Вы находитесь на странице: 1из 12

Tunnelling and Underground Space Technology 86 (2019) 10–21

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology


journal homepage: www.elsevier.com/locate/tust

Underground survey to locate weathered bedrock depth using noninvasive T


microtremor measurements in Jurong sedimentary formation, Singapore

Palanidoss Subramaniam, Yunhuo Zhang, Taeseo Ku
Department of Civil and Environmental Engineering, National University of Singapore, 1 Engineering Drive 2, Singapore 117576, Singapore

A R T I C LE I N FO A B S T R A C T

Keywords: Determination of bedrock depth is an essential part of design/alignment/initial planning of any infrastructure
Bedrock detection projects such as tunneling, foundations and underground utilities. Drilling a borehole to identify the depth of
HVSR bedrock is often time consuming, costly and cumbersome process. In the present study, microtremor methods are
Microtremor measurement used to identify the depth of weathered bedrock. Series of ambient microtremor array measurements (MAM) and
Shear wave velocity
3-component microtremor methods are conducted in the weathered rock of Jurong formation in Singapore.
Surface wave
Based on these surface wave measurements and available reference borelog information, three different ap-
proaches – (1) shear wave velocity (Vs) based approach; (2) wavelength based approach and (3) Horizontal to
Vertical Spectral Ratio (HVSR) approach – are proposed to identify the depth of bedrock in the Jurong formation.
For the Vs based approach, a specific reference value of Vs is suggested to estimate the depth of bedrock in the
Jurong formation, which is readily determined by the Vs profile. In the wavelength based approach, an ex-
perimental dispersion curve in the form of wavelength and phase velocity is found able to identify a deviation
point that can be used for predicting the corresponding bedrock depth. A relationship between the deviation
point in wavelength and bedrock depth has been established. In the HVSR based approach, an empirical cor-
relation model is formed between the natural frequency obtained from the HVSR profile and the bedrock depth
information. To conclude, all three approaches can provide overall reasonable estimates in the depth of bedrock
although the Vs based approach is slightly better in statistics, followed by HVSR based approach. In practice, the
HVSR approach would be easy to implement in terms of field testing and analysis.

1. Introduction et al. (2017) used Multichannel Analysis of Surface Waves (MASW)


(Park et al., 1999), Refraction Microtremor (ReMi) (Louie, 2001) and
Identifying the depth of bedrock is very important in planning ideal SPatial Auto Correlation (SPAC) (Aki, 1957) methods to estimate the
utilization of underground space, which involves determining founda- average shear wave velocity of 30 m depth (Vs30) and to determine the
tion system, alignment of underground tunnels, TBM selection and/or depth of engineering and seismic bedrock in the Izmir Inner bay area in
excavation planning, as well as an appropriate design of buried drai- Turkey. They have used an equal thickness layered approach in the
nage network. Conventional vertical boring is commonly carried out to inversion of dispersion curve where each soil/rock layer is assumed to
locate the depth of bedrock. However, drilling a borehole to locate the have an equal thickness. Similarly, Ridwan et al. (2015) conducted
bedrock depth in a large infrastructure project is often a time con- microtremor array measurements using a triangular array to locate the
suming process. In addition, the cost of drilling a borehole is very ex- engineering bedrock in Jakarta where the bedrock depth varies be-
pensive and subjected to environmental concern which involves cutting tween 353 and 662 m. Akgün et al. (2013) used microtremor and
of plants and trees for the transportation of drilling equipment. MASW methods to find ground parameters such as Vs30, bedrock depth,
Near surface geophysics are widely used in various infrastructure, attenuation and number of layers for ground response analysis. Most of
agricultural, mining, environmental and hydrological projects. Various these studies were conducted in the view of seismic microzonation.
types of near surface geophysical methods and their applications in Coulouma et al. (2013) investigated the efficacy of Electrical Resistivity
geotechnical engineering are listed in Anderson et al. (2008). Apart Tomography (ERT) (Shima and Sakayama, 1987), diachronic ERT and
from these, microtremor methods have been also used in the health Spectral Analysis of Surface wave (SASW) (Nazarian and Stokoe, 1984)
assessment of various structural components (Gao et al., 2016). Pamuk methods to determine the depth of bedrock. However, the intended


Corresponding author.
E-mail addresses: ceesubp@nus.edu.sg (P. Subramaniam), zhangyunhuo@u.nus.edu (Y. Zhang), ceekt@nus.edu.sg (T. Ku).

https://doi.org/10.1016/j.tust.2019.01.005
Received 29 May 2018; Received in revised form 21 December 2018; Accepted 9 January 2019
0886-7798/ © 2019 Elsevier Ltd. All rights reserved.
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

depth of investigation was less than 5 m and it was found that ERT is folds and faults, the rock type, layering and degree of weathering vary
applicable only when high contrast in electrical resistivity is expected widely, both vertically and laterally. Based on the British weathered
between soil and rock. Tamrakar and Luke (2013) proposed a disper- rock classification system (BS 5930, 1999), the rock can be classified as
sion curve based approach to find the depth of bedrock. However, the S(I) to S(VI) where S(I) is fresh intact rock and S(VI) is residual soil
depth of investigation was less than 10 m and based on active MASW respectively. The most commonly found geomaterials in Jurong for-
testing. They concluded that the approach is only applicable when a mations are (i) residual soil (gravel to clay in various proportions), (ii)
sharp bend in the dispersion curve is visible i.e., clear contrast between limestone, (iii) sandstone, (iv) siltstone, and (v) mudstone. The SPT
soil and rock layer. Moon et al. (2017) suggested a normalized phase refusal depth in particular testing location depends upon the type of
velocity approach and an average Vs30 based approach to find the depth rock and degree of weathering. In this study, the subsurface information
of weathered Bukit Timah Granite in Singapore. These methods were of all the testing locations are inferred based on SPT testing and rock
based on combined active and passive MASW testing conducted at one coring data.
location and validated with a narrow bedrock range (28.2–35.6 m). In
contrast to array measurements, few studies have also been conducted 2.2. Field surface wave measurement
using HVSR (horizontal to vertical spectral ratio) to find the bedrock
depth. In this method, a fundamental natural frequency of the site is Microtermor Array Measurement (MAM) was carried out using
correlated with the known bedrock depth (Yilar et al., 2017; Trevisani 4.5 Hz vertical geophones in L shaped/angular shaped array and the
et al., 2017; Parolai et al., 2002; Seht and Wohlenberg, 1999). From the geophones were connected through geophone cables. Ambient micro-
literature, it is evident that the past studies were based on either active tremors were recorded for 40 min at the sampling frequency of 2 ms at
testing (< 10–15 m) or seismic microtremor measurements to identify each testing location using a 24 channel seismograph. In order to cal-
engineered/seismic bedrock located at deeper depths (> 100 m). culate the horizontal to vertical spectral ratio, 3-component geophones
In the present study, the depth of weathered bedrock (weathering (2 horizontal - HEW, HNS; 1 vertical) were used. Since all the tested sites
grade - SIV/SIII; According to BS 5930 (1999), the weathering grade were of natural ground, spike type geophones were used and the ver-
SIV is highly to very highly fractured core with the rock quality des- ticality of the geophones were ensured through bubble level attached in
ignation (RQD) value of 0 whereas SIII is classified as weakened & the geophones. Totally eleven MAM tests at eight different locations in
discolored large pieces with RQD > 0) is estimated based on passive the Jurong formations with L- shaped/angular shaped array were
surface wave testing. Conducting active MASW testing is often un- conducted. The array shape, size and number of geophones were
favorable in the urban environment like Singapore due to the inter- decided based on the site conditions such as ground clearance, obstacles
ference of noises generated by heavy traffic and anthropological ac- and availability of area. It should be noted that the depth of in-
tivities. Due to the space constraints and environmental concerns, it vestigation is approximately equal to the size of the array. The in-
would be also not possible to plan for farther source (shot) location and formation on testing locations and geophone array (shape, size and
use a higher energy impact source to identify the deeper bedrock number of geophones) are listed in Table 1. The sites where larger ar-
(> 15 m) depth. In such situations, Microtremor Array Measurements rays are not possible to form, centerless circular array method (Cho
(MAM) is more suitable to locate the bedrock in which passive ambient et al., 2006) or miniature array analysis of microtremors (Cho et al.,
noise (e.g., traffic noise) is used as a source. In addition to this, in the 2013) shall be tried to construct the dispersion curve.
case of weathered rock, locating the rock depth is more challenging Fig. 2(a–g) illustrates the overall process involved in the generations
where the impedance contrast between the soil and rock is not sig- of Vs profile and HVSR plot. The acquired signals were checked for data
nificantly high enough to identify the bedrock depth from the layered consistency and the inconsistent channels were removed. For the mi-
shear wave velocity profile. With this view, series of microtremor array crotremor array data in frequency domain, dispersion image (frequency
measurements (MAM) and 3-component geophone measurements were versus Rayleigh wave velocity) was constructed by Extended Spatial
taken in various locations of Jurong formations in Singapore. However, Auto-Correlation method (ESPAC) (Ling and Okada, 1993; Okada,
noise sources within the MAM array/closer to the one leg of the array 1994). ESPAC is originally derived from the Spatial Auto-Correlation
need to be filtered out. The method is also not advisable for very (SPAC) algorithm proposed by Aki (1957). In the urban environment
shallow bedrock depths (< 8 m). The depth of bedrock in the Jurong like Singapore, it is often not possible to construct a circular array with
formation varies between few meters to tenths of meters and the Jurong a central geophone. Unlike SPAC, ESPAC considers frequency as a
weathered rock formation consists of sandstone, siltstone, limestone constant term instead of radius in the calculation of apparent phase
and mudstone. Several important underground construction projects velocity (Ohori et al., 2002) and this eliminates constraint in forming
are ongoing and planned to expand the present underground infra- the L-shaped or angular shaped array. Moreover, the number of geo-
structure network such as deep tunnel sewerage system, Jurong rock phones to be used in the circular array in order to calculate the SPAC
caverns, and Jurong region line. In terms of ways of inferring bedrock coefficient is not known clearly (Okada, 2006). The derivation of the
depth, three different approaches are explored to estimate the depth of ESPAC method is explained in detail by Parolai et al. (2006).
weathered bedrock (SIV or SIII) and Standard Penetration Test (SPT) Once the dispersion image is generated from the ESPAC velocity
refusal depth for the Jurong formation. spectrum, a theoretical dispersion curve is calculated based on the
forward modelling proposed by Lai and Rix (1998). The generation of
2. Experimental program theoretical dispersion curve and inversion were performed through
winMASW® software. This matlab based environment uses a genetic
2.1. Testing location algorithm (GA) based optimization code to perform inversion (Dal
Moro et al., 2007). The number of populations and generations in the
In the present study, MAM and 3-component geophone measure- genetic algorithm are adjusted in such a way to achieve the best fit in
ment methods were carried out at eight different locations of Jurong the experimental and theoretical dispersion curves. Fig. 3(a–e) depicts
formations in Singapore. The testing locations are shown in Fig. 1 along the various process involved in the generation of shear wave velocity
with the various geological formations of Singapore (modified map profile corresponding to site #4. Fig. 3(a) shows the dispersion image
from Lee and Zhou, 2009). Jurong formations are basically sedimentary together with the experimental dispersion curves and theoretical dis-
rock spread over west and south west part of Singapore. The Jurong persion curve (i.e., associated to initial model). It is assumed that the
formation’s origin dated back to late Triassic (235 myr) and late Jur- shear wave velocity increases with depth. Initial guess of the shear
assic (175 myr) (Zhao et al., 1999). Jurong formation predominantly wave velocity model can be constructed approximately by using the
consists of limestone, sandstone, siltstone and mudstone. Due to heavy following relationship (Brown et al., 2000):

11
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

6
9 2
5
4 3

1
7
8

Fig. 1. Testing locations in the Jurong formation (shown in green) and geological map of Singapore (inset) (modified from Lee and Zhou 2009). (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1
Testing program of microtremor array measurement (MAM).
#. Testing Location Reference Borehole IDa Testing methods Array size Geophone base (Number of Geophone)

1. Jurong Central Park 1.1 Passive 40 m × 40 m Spike (11)


2. Jalan Boonlay 2.1 Passive 40 m × 40 m Spike (11)
2.2 Passive 40 m × 40 m Spike (11)
3. Bukit Batok 3.1 Passive 40 m × 40 m Spike (11)
3.2 Passive 28 m × 20 m Spike (14)
4. Jurong west st 82 4 Passive 36 m × 36 m Spike (19)
5. PIE Road 5.1 Passive 36 m × 36 m Spike (19)
6. Nanyang Technological University (NTU site 1) 6.1 Passive 36 m × 36 m Spike (19)
6.2 Passive 40 m × 28 m Spike (18)
7. Jurong Town Hall Road 7 Passive 40 m × 40 m Spike (21)
8. West Coast Road 8 Passive 32 m × 34 m Spike (17)

a
Testing at the same site but in different reference borehole is indicated in decimal.

12
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

(a) (b)
Battery 10 9 8 7 6 Geophone Number
5 0 1 2 3 4 5 6 7 8 9 10
0
Geophones 4 5
3

Time (sec)
10
2 15
1 20
Laptop for 0 25
data logging 30
Seismograph

(d) (c)
Shear wave velocity (m/s)
0 200 400 600 800

Phase Velocity (m/s)


0
700

20 500
Depth (m)

300
40 100
0 2 4 6 8 10 12 8 10 12
Frequency (Hz)
60

(e) Battery (f) (g)


E-W V N-S
3 component 0
Geophone 5
8
Time (sec)

10 6
15
HVSR

4
20 2
25 0
Laptop for 1 10
30 Natural Frequency (Hz)
data logging
Seismograph

Fig. 2. MAM – (a) experimental setup used for MAM testing, (b) typical records from microtremor array, (c) dispersion image, (d) shear wave velocity profile after
inversion. MM – (e) experimental setup used to measure 3-components of microtremor, (f) typical records from 3-component microtremor, (g) HVSR profile.

Vs = 1.1VR at z =
λ find the depth of medium to hard weathered bedrock directly from the
3 (1) Vs profile due to the poor impedance between soil and rock. Herein,
three different approaches are proposed to estimate the depth of bed-
where VR and Vs are the Rayleigh wave phase velocity and shear wave
rock at Jurong formation.
velocity respectively. z and λ denote the depth of layer and corre-
sponding wavelength respectively.
(i) Preselected Vs based approach
The initial shear wave velocity model and the search space for the
(ii) Wavelength based approach
best model are shown shaded green1 in Fig. 3(b). The calculated misfit
(iii) HVSR approach
at each generation is shown in Fig. 3(c). The Figure also shows the
misfit variation of the fittest model (i.e., best model) among all the
The first two methods are based on MAM while the third is using a
generated models in the framework of genetic algorithm within the
HVSR profile. Fig. 4 shows a flowchart that illustrates the process of
search space. The dispersion curves corresponding to the fittest model
detecting the bedrock based on all the three approaches. The accuracy
(i.e., best model) is shown in Fig. 3(d) and the mean model implies the
in the bedrock detection of all the tested sites based on these proposed
average of all the generated models in the search space. The shear wave
approaches are discussed in the next section.
velocity (Vs) profile corresponding to the best generated model
(Fig. 3(e)) in the GA is further used to find the depth of weathered rock
(SIV or SIII). The depth of bedrock at each testing location and the type 3.1. Preselected VS based approach
of bedrock are shown in Table 2.
This method is proposed based on the assumption that the variation
in the Vs of weathered sedimentary rock in Jurong formation is within a
3. Methodology
narrow band. Using the Vs profile generated from the MAM testing and
the known borelog information, a reference Vs is suggested to locate the
In the case of weathered bedrock, it may not be straightforward to
bedrock, which is determined based on the data collected from all the
test sites in Jurong formation. Fig. 5 illustrates a typical procedure to
1
For interpretation of color in Fig. 3, the reader is referred to the web version find the depth of bedrock. A horizontal line is drawn at the bedrock
of this article. depth (shown in borelog) to Vs profile, which allows to measure the Vs

13
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

(a) (b)
900 0
Experimental dispersion curve
Initial model

Phase Velocity (m/s)


15

Depth (m)
600

30

300

45
200 400 600 800
4 8 12 16 20 24
Shear wave velocity (m/s)
Frequency (Hz)

(c) -2

-3

-4
Misfit

-5

-6

-7
0 20 40 60 80 100
Generation

(d) (e)
Experimental dispersion curve Shear wave velocity (m/s)
700 Best model 0 400 800 1200
Mean model
600 0
Phase Velocity (m/s)

500

400
20
Depth (m)

300

200

100 40
4 8 12 16 20
Frequency (Hz)

60
Fig. 3. (a) Experimental and initial model dispersion curves (b) initial shear wave velocity model in the shaded search space (c) calculated misfit for fittest model and
average of all the considered models (d) dispersion curves for best model, mean model (considering all the models) together with experimental values (e) shear wave
velocity profile corresponding to the best model.

at the bedrock depth. In this case, the depth of bedrock is located at 9 m drawn from the abscissa (500 m/s) to Vs profile, then a horizontal line is
and the corresponding Vs is about 420 m/s. Similarly, shear wave ve- drawn from the intersection point at Vs profile to ordinate (depth). This
locities at engineering bedrock depths of all the testing sites in the implies the depth of bedrock (12 m) estimated using the preselected Vs.
Jurong formation are examined. From this, as a reasonable bedrock
velocity for all the testing sites, a reference Vs of bedrock for Jurong
formation is suggested. This reference Vs can be applied to identify the 3.2. Wavelength based approach
bedrock depth directly from the shear wave velocity profile. In this
study, after careful reviews on all the testing sites at Jurong formation, The experimental Rayleigh wave dispersion curve, calculated based
the reference Vs is identified as 500 m/s. In Fig. 5, a vertical line is on ESPAC approach, is further analyzed to examine the possibility of
identifying the bedrock depth. Tamrakar and Luke (2013) developed

14
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

Table 2
Type of bedrock at various testing locations in Jurong formation.
Testing Location Reference Borehole ID Bedrock depth (m) Grade Type of bedrock

Jurong Central Park 1.1 27.5 SIII Limestone (27.5–37.5 m)


Alternative layers of SIV and Siltstone (37.5–41 m)
SIII
Jalan Boonlay 2.1 29 SIII Sandstone (29–34.8 m)
Limestone (34.8–40 m)
Siltstone (40–45 m)
2.2 34 SIII Sandstone (34–37 m)
Limestone (37–44 m)
Siltstone (44–45 m)
Bukit Batok 3.1 9 Alternative layers of SIV and Sandstone (9–40 m)
SIII
3.2 16 SIII Sandstone (16–27.5 m)
Alternative layers of SIV and Sandstone (27.5–40 m)
SIII
Jurong west st 82 4 21 SIV Siltstone (21–23 m)
Sandstone (23–26 m)
SIII Siltstone (26–40 m)
PIE Road 5.1 11 SIV Siltstone (11–30 m)
SIII Siltstone (30–31 m)
SIV Siltstone (31–40 m)
Nanyang Technological University (NTU site 1) 6.1 12 SIV Sandstone (12–15 m)
SV Gravelly sand (15–27 m)
SIV Siltstone (27–30 m)
SIII Mudstone (30–31 m)
SIII Sandstone (31–40 m)
6.2 17 SIII Sandstone (17–54 m)
Jurong Town Hall Road 7 19 SIII Sandstone (19–25 m)
Alternative layers of SIV and Sandstone (25–35 m)
SIII
SIII Siltstone (35–40 m)
West Coast Road 8 19.5 Alternative layers of SIV and Sandstone (19.5–23.5 m)
SIII
Alternative layers of SIV and Alternative layers of sandstone and siltstone
SIII (23.5–40 m)

Converting into Spatially auto


MAM test Generating
frequency correlation
dispersion curve
domain (ESAC)

Preselected wave Shear wave Phase velocity - Change point in


Inversion
velocity velocity wavelength plot wavelength

Bedrock
estimation
3 Component
H/V Natural
(H, V) Data
Spectrum Frequency
acquisition
Fig. 4. Methodology to estimate the depth of bedrock.

series of theoretical dispersion curves by varying the depth to bedrock estuarine layer (∼3 m) which is not seen in the rest of the sites. Though
between 2 m and 10 m in order to study how it affects the variation in boreholes #3.1 and #3.2 are 100 m apart, the dispersion curve corre-
the wavelength. In the view of active MASW testing and for very sponding to #3.2 shows a higher phase velocity profile. This may be
shallow bedrock sites (< 10 m), an apparent relationship was found due to dominant higher mode caused by closer heavy road traffic. The
between the bending point of wavelength and the bedrock depth. On dispersion curves are further converted into ‘wavelength vs frequency’
contrast to this, in the present approach, the dispersion curves are plots (Fig. 6(b)) through the following relationship (Eq. (2)).
generated from microtremor array measurement (i.e., passive MASW)
in the weathered rock sites. The depth of bedrock varies between 9 m Phase velocity = Frequency × Wavelength (2)
and 34 m. Fig. 6(a) shows the experimental dispersion curves of all the
Each phase velocity point is divided by the corresponding frequency
testing sites in the Jurong formation. It can be seen that at most of the
in order to obtain the wavelength and is plotted against the frequency.
sites, the dispersion curves fall in the close band. However, boreholes
Although the bedrock depths of all the testing sites vary, wavelength
#1 and #8 show relatively lower phase velocity profiles. From the
curves fall in very narrow band. However, it is observed that the
borelog information at these locations, it shows the presence of
maximum bending point in the wavelength curve varies according to

15
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

Shear wave velocity (m/s) 120 50

0 300 600 900 1200 45

Wavelength (m)
0 100
40

Wavelength (m)
35
80
Actual Bedrock depth = 9 m 30
3m 450 500 550 600
60 Phse velocity (m/s)
Estimated bedrock
depth 40
20
20
Depth (m)

V = 500 m/s
s(Pre) 0
400 600 800 1000
Phase velocity (m/s)
Fig. 7. Typical phase velocity vs wavelength plot to identify the change in
40 wavelength.
Fill (Sand with rock fragments)
S VI (Residual soil) 3.3. HVSR method
S V (Weak to moderately weak)
S IV (moderately weak to Horizontal to vertical spectral ratio (HVSR) is calculated using Eq.
moderately strong) (3) from the ambient noise captured by three-component geophone (2
S III (Moderately strong) horizontal geophones – HEW, HNS; 1 vertical geophone) (Nakamura,
60 1989).

Fig. 5. Shear wave velocity profile corresponding to the reference borehole |fEW |2 + |fNS |2
#3.1. HVSR (f ) =
2 |fV |2 (3)

the bedrock depth. Thus, a novel attempt has been made to establish a where |fEW | and |fNS | are amplitude spectra of the horizontal components
relationship between the wavelength bending point and the known in east-west and north-south directions respectively. |fV | is the ampli-
bedrock depth for Jurong formation. The procedure is explained briefly tude spectrum of the vertical component.
as follows. From the HVSR spectrum, the frequency corresponding to a peak
At first, wavelength is calculated from the experimental dispersion HVSR value is selected. In order to identify the reliable fundamental
curve and plotted against phase velocity as shown in Fig. 7. In the frequency, Site EffectS assessment using AMbient Excitations (SESAME)
‘phase velocity vs wavelength’ plot, a straight line is drawn along the guidelines (Bard and Sesame-Team, 2004) are followed. During the
wavelength curve to find the bending/deviation point as shown in HVSR calculation, spectral smoothing of 15% is carried out in trian-
Fig. 7. Using the bending point in wavelength of all the testing sites and gular window. Tapering of 5% is allowed and the window length is
the corresponding depth of bedrock (from the borehole profile), a adjusted by trial and error method so as to satisfy the SESAME criteria.
correlation is established. This apparent relationship can be used to In few cases, occurrence of double peaks could not be avoided since the
estimate the bedrock depth. testing has been done along the express way. The highest peak is con-
sidered as the predominant frequency to establish the correlation be-
tween the fundamental frequency of the site and the bedrock depth. The
occurrence of secondary peak is attributed primarily to heavy traffic
and it does not significantly affect the predominant frequency of the site

800 200
BH ID BH ID
1 1
2.1 2.1
Phase Velocity (m/s)

2.2 150 2.2


600
Wavelength (m)

3.1 3.1
3.2 3.2
4 4
5 5
6.1 100 6.1
6.2 6.2
400 7 7
8 8

50
200

0
0 10 20 30 0 10 20 30
Frequency (Hz) Frequency (Hz)
Fig. 6. (a) Dispersion curves, (b) wavelength vs frequency of all the tested sites in Jurong formation.

16
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

Shear wave velocity (m/s) Shear wave velocity (m/s)


0 200 400 600 800 0 300 600 900 1200
0 0
500 m/s
500 m/s
10 10

20 20
Depth (m)

Depth (m)
30 1 30 3.1
2.1 3.2
2.2 4
40 40
1 2.1 3.1 3.2 4
Fill Fill
S VI S VI
50 SV 2.2
50 SV
S IV S IV
S III S III
60 60
Shear wave velocity (m/s) Shear wave velocity (m/s)

0 300 600 900 1200 0 300 600 900 1200


0 0
500 m/s 500 m/s

10 10

20 20
Depth (m)

Depth (m)

30 5 30 7
6.1 8
6.2
40 40
Fill 5 6.1 6.2 Fill 7 8
S VI S VI
SV SV
50 S IV 50 S IV
S III S III
S II
60 60
Fig. 8. Shear wave velocity profiles of all the tested sites in Jurong formation.

Table 3
Evaluation of proposed approaches in the bedrock estimation.
Testing Location Reference Borehole ID Actual Bedrock Depth (m) Preselected Vs based approach Wavelength based approach HVSR approach

Jurong Central Park 1.1 27.5 −2.7 −1.6 −0.3


Jalan Boonlay 2.1 29 −2.7 −4.8 +7.1
2.2 34 −7.0 −7.0 x
Bukit Batok 3.1 9 +3.4 −0.6 −2.3
3.2 16 −0.4 −1.1 x
Jurong west st 82 4 21 +1.1 −4.6 +3.9
PIE Road 5.1 11 +1.0 +1.5 −0.9
NTU site 1 6.1 12 +2.4 +2.2 −2.9
6.2 17 +2.5 +4.0 +4.2
Jurong Town Hall Road 7 19 −0.5 +6.8 −2.2
West Coast Road 8 19.5 +1.5 −1.1 −5.5
NTU site 2 9.1 10 * * +0.34
9.2 20 * * +1.88

Note: (+) indicates overestimation of depth & (−) indicates underestimation of depth, x – Reliable peak was not identified due to higher traffic volume; * – Space
constraint to form microtremor array.

(Harutoonian et al., 2013). In this study, the fundamental frequencies of formation in Singapore.
all the tested sites are correlated with corresponding bedrock depths
through a power function. Eventually, this empirical relationship can be
used to estimate the depth of weathered bedrock at the Jurong

17
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

60 120

100

Wavelength (m)
Wavelength (m)
40 80

60
1 3.1
20 2.1 40 3.2
2.2 4
20

0 0
400 500 600 400 600 800
Phase velocity (m/s) Phase velocity (m/s)

80 80

70
Wavelength (m)

60

Wavelength (m)
60

50
40 40
5 30
6.1 7
20 6.2 20 8
10
0 0
400 500 600 400 600 800
Phase velocity (m/s) Phase velocity (m/s)
Fig. 9. Change in phase velocity vs wavelength plot of all the tested sites in Jurong formation.

40 wavelength (λ) at lower bending point and bedrock depth (D) is es-
(b). Wavelength based approach
tablished as follows.
1 D = 345.9 λ−0.881 (R2 = 0.78) (4)
2.1
Outliner 2.2 The variations in wavelength with respect to phase velocity of all
Bedrock depth (m)

3.1 the testing sites are shown in Fig. 9. Nearer to the lower bending por-
3.2
tion, a straight line is drawn along the straight portion of the wave-
4
5 length profile to identify the lower deviation point. In all the cases the
6.1 bending portion is magnified to exactly identify the deviation/bending
20
6.2 point. At most care should be taken to identify the bend point and it
7 also depends on the quality of the microtremor signal. The wavelength
8 corresponding to this deviation point is plotted against the bedrock
depth for all the testing sites in Jurong formation as shown in Fig. 10.
Boreholes #2.1 and #2.2 are approximately 25 m apart, and MAM tests
were conducted using ‘L & shaped arrays’. The boreholes #2.1 and
R2 = 0.78
D = 345.9 λ−0.881 #2.2 were located at the corner of the ‘L & shaped arrays’ respec-
tively and the horizontal legs of the arrays were overlapped between
0 the boreholes during the tests. No significant change between the two
0 20 40 60 80 dispersion curves was identified. However, the depths of bedrock cor-
Wavelength (λ)(m) responding to bore logs #2.1 and #2.2 are 29 m and 34 m respectively.
From Fig. 10, it can be seen that data corresponding to the borehole
Fig. 10. Change in wavelength vs corresponding bedrock depth of all the tested
#2.2 does not follow the trend and it is considered as an outliner.
sites in Jurong formation.
Horizontal to vertical spectral ratio of all the tested sites are shown
in Fig. 11. Frequency corresponding to the peak HVSR is considered as a
4. Results & discussions natural frequency of the site. The observed natural frequencies of all the
tested sites in the Jurong formation are plotted against the corre-
As discussed in the preceding Section 3.1, the Vs profiles corre- sponding bedrock depths, which exhibits an inverse relationship as
sponding to all the testing sites were used in the identification of bed- shown in Fig. 12. A site-specific empirical correlation model is proposed
rock depth. Fig. 8 shows the Vs profiles of all the testing locations along between the natural frequency (f) and the bedrock depth (D) as shown
with the borehole profiles. A vertical line is drawn from abscissa cor- below.
responding to 500 m/s to locate the bedrock depth. The difference in
actual and estimated bedrock depth is shown in Table 3. D = 69.4 f −0.996 (R2 = 0.75) (5)
As stated in wavelength based approach, a relationship between These relationships can be applied to estimate the depth of bedrock

18
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

6
1
2.1
4
2.2

HVSR 2

0
1 10
Frequency (Hz)

8
3.1
6 4
5
HVSR

0
1 10
Frequency (Hz)

20

16 6.1
6.2
12 7
HVSR

8
4
0
1 10
Frequency (Hz)

10

8
8
6 9
HVSR

4 10

0
0.01 0.1 1 10
Frequency (Hz)
Fig. 11. HVSR profiles of all the tested sites in Jurong formation.

in Jurong formation. The estimated average shear wave velocity of is 5 m, there may not be considerable variation in bedrock depth within
sedimentary layer considering quarter wavelength law (Delgado et al., the overlapped array area. This may be the possible reason for the
2000) from Eq. (5) is about 279 m/s. Fig. 13(a)–(c) compare the actual underestimation of the bedrock depth.
bedrock depth from borelog and the estimated bedrock depth using the It is seen from Fig. 13(b) that the wavelength based approach can
preselected Vs approach, the wavelength based approach, and the HVSR estimate the bedrock depth reasonably well (RMSE = 4.55) at most of
approach respectively. As can be seen in Fig. 13(a), most of the data are the testing sites except for boreholes #2.2 and #7. The possible un-
scattered around the line of equality and the preselected method shows derestimation at borehole #2.2 was explained in the previous para-
the best estimate (e.g., root mean square error, RMSE = 2.79) among graph. Similarly at the testing location #7, a gentle slope in the bedrock
the three proposed methods. The higher RMSE value is due to some is observed from the available nearby bore log information. This may
outliners and site heterogeneity. For example, borehole #2.2 shows have resulted in some uncertainties in applying the wavelength based
considerable underestimation of bedrock depth (variation = 7 m) as approach. In addition, it should be noted that the approach is more
shown in Fig. 13(a). As discussed in the Section 3.2, boreholes #2.1 and sensitive in prediction at lower wavelength regime.
#2.2 are 25 m apart and one leg of the both microtremor arrays were HVSR based approach estimates the bedrock depth within ± 5 m
overlapped between the boreholes during the tests. Although the var- accuracy (RMSE = 3.52) at all the testing sites excepting the borehole
iation in bedrock depth (obtained from borelog) between the boreholes location #2.1. This is probably attributed to the uneven bedrock profile

19
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

40 result from the existence of different kinds of rock (limestone, sand-


stone and siltstone) materials that can affect the fundamental period of
1 the site.
2.1
Actual bedrock depth (m)

3.1 5. Conclusions
4
5
6.1 Surface wave tests (microtremor array method and 3-component
6.2 microtremor method) were carried out in the weathered Jurong for-
7 mation in Singapore to identify the depth of bedrock. Based on the
20 8
9
wave measurements, three independent methods were proposed to es-
10 timate the depth of bedrock: (i) preselected Vs based approach, (ii)
wavelength based approach, and (iii) HVSR based approach. The fol-
lowing conclusions are drawn based on the experimental results.

D = 69.4*( f -0.996) 1. All the proposed methods in this study are based on non-invasive
R2 = 0.75 seismic wave measurements and can be efficiently used for rea-
sonable estimation of the bedrock depth at Jurong formation.
0 Conceptually, the methodologies would be easily extended for si-
0 2 4 6 8 10 milar geologic formations (e.g., weathered sedimentary rock for-
Natural Frequency (Hz) mation).
2. In the present case, the preselected Vs based approach provided the
Fig. 12. Natural frequency and corresponding bedrock depth of tested sites in
Jurong formation.
best prediction of bedrock depth in Jurong formation. ESPAC ana-
lysis was adopted to generate the dispersion curve. Subsequently,
the Vs profile was derived from the best fit of theoretical dispersion
in the Jalan Boonlay testing site as explained in the previous para- curve with the experimental dispersion curve. Using the Vs profile,
graphs. Although the general prediction trend looks reasonable, rela- Vs of 500 m/s can be reasonably used as a guideline to identify the
tively wide scatters were observed around the line of equality. This may depth of bedrock in Jurong formation.

40 40
(a). Preselected Vs approach (b). Wavelength based approach
Estimated bedrock depth (m)
Estimated bedrock depth (m)

1 1
2.1 2.1
20 2.2
20 2.2
3.1 3.1
3.2 3.2
4 4
5 5
6.1 6.1
6.2 6.2
RMSE = 4.55
RMSE = 2.79 7 7
8 8
0 0
0 20 40 0 20 40
Actual bedrock depth (m) Actual bedrock depth (m)

40
(c). HVSR approach
Estimated bedrock depth (m)

1
2.1
3.1
20 3.2
4
5
6.1
6.2
7
8
RMSE = 3.52 9
10

0
0 20 40
Actual bedrock depth (m)
Fig. 13. Actual and estimated bedrock depth: (a) preselected Vs approach, (b) wavelength based approach, (c) HVSR approach.

20
P. Subramaniam et al. Tunnelling and Underground Space Technology 86 (2019) 10–21

3. Wavelength based approach used the experimental dispersion curve vibration intensity of microtremors. Tunn. Undergr. Sp. Technol. 51, 338–345.
to establish a relationship between the change in wavelength and https://doi.org/10.1016/j.tust.2015.10.032.
Harutoonian, P., Leo, C.J., Tokeshi, K., Doanh, T., Castellaro, S., Zou, J.J.,
the corresponding bedrock depth. Since the method is very sensitive, Liyanapathirana, D.S., Wong, H., 2013. Investigation of composite compacted ground
care should be taken in the picking of theoretical dispersion curve using microtremors. J. Geotech. Geoenviron. Eng. 139, 1539–1553. https://doi.org/
and in the identification of change point. 10.1061/(ASCE)GT.1943-5606.0000881.
Lai, C.G., Rix, G.J., 1998. Simultaneous inversion of Rayleigh phase velocity and at-
4. HVSR based approach might be the simplest method in terms of tenuation for near-surface site characterization. Georg. Inst. Technol. Report No: GIT-
easiness in field testing and analysis. The method is most convenient CEE/GEO-98-2, 1-258.
where the geophone array formation is not possible. Lee, K.W., Zhou, Y., 2009. Geology of Singapore. Defense Science and Technology Agency
(DSTA).
Ling, S., Okada, H., 1993. An extended use of the spatial autocorrelation method for the
In the present study, dispersion curves and HVSR were analyzed estimation of structure using microtremors. In: 89th SEGJ Conference. Society of
separately. However, the combination of these two methods may im- Exploration Geophysicists of Japan, Nagoya, pp. 44–48 in Japanese.
Louie, J.N., 2001. Faster, better: Shear-wave velocity to 100 meters depth from refraction
prove the accuracy of bedrock estimation. These non-invasive based
microtremor arrays. Bull. Seismol. Soc. Am. 91, 347–364. https://doi.org/10.1785/
methods would minimize the number of drilling required during the 0120000098.
initial planning of any large infrastructure projects such as design and Moon, S.-W., Hayashi, K., Ku, T., 2017. Estimating spatial variations in bedrock depth and
alignment of underground structures. However, the recommended va- weathering degree in decomposed granite from surface waves. J. Geotech.
Geoenviron. Eng. 143, 1–11. https://doi.org/10.1061/(ASCE)GT.1943-5606.
lues and equations given in the paper are regional specific which de- 0001679.
pend on the type of rock and weathering conditions. Nakamura, Y., 1989. A method for dynamic characteristics estimation of subsurface using
microtremor on the ground surface. Railway Tech. Res. Inst., Quart. Rep. 30 (1),
25–33.
Acknowledgment Nazarian, S., Stokoe, K., 1984. In situ shear wave velocities from spectral analysis of
surface waves. In: Proc. 8th world Conf. Earthq. Eng., vol. 3, pp. 31–38.
The authors appreciate the financial support from Singapore Land Ohori, M., Nobata, A., Wakamatsu, K., 2002. A comparison of ESAC and FK methods of
estimating phase velocity using arbitrarily shaped microtremor arrays. Bull. Seismol.
Transport Authority (LTA, Award Number: R-302-000-164-490). We Soc. Am. 92, 2323–2332. https://doi.org/10.1785/0119980109.
also thank Mr. Vinoth Ganapathiraman for his assistance in the field Okada, H., 1994. A research on the practical application of microtremor exploration
testing. technique to a wide area survey of a underground structure under 3,000 m in depth
(in Japanese). Report No. 03554009. Japan Science. Tech.
Okada, H., 2006. Theory of efficient array observations of microtremors with special
References reference to the SPAC method. Explor. Geophys. 37, 73–85. https://doi.org/10.1071/
EG06073.
Pamuk, E., Akgün, M., Özdağ, Ö.C., Gönenç, T., 2017. 2D soil and engineering-seismic
Akgün, M., Gönenç, T., Tunçel, A., Pamukçu, O., 2013. A multi-approach geophysical
bedrock modeling of eastern part of Izmir inner bay/Turkey. J. Appl. Geophys. 137,
estimation of soil dynamic properties in settlements: a case study in Güzelbahce-İzmir
104–117. https://doi.org/10.1016/j.jappgeo.2016.12.016.
(Western Anatolia). J. Geophys. Eng. 10. https://doi.org/10.1088/1742-2132/10/4/
Park, C.B., Miller, R.D., Xia, J., 1999. Multichannel analysis of surface waves. Geophysics
045001.
64, 800–808. https://doi.org/10.1190/1.1444590.
Aki, K., 1957. Space and time spectra of stationary stochastic waves, with special re-
Parolai, S., Bormann, P., Milkereit, C., 2002. New relationships between Vs, thickness of
ference to microtremors. Bull. Earthquke Res. Inst. 35, 415–456.
sediments, and resonance frequency calculated by the H/V ratio of seismic noise for
Anderson, N., Croxton, N., Hoover, R., Sirles, P., 2008. Geophysical methods commonly
the cologne area (Germany). Bull. Seismol. Soc. Am. 92, 2521–2527. https://doi.org/
employed for geotechnical site characterization. Transp. Research Circular, E-C 130,
10.1785/0120010248.
TRB, Wshington DC.
Parolai, S., Richwalski, S.M., Milkereit, C., Fäh, D., 2006. S-wave velocity profiles for
Bard, P., Sesame-Team., 2004. Guidelines for the implementation of the H/V spectral
earthquake engineering purposes for the Cologne area (Germany). Bull. Earthq. Eng.
ratio technique on ambient vibrations measurements, processing and interpretation.
4, 65–94. https://doi.org/10.1007/s10518-005-5758-2.
SESAME Eur. Res. Proj. WP12 – Deliv. D23.12 1–62.
Ridwan, M., Widiyantoro, S., Afnimar, Irsyam, M., 2015. Identification of engineering
Brown, L., Boore, D., Stokoe, K., 2000. Comparison of shear-wave velocity profiles from
bedrock in Jakarta by using array observations of microtremors. Proc. Earth Planet.
SASW and downhole seismic tests at a strong-motion site. In: Proc. 12th World Conf.
Sci. 12, 77–83. https://doi.org/10.1016/j.proeps.2015.03.040.
Earthquake Eng, pp. 1–8.
Seht, M.I.-V., Wohlenberg, J., 1999. Microtremor measurements used to map thickness of
BS 5930, 1999. Code of practice for site investigations. https://doi.org/10.3403/
soft sediments. Bull. Seismol. Soc. Am. 89 (1), 250–259.
01897862.
Shima, H., Sakayama, T., 1987. Resistivity tomography: an approach to 2-D resistivity
Cho, I., Senna, S., Fujiwara, H., 2013. Miniature array analysis of microtremors.
inverse problems. In: {SEG} Technical Program Expanded Abstracts 1987. Society of
Geophysics 78, KS13–KS23. https://doi.org/10.1190/geo2012-0248.1.
Exploration Geophysicists. https://doi.org/10.1190/1.1892038.
Cho, I., Tada, T., Shinozaki, Y., 2006. Centerless circular array method: Inferring phase
Tamrakar, P., Luke, B., 2013. Feasibility of approximating depth to shallow bedrock di-
velocities of Rayleigh waves in broad wavelength ranges using microtremor records.
rectly from the rayleigh wave dispersion curve. Geotech. Test. J. 36 (6), 882–890.
J. Geophys. Res. 111. https://doi.org/10.1029/2005jb004235.
https://doi.org/10.1520/GTJ20120080.
Coulouma, G., Lagacherie, P., Samyn, K., Grandjean, G., 2013. Comparisons of dry ERT,
Trevisani, S., Boaga, J., Agostini, L., Galgaro, A., 2017. Insights into bedrock surface
diachronic ERT and the spectral analysis of surface waves for estimating bedrock
morphology using low-cost passive seismic surveys and integrated geostatistical
depth in various Mediterranean landscapes. Geoderma 199, 128–134. https://doi.
analysis. Sci. Total Environ. 578, 186–202. https://doi.org/10.1016/j.scitotenv.
org/10.1016/j.geoderma.2012.07.026.
2016.11.041.
Dal Moro, G., Pipan, M., Gabrielli, P., 2007. Rayleigh wave dispersion curve inversion via
Yilar, E., Baise, L.G., Ebel, J.E., 2017. Using H/V measurements to determine depth to
genetic algorithms and Marginal Posterior Probability Density estimation. J. Appl.
bedrock and Vs30in Boston, Massachusetts. Eng. Geol. 217, 12–22. https://doi.org/
Geophys. 61, 39–55. https://doi.org/10.1016/j.jappgeo.2006.04.002.
10.1016/j.enggeo.2016.12.002.
Delgado, J., López Casado, C., Giner, J., Estévez, A., Cuenca, A., Molina, S., 2000.
Zhao, J., Liu, Q., Lee, K.W., Choa, V., Teh, C.I., 1999. Underground cavern development
Microtremors as a geophysical exploration tool: applications and limitations. Pure
in the Jurong sedimentary rock formation. Tunn. Undergr. Sp. Technol. 14, 449–459.
Appl. Geophys. 157, 1445–1462. https://doi.org/10.1007/PL00001128.
https://doi.org/10.1016/s0886-7798(00)00007-9.
Gao, Y., Jiang, Y., Li, B., 2016. Voids delineation behind tunnel lining based on the

21

Вам также может понравиться