, gravel and boulder) present in the area can play a crucial role in strong motion excitation.

Methodology

The ambient noise measurements were recorded from 200 sites in the Kangra Valley to derive the resonance frequency (f0) (Fig. 1). Among these, 85 sites were covered using both the multichannel simulation with one receiver (MSOR) technique and ambient noise measurements and the HVSR curves derived from these sites can be modelled using joint-fit-inversion of HVSR and dispersion curve. Finally, the HVSR data and the thickness derived using inversion analysis were used to obtain a new empirical relationship between resonance frequency and thickness of the soil column.

Data acquisition and processing of ambient noise

Extensive ambient noise measurement survey was carried out covering the entire Kangra Valley within and along the fringes of the Kangra basin. The instrument used is a tromograph named “Tromino” (model ENGY), consisting of a three component electro-dynamic sensors having an operating frequency range of 0.1–1024 Hz. Better coupling of the instrument with ground was ensured by fixing the instrument using three long spikes. The instrument was set up inside a pit to avoid any interference from direct wind noise or footsteps, etc. (Fig. 2a). The data were acquired for 20 min with a sampling frequency of 128 Hz, considering the SESAME (Site EffectS assessment using AMbient Excitations) criteria (SESAME 2004). The acquired data was imported and analysed using ‘Grilla’ software (Fig. 2b-f). The ambient noise measurement records were analysed in 60 non-overlapping time windows of 20 s with different smoothing techniques, i.e. triangular window (TW), cosine window (CW), rectangular window (RW) and Konno and Omachi window (KO) with b value = 40 (Konno and Omachi 1998). The data processing was performed in the frequency range of 0.2–60 Hz, but only the peaks within or close to the 0.2–30 Hz were considered for the study purpose. Different smoothing techniques were attempted and finally the triangular window (TW) filtering was applied to all the data set, thus reducing the standard deviation (SD) of H/V spectral ratio for better discrimination of stratigraphic peaks. The HVSR analysis led to single or multiple frequency peaks and the frequency with maximum amplitude was marked as ‘f0’ and the corresponding amplitude as ‘A0’. The sites with secondary and tertiary peak also had distinctive frequencies which were denoted as ‘f1’ and ‘f2’ with amplitude ‘A1’ and ‘A2’, respectively (Fig. 3a). The peak with maximum amplitude (A0) was used to select the main resonance frequency (f0) of the site. The resonance frequency was selected in the frequency band of engineering interest (0.5–20 Hz), which, in Fig. 3a, falls within the dotted green rectangle, while the higher frequency range (> 20 Hz) in the red dotted rectangle represents a very shallow contrast and so generally excluded from the final analysis. The noisy transients were removed during final processing to get clear spectrum peaks that further helped to distinguish anthropic peaks (dirac like) from stratigraphic one (‘eye’ shaped structure in spectral amplitude curves; Fig. 3b).

Fig. 2
figure 2

Example of ambient noise measurement in passive mode; a field setup showing Tromino in a ~ 0.07 m deep pit; b time variation of the frequency component with rectangular frames marking the disturbance related with transient noise, c which was further removed from the HVSR analysis; d directional HVSR with colour showing the amplitude of HVSR spectra; e HVSR curve and f amplitude spectra of the three components

Full size image
Fig. 3
figure 3

Multiplicity of HVSR peak (site no. 139-II) with subsurface heterogeneities; a HVSR peak with distinctive frequency (f0, f1,) and amplitude (A0, A1); b corresponding stratigraphic peaks with ‘eye’ shaped structure in up–down component; c site condition showing subsurface heterogeneities (boulders) excavated from the area

Full size image

The removal of noisy transient from the time window (Fig. 2b, c) provides a clearer HVSR curve. After identifying the stratigraphic peaks, the corresponding H/V curves were checked using SESAME criteria that define conditions for clear and reliable identification of H/V peaks (SESAME 2004; Rezaei et al. 2013; Gabas et al. 2014; Rezaei and Choobbasti 2017; Mahajan and Kumar 2018). The H/V curves were also used to identify different layers with significant impedance contrast, in case of curves having more than one peak. From this check, the identification of peaks found at some of the sites showed less reliability, as per SESAME criteria, so such sites were excluded from the final analysis.

The HVSR technique has a lot of applications including identification of the resonance contrast between sediments and the bedrock comprising variable stratigraphic succession (Ansal 2004; Tarabusi and Caputo 2016; Gosar 2017; Mi et al. 2019). The HVSR peaks were used to select dominant velocity contrast between the overlying soft sediments and underlying stiff strata/bedrock. The analysis further reveals that the Kangra Valley is marked by different typologies of HVSR curves (i.e. broad, clear and flat peak). Clear HVSR peaks were identified in the central part of the study area, where sharp contrast is observed between the overlying soft sediment and the underlying stiff sediment (Fig. 4a). Broader HVSR peaks are observed in the regions that are characterized by sloping interface or shallow structures (i.e. the JMT) between softer and very stiff material (i.e. site no. 49; Fig. 4b). Multiple HVSR peaks were observed in the areas having moderate topographic relief that are characterized by multiple contrast. These multiple contrasts may be due to the material composition of the Dharamsala Group of rocks, which consists of claystone, mudstone and sandstone (Fig. 4c). Flat HVSR curves were observed from some of the isolated locations that are characterized by hard bedrock lithology, mainly Middle Siwalik Sandstone and Upper Siwalik Boulder Conglomerates (Fig. 4d; site no. 183).

Fig. 4
figure 4

Different typologies of HVSR curve; a clear, b broad, c multiple and d flat HVSR peaks in Kangra Valley

Full size image

The presence of fan sediment subsurface lithology made it difficult get a clear peak at low frequency, because of very low impedance contrast between the underlying bedrock and overlying fan sediments.

Multichannel simulation with one receiver (MSOR) data acquisition

The MASW technique deploys multiple receivers laid down along a linear array connected to a common recorder [analog-to-digital (A/D) convertor] named as engineering seismograph. The multichannel data acquisition requires a relatively long spread (100–150 m) on ground to penetrate deeper, as in general, the penetration depth is equal to half of the spread length, although it also depends on three factors, i.e. impact source energy, frequency of geophone and the stiffness of the material underneath (Mahajan 2009). In the frontal part of Himalaya, where large open grounds are not easily available, instead of using MASW, the MSOR mode of data acquisition was used to obtain information on shear wave velocity from the Rayleigh wave dispersion curve. Due to irregular topographic conditions, depth of bedrock varies from a few meters to tens of meters. The MSOR technique can be used on smaller area as shown by various authors (Ryden et al. 2004; Tokshi et al. 2013; Lin and Ashlock 2016). The technique relies on active excitation of seismic waves recorded by a single receiver and a trigger geophone, instead of relying on a set of geophones as per the MASW technique. In hilly terrains, the use of MSOR was found much more advantageous than MASW in term of speed, portability and space. The same tromograph has been used in active and passive mode for recording seismic waves using a combination of analysis methods for subsoil velocity modelling (Harutoonian et al. 2013). The impact source and trigger were moved with respect to the sensor in an incremental manner to cover a 1-D array of ~ 20–25 m on ground (Fig. 5). From each excitation, seismic traces were acquired and finally, combining all individual seismic traces, a simulated multichannel record was compiled for further processing (Fig. 6). The seismic traces were processed using ‘Grilla’ software to generate a dispersion image in the form of an overtone image (Fig. 6f).

Fig. 5
figure 5

Schematic diagram of field configuration for data acquisition using multichannel simulation with one receiver (MSOR) mode, showing location of sensor, trigger and source with respect to each other. The dotted geophone and plate show their next location while acquiring data in an incremental manner

Full size image
Fig. 6
figure 6

Processing steps of multichannel simulation with one receiver (MSOR) data using ‘Grilla’ software: a acquired raw data along different channels, b picking of first arrival, c picking of shot times with respect to trigger channel, d combining different multiple records in a single seismic record, e selecting an appropriate portion of signal and f generation of dispersion curve

Full size image

The dispersion curves obtained using MSOR can have fundamental and higher modes, whereas in MASW technique, mostly the fundamental mode is obtained and used for dispersion analysis. The overtone image of the dispersion curve represents three variables, i.e. phase velocity versus frequency, while colour represents the amplitude of the signal, which is a function of wavelength (λ), as the maximum depth (Zmax) for which shear wave velocity (Vs) can be reasonably calculated (Rix and Leipski 1991).

Combining the dispersion curves derived from MSOR and the H/V curves from ambient noise analysis, a joint-fit-inversion modelling was performed using the Occam’s razor principle that suggests to select the minimum number of model parameters, with a number of peaks (n) and layers (n + 1) to be used for a reliable result. The dispersion curve obtained from the MSOR technique provided the initial information on phase velocity and phase frequency. According to Kramer (1996), the fundamental frequency (f0) is a function of shear wave velocity (Vs) and thickness (H) of the soil column according to the relation:

$$f_{0} = {\text{ Vs}}/4H.$$
(1)

So using this equation, one can derive the thickness of the soil column if the shear wave velocity of that soil column is known, which could be derived by collecting the seismic data using either the MASW or the MSOR technique.

Results

Joint-fit-inversion modelling

To estimate the approximate depth of the first impedance contrast responsible for resonance in the frequency range of engineering interest (< 20 Hz), the phase velocity spectra derived using MSOR technique were used to provide constraints to the inverse modelling procedure of the HVSR curve. This modelling was based upon the assumption of plane parallel stratigraphy, so the presence of lateral heterogeneities can lead to incorrect interpretations of phase velocity spectra. The first HVSR peak below 20 Hz was modelled using the information of thickness and shear wave velocity from the dispersion curve. The second HVSR peak was modelled with the support of the same phase velocity spectra. Further, low frequency HVSR peaks of the same curve were further modelled using trial and error method. During the inversion procedure, the synthetic HVSR curve obtained from initial modelling was matched with the experimental average HVSR curve (e.g. Castellaro and Mulargia 2009; Roser and Gosar 2010). The modelling of the higher mode were taken into consideration in estimating the thickness of shallow layer. The best fitting criterion between experimental and synthetic HVSR curves was used in inversion analysis after taking base information of shear wave velocity of the soil type from the dispersion curve (Fig. 7a, b).

Fig. 7
figure 7

Diagrams showing the outcome of the joint-fit-inversion modelling for three different sites, with modelled HVSR curves (a, d, g), MSOR driven dispersion curves (b, e, h) and correspondingly derived 1-D shear wave velocity profiles (c, f, i). Red dotted lines mark maximum penetration depths deduced by the MSOR survey

Full size image

In the present study, the medium frequency HVSR peaks (3–10 Hz) are dominant, while the HVSR curves sometime become more chaotic and irregular towards the ends. At few locations, the HVSR peaks were almost flat maybe because of the presence of material properties and near surface bedrock which bring impedance contrast very low.

Figure 7a shows the H/V curve of the site located in the extreme southern part of the valley with HVSR peak at 6.81 Hz (site no. 36). The dispersion curve obtained from the same site using MSOR survey shows a phase velocity (c) of 430 m/s corresponding to phase frequency (f) of 8.0 Hz. Since, the penetration depth (Zmax) is half of the wavelength (λ), where λ = c/f, i.e. 430 m/s/8.0 Hz = 53.75 m, thus the penetration depth achieved is ~ 27 m (Fig. 7b). In Fig. 7, the frequency band 8–14 Hz corresponds to the higher mode and the input from array and HVSR were used in the joint-fit-inversion modelling. The first predominant HVSR peak provides the shallower contrast at a depth of 6 m, which is in good agreement with the shear wave velocity (Vs) profile obtained from MSOR data. The maximum penetration depth deduced by the MSOR survey is marked by red dotted line (Fig. 7c), whereas the velocity model was extended by inverting the low frequency peaks of the HVSR curve (Castellaro and Mulargia 2009; Castellaro 2016; Gupta et al. 2019). Another site located in the central part of the valley shows the HVSR peak at 4.97 Hz (site no. 103; Fig. 7d), and the maximum depth of penetration achieved using MSOR survey is estimated to be ~ 28 m, i.e. resulting from c = 386 m/s and f = 7.0 Hz (Fig. 7e, f). The joint-fit-inversion modelling of HVSR and dispersion curves provide shallow depth information (Fig. 7f). The third site (site no. 144), from the northern part of Kangra Valley, shows the main peak was at 11.88 Hz (Fig. 7g). The analysis of dispersion curve allowed to achieve a maximum depth penetration of 30 m (Fig. 7h), whereas the inversion of HVSR low frequency peaks allowed to investigate up to 75 m depth (Fig. 7i).

New empirical relationships between f0 and H

The Kangra Valley is highly rugged and criss-crossed by southern flowing drainages due to the presence of ridges and valleys. The valley is underlain by fan sediments, whereas the ridges are underlain by either tertiary sedimentary rocks or moraine (glacial) deposits. Under such conditions covering all sites using MASW or MSOR was not feasible due to non-availability of space for longer linear profile. However, conducting ambient noise measurement was possible even in such locations. In such area, site characterization has been achieved by establishing an empirical relationship between fundamental frequency and pseudo-depth (H), defined as the empirically estimated depth of the main impedance contrast. A number of empirical relationships between fundamental frequency and pseudo-depth were previously established for different regions of the world by various authors, i.e.:

$$H = \, 96f_{0}^{ - 1.388} ,$$
(2)

for the Lower Rhine Embayment in Germany (Ibs-von Seht and Wohlenberg 1999)

$$H = \, 108f_{0}^{ - 1.551} ,$$
(3)

for the Cologne area—Germany (Parolai et al. 2002)

$$H = \, 105.99f_{0}^{ - 1.1531},$$
(4)

for the coast South of Istanbul (Birgoren et al. 2009)

$$H = \, 141f_{0}^{ - 1.27} ,$$
(5)

for the Izmit Bay area—Turkey (Ozalaybey et al. 2011)

$$H = \, 146.01f_{0}^{ - 1.2079} ,$$
(6)

for the Kathmandu Valley—Nepal (Paudyal et al. 2013).

Most of these equations were obtained using rigorous geophysical investigations and their comparison with borehole record, whereas for the present study area, limited information was available from sparse borehole records, so the thickness of each layer and its shear wave velocity was derived using surface wave analysis and joint-fit-inversion modelling. Thus, to calculate the sediment thicknesses, the use of area specific shear wave velocity and thickness have been used for reliable results. In the present study, the thickness of sediments derived using joint-fit-inversion modelling of HVSR and dispersion curve was plotted against the fundamental frequency (f0) of 85 sites for which HVSR and MSOR data were available (Fig. 8).

Fig. 8
figure 8

a First order pseudo-depth with respect to the main contrast (black numbers) and the respective shear wave velocity (red numbers) from the modelled HVSR peak. b Plot of thickness versus resonance frequency derived from joint-fit-inversion and ambient noise measurements, respectively; solid line shows the non-linear fit to derive an empirical relationship between fundamental frequency and experimental pseudo-depth (H) as Eq. 7

Full size image

Although due to physiographic constraints, the MSOR data points are not homogeneously distributed, however the HVSR and MSOR measurements from 85 sites gave enough data to derive a new empirical relationship for the study area through a non-linear regression analysis between thickness and resonance frequency. The fundamental frequency derived for 85 sites was plotted against thickness of sediment and the resulting equation is

$$H \, ({\text{MSOR}}) \, = \, 183.13f_{0}^{ - 1.542} .$$
(7)

The thickness of sediment at a particular site along with their shear wave velocity with respect to the main contrast is shown in Fig. 8a, whereas the Fig. 8b shows the non-linear regression fit between fundamental frequency and sediment thicknesses. The final equation derived from experimental data was used for calculating the thickness of sediment of other sites, where it was not possible to acquire data with MSOR as well MASW. Thus, it has been concluded that in the absence of any shear wave velocity data, thickness of sediment can be derived using fundamental frequency of that site and applying the same in Eq. 7. Further to compare our results the present study data has been plotted in double logarithmic coordinates as shown in Fig. 9 using different empirical relationship of the world. The empirical relationship for this study between fundamental frequency and depth shows good fitting with the derived regression curve (R2 = 0.868). In a double logarithmic plot, the slope is represented by ‘b’ value, whereas ‘a’ value represents the y intercept. The ‘b’ value of Kangra Valley is found to be significantly higher than in other regions of the world except Germany (Ibs-von Seht and Wohlenberg 1999; Parolai et al. 2002). The higher ‘b’ value of the study region could be due to variability in the thickness of sediment and the shear wave velocity parameters. Actually, in several studies the resulting b values were found close to − 1, but for the Kangra Valley (and the Cologne zone in Germany) ‘b’ is around − 1.5, which implies that thickness decreases with the increase of the resonance frequency more rapidly than what expected for horizontal homogeneous layers. This can be due to the fact that local conditions can differ from those of a soft horizontal homogeneous layer overlying a stiff bedrock, thus causing resonance frequencies modified by 2D/3D effects and/or lateral variations of Vs. On the basis of shear wave velocity, in Kangra Valley three types of lithology can be distinguished, namely, the bedrock having shear wave velocity of the order of > 760 m/s, moraine sediments having an average shear wave velocity of the order of 450–500 m/s, fan sediments and alluvial soil having an average shear wave velocity of the order of 200–300 m/s (Mahajan and Kumar 2018).

Fig. 9
figure 9

Different fitting curves and the obtained empirical equation for the Kangra Valley

Full size image

Based on Eq. 7, the spatial distribution of thickness value in the study area is shown in Fig. 10. According to this map the thickness of the alluvial sediments is very thin the northern part of the basin and are underlain by hard rock (sandstone or Upper Siwalik Boulder Conglomerate), thus the impedance contrast is very high in such area especially in the northern part of the valley. The north-central and central part of the basin have thick sedimentary fans but the constituents of these sediments are as that of Upper Siwalik Boulder Conglomerate, thus having a low velocity contrast between the overlying layer and underlying hard rock. Therefore, HVSR peak frequency (f0) of these sites have almost flat HVSR curve in contrast to the northern and south-western area which is characterized by hard rock underlying alluvial sediments. Some of the sites are directly exposed on the surface or have very-very thin alluvial cover, thus it has found that the sediment thickness estimated using Eq. 7 of such sites is found to be overestimated (encircled in Fig. 10) as compared to other sites of the basin.

Fig. 10
figure 10

First order pseudo-depth derived using new empirical relationship for the Kangra Valley

Full size image

Discussion and conclusion

Although, Kangra Valley had experienced a devastated effect during 1905 Kangra earthquake (Ms 7.8; Ambraseys and Douglas 2004), since then most of the frontal part of Himalayan region has witnessed unplanned urbanization which is prone to site amplification. The reported variation in the damage within the valley and on the fringes of the basin and upcoming non-resistant construction activity in the region prompted us to carry out site characterization studies. Since the properties of the near surface material are required for estimating site amplification parameters of the Kangra Valley, the present study was planned to give insight into the needs of seismic resistant structures to reduce the impact of future earthquakes. Until today, most of the Himalayan frontal belt is marked under seismic zone IV and V, based on the damage distribution of previous earthquakes. There are no data on the thickness of the soil column and its shear wave velocity by virtue of which one can calculate the seismic parameters for the site response analysis. Thus, efforts were made to estimate the shear wave velocity and thickness of each site using seismic method. A new empirical relationship has also been established for the Kangra region to cover those sites where topographic and physiographic hurdles prevent long seismic profiling. The study revealed that the applications of joint-fit-inversion modelling can be very useful in estimating the shear wave velocity parameters of different sites in the frontal part of the Himalayan region, because it needs much less space and is cheaper and faster as compared to other seismic methods like MASW. To estimate the thickness of the soil column above bedrock, seismic data were collected in active and passive mode.

The HVSR analysis of ambient noise from passive records provided different typologies of HVSR peaks which enabled us to infer the presence of subsurface heterogeneities. The central part of the valley has clear peaks indicating sharp contrast between the soft alluvium and hard substratum, whereas multiple peaks were mainly observed from northern and southern fringes of the basin and along the fault zones, reflecting lithological complexities underneath. The multiplicity of HVSR peaks is inferred to be related with the presence of more levels of impedance contrast and implies a greater damage potential because different frequency peaks at the same site may cause amplifications at multiple frequencies during strong motion excitation. The same phenomenon has also been observed in the Kathmandu Valley by Paudyal et al. (2012). Broad peaks were observed from the sites with sloping interface between different strata (SESAME 2004). Most of the sites in the Kangra Valley are mainly characterized by the presence of fan sediments (boulder, cobbles and sand with clay as matrix material) underlain by Upper Siwalik Boulder Conglomerates or Middle and Lower Siwalik Sandstone or Dharamsala Group (sandstone, mudstone and claystone). Hence, the material composition of the overburden material is almost the same except the variation in their compaction, because most of the alluvial material is a product of weathering from the same rock. Thus, at places where the underneath bedrock and the overlying material have similar composition, the HVSR curve is found to present low spectral ratios and flat HVSR curve, reflecting very poor contrast (Tun et al. 2016; Mahajan and Kumar 2018).

Since, the HVSR analysis only provides the fundamental frequency of each site, a multichannel simulation with one receiver (MSOR) was performed in active mode at some of the sites where the HVSR data were obtained. The active mode allowed us to penetrate up to bedrock with joint-fit-inversion modelling of HVSR and dispersion curve. These multiple simulations along a linear profile allowed to drive a 1-D shear velocity profile to understand the thickness of the soil column above bedrock, which was validated from the MASW results obtained by Mahajan and Kumar (2018). This modelling not only provides 1-D shear wave velocity profile but also provides depth with respect to the main impedance contrast to deduce new empirical relationship from non-linear regression analysis.

Since, the Kangra Valley is highly rugged and it is often impossible to deploy linear profiles using MSOR technique or any other seismic reflection method, because most of the sites are inaccessible, ambient noise measurements were used to determine overburden characteristics from the fundamental frequency of each site. To determine the thickness of the surface layer at such sites, the new empirical relationship derived from experimental data by taking input from joint-fit-inversion modelling, i.e. H(MSOR) = 183.13f0−1.542 (Eq. 7) between frequency and thickness derived from MSOR data, were utilized to derive sediment thickness from 200 sites. This relationship can also be used for estimating the thickness of sediment above bedrock where only resonance frequency is available. Thus, the present study will be very useful in conducting site characterization studies in other parts of the frontal Himalayan region and areas with young sedimentary basin.