You are on page 1of 8

Journal of Applied Geophysics 104 (2014) 106–113

Contents lists available at ScienceDirect

Journal of Applied Geophysics


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

Two-dimensional pre-stack reverse time imaging based on


tunnel space☆
Fei Cheng a, Jiangping Liu a,⁎, Niannian Qu b, Mao Mao a, Liming Zhou c
a
Subsurface Imaging and Sensing Laboratory, Institute of Geophysics and Geomatics, China University of Geosciences, 388 Rumo Rd., Wuhan, Hubei, 430074, China
b
Guizhou Geology Survey, Guiyang, Guizhou, 550004, China
c
Changjiang River Scientific Research Institute, Wuhan, Hubei, 430010, China

a r t i c l e i n f o a b s t r a c t

Article history: In order to increase the safety and efficiency in tunnel constructions, there is a need to carry out an effective and
Received 20 July 2013 precise tunnel prediction method to detect unexpected lithological and structural heterogeneities ahead of
Accepted 21 February 2014 tunnel face. Seismic prediction is considered as one correct and efficient method. The assumption, which differs
Available online 28 February 2014
from the reality, taken in most of the current tunnel seismic imaging methods is that the tunnel space is a
homogeneous medium with surrounded layers with the same elastic characters. In this paper, taking into
Keywords:
Advance predication
account the actual situation of tunnel space, we propose some new tunnel geological models that are closer to
Tunnel space the reality using the first-order coupled elastic equations of particle velocity and stress, and high order staggered
Reverse time migration grid finite-difference algorithm to fulfill numerical simulation of seismic full-wave fields in tunnel space. Then for
Multiwave imaging these synthetic simulated records, we utilize reverse time migration operator based on non-conversion wave
equation with decoupled P- and S-waves, and excitation time imaging condition to achieve reliable two
dimensional (2D) reverse time migration imaging (RTM) based on tunnel space effectively. Results demonstrate
that (1) it is able to achieve synthetic simulation and reverse time migration imaging correctly by using a
staggered grid finite-difference (FD) algorithm with second-order accuracy in time and fourth-order accuracy
in space, and reverse time operator based on non-conversion wave equation with decoupled P- and S-waves;
(2) tunnel-based reverse time migration imaging can effectively suppress mirror artifact occurring in
conventional imaging approaches; and (3) as the dip angle of lithological interface decreases, the energy of
P wave imaging increases while the energy of S wave imaging decreases when shooting and receiving at the
same side of interface, while when the dip angle of interface is 90°, common-source gather with shots near the
tunnel face is beneficial to the imaging of P wave.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction hampering failure-free tunnel construction. Advance borehole drilling is


relatively more accurate than surface measurements, but is also normal-
Infrastructure construction in urban areas and larger traffic deve- ly restricted to a limited prediction distance which will significantly
lopment projects often involve tunnel excavation. Tunnel is an under- delay the construction progress. Hence, to increase the safety and
ground work with many unknown geological conditions such as efficiency of tunnel constructions, geophysical exploration ahead of a
fracture zones, faults, caves and aquifers in front of it. Failing to tunnel has become a valuable tool.
recognize and predict these geological conditions ahead of the tunnel Nondestructive geophysical methods are efficient techniques inves-
front represents dangers ranging from construction progress delay tigating and predicting lithological and structural heterogeneities for
and instrument damage to tremendous casualties. Therefore, using distances up to several hundred meters from the tunnel wall. Among
advanced technology to predict these lithological and structural all these methods, seismic imaging is the most effective one because
heterogeneities ahead of a tunnel front is of major importance before of its relatively deep penetration range and high spatial resolution.
excavating tunnels. Surface measurements can provide general Different tunnel seismic prediction systems have been applied since
geological information of the ground, but the data resolution is not the early 1990s in tunneling projects worldwide. The Tunnel Seismic
sufficient to identify lithological boundaries or other geological features Prediction (TSP) software developed by AMT, Switzerland (Dickmann
and Sander, 1996, Bohlen et al., 2007), identifies reflections from front
of the tunnel face in the recordings by their traveltime curves and
☆ This is publication of no. 41202223 project funded by the National Natural Science
Foundation of China (NSFC).
their intersection with the tunnel axis derived from depth migration
⁎ Corresponding author. Tel.: +86 27 67883525. images. Zeng (Zeng, 1994) proposed the negative apparent velocity
E-mail address: liujp_geop@126.com (J. Liu). method which uses apparent velocity gradient of direct wave and

http://dx.doi.org/10.1016/j.jappgeo.2014.02.013
0926-9851/© 2014 Elsevier B.V. All rights reserved.
F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113 107

reflected wave to predict faults ahead of tunnel face. OYO, Japan, put for- where ux and uz denote x component and z component, respectively,
ward a method named HSP (Horizontal Seismic Profiling) by using a while Vp and Vs are the velocity of P and S waves, respectively. With
source and receiver geometry as in a surface refraction seismic survey introducing:
(Inazaki et al., 1999). NSA Engineering in the United States developed 8 2
the True Ref1ection Tomography (TRT) technology which uses multi- >
< τ p ¼ V p ð∂ux =∂x þ ∂uz =∂zÞ
τ ¼ V s ð∂ux =∂z−∂uz =∂xÞ ; ð2Þ
point shooting and receiving observation system to identify the rock 2
>
: sx
mass conditions ahead of tunnel excavation. The spatial distribution of 2
τ sz ¼ V s ð∂uz =∂x−∂ux =∂zÞ
receivers and sources contributes to obtain sufficient spatial wave field in-
formation (Neil et al., 1999). In 1999, a seismic exploration system called Taking the derivative of time on both hand sides of the equations,
the Integrated Seismic Imaging System (ISIS) has been developed to ob- the order of Eq. (1) can be reduced to obtain generalized first order
tain the knowledge of the upcoming rock conditions during tunneling at stress and velocity non-conversion wave equation (Eq. (3))
the GFZ, Germany (Borm et al., 2001, 2003). In this system, TBM is used
8
to generate wave field and three-component geophones situated on the > ∂vx =∂t ¼ ∂τp =∂x þ ∂τ sx =∂z
>
>
side wall are used to register records. Throughout the previous research >
>
< ∂vz =∂t ¼ ∂τp =∂z þ ∂τ sz =∂x
>
the ignorance of tunnel space does not correspond to reality obviously. 2
∂τ p =∂t ¼ V p ð∂vx =∂x þ ∂vz =∂zÞ ; ð3Þ
In recent researches about tunnel prediction using seismic records, >
>
>
>
2
∂τ sx =∂t ¼ V s ð∂vx =∂z−∂vz =∂xÞ
Bohlen (Bohlen et al., 2007) and Stefan (Stefan et al., 2010, 2011) consid- >
>
: 2
ered the impacts of tunnel space and used surface wave along with con- ∂τ sz =∂t ¼ V s ð∂vz =∂x−∂vx =∂zÞ
verted S wave to predict geological structures ahead of tunnel. They
vx and vz are mixed wave field including both P wave and S wave. We
discussed the conversion mechanism between these two kinds of waves
need to decouple the wave field to obtain independent P and S waves
as well. Petronio (Petronio et al., 2007) used the Tunnel-Seismic While-
for reverse time migration. Setting vx = vpx + vsx,vz = vpz + vsz and
Drilling (TSWD) method to predict geologic interfaces ahead of a tunnel
substituting them into Eq. (3), we calculate the first order equation
front. In the research area of seismic imaging for tunnel excavation,
with decoupled P and S waves (Eq. (4))
Rechlin (Rechlin et al., 2009) implemented Kirchhoff migration for
TSWD data. Lüth (Lüth et al., 2008) applied “Kirchhoff Pre-Stack Depth
8
Migration” (KPSDM) to tunnel predication. Furthermore, Tzavaras > ∂v =∂t ¼ ∂τ p =∂x
>
> px
(Tzavaras et al., 2012) applied “Kirchhoff Pre-Stack Depth Migration” >
> ∂vpz =∂t ¼ ∂τ p =∂z
>
>
(KPSDM), “Fresnel Volume Migration” (FVM), and “Reflection-Image- >
>
< ∂vsx =∂t ¼ ∂τsx =∂z
>
Spectroscopy” (RIS) to tunnel predication, respectively. However, only a ∂vsz =∂t ¼ ∂τsz =∂x ; ð4Þ
few researches are related to using reverse time migration to tunnel pred- >
> 2
> ∂τ p =∂t ¼ V p ð∂vx =∂x þ ∂vz =∂zÞ
>
>
>
ication. It is known that restricted survey layout and rock mass location af- >
> ∂τ sx =∂t ¼ V 2s ð∂vx =∂z−∂vz =∂xÞ
>
>
fect resolution to different extents. Therefore, it is necessary to consider : 2
∂τ sz =∂t ¼ V s ð∂vz =∂x−∂vx =∂zÞ
the influence of tunnel space and select an accurate imaging method to
do the prediction of geological conditions in front of the tunnel.
In this paper, we present a new tunnel seismic prediction method. De-propagation process, an extrapolation of the recorder wave field
Taking into account the tunnel space, we build tunnel geological models (seismogram), which is prescribed in the reverse order at the receiver
that are closer to the reality by using a staggered grid finite-difference location, includes calculating imaging time for each grid point and
(FD) algorithm with second-order accuracy in time and fourth-order ac- applying imaging condition to extrapolated wave field in every extrap-
curacy in space to realize numerical simulation of seismic full-wave field olation. Reverse time extrapolation is an initial boundary value problem
based on tunnel space (Liu et al., 2012). Then for these simulated re- based on the velocity–stress equation (Eq. (4)). It begins from the
cords from different models, we utilize non-conversion wave equation maximum recorded time of the (x, z) plane and takes seismic record
with decoupled P- and S-waves to achieve reliable reverse time migra- f(x, z = 0, t) as the boundary condition to calculate back through the
tion imaging in tunnel space effectively. At last, we verify and demon- time axis.
strate the effectiveness of this method in field tunnel data.
2.2. Imaging condition
2. Methodology
We use the dynamic programming traveltime computation method
2.1. Elastic wave pre-stack reverse time migration proposed by Schneider (Schneider et al., 1992) to calculate imaging
time. This method will accurately determine the first arrival traveltime
In wave field extrapolation, incidence of seismic wave to the inter- through arbitrary, discrete, and discontinuous velocity distributions. It is
face reproduces converted wave. This converted wave leads to disturbed more stable, needs less calculation time, and is not going to cause any
wave in the imaging process because it does not meet the imaging shaded region.
condition. Therefore it is necessary to develop a non-conversion wave We use excitation time image condition as shown in Eq. (5)
equation and decouple elastic wave field to guarantee that P wave and
S wave propagate independently. In 2009, Zhang (2009) put forward (   0 0
X
T
1 0 0
x ¼ x ; z ¼ z ; t ¼ td x ; z
the non-conversion wave equation with decoupled P- and S-waves. Iðx; zÞ ¼ Rðx; z; T−t Þ f ðx; z; T−t Þ f ðx; z; t Þ ¼ ; ð5Þ
0 other
t¼0
When propagating in inhomogeneous medium in which ρ and μ are
constant while λ is variable, the completely decoupled S and P waves
propagate independently and no converted wave is reproduced. There- where T is the maximum recorded time of wave field record. The corre-
fore assuming that ρ and μ are constant, we could obtain the non- sponding wave field is picked up from reverse time extrapolation wave
conversion elastic displacement equation as follows (Eq. (1)) field R(x, z, t) according to excitation time td(x′, z′).

2.3. Boundary condition


8    
< ∂2 ux =∂t 2 ¼ ∂=∂x V 2p ð∂ux =∂x þ ∂uz =∂zÞ þ ∂=∂z V 2s ð∂ux =∂z−∂uz =∂xÞ
   ; ð1Þ Absorbing boundary condition around the boundary of models in
: ∂2 u =∂t 2 ¼ ∂=∂z V 2 ð∂u =∂x þ ∂u =∂zÞ þ ∂=∂x V 2 ð∂u =∂x−∂u =∂zÞ
z p x z s x z
the de-propagation process is employed in our models (Cerjan et al.,
108 F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113

shown in Fig. 2, in which models are illustrated by the velocity of


P wave. These models can help us to verify the effectiveness of the
proposed numerical simulation method and to study wave field charac-
teristics for different models. ModelI-1 is designed for investigating the
influence of survey layout and lithological interfaces with large dipping
angles (the angle between interface and tunnel axis) on reverse time
migration imaging. For the purpose of analyzing the variation regulato-
ry of imaging according to the dipping angle of interface, ModelI-2 and
ModelI-3 are designed with an interface with a dipping angle of 75° and
60°, respectively. The normal depth of tunnel face in all of the three
models is 65 m. Model II is the three layer wedge model, which is
designed for studying the imaging of the complicated geology structure.
Fig. 1. Acquisition geometry. The tunnel is 100 m in length and 12 m in width. 83 receivers
Model III is similar with Model I-3 except that it considers the tunnel as
with 35 in both sides of the side walls respectively and 13 in the tunnel face are located
counterclockwise at 1 m depth. Source1 is 16 m away from the first geophone and Source2 a homogeneous medium with same elastic features of surrounded
which is also at 1 m depth is located at the intersection of sidewall and tunnel face. Solid layers. Model III aims at the different results of imaging when not
dots and asterisks represent receivers and sources, respectively. considering tunnel space. Table 1 shows detailed parameters of these
theoretical models.
1985). That means in certain ribbon-like regions at the boundary, the
stress and particle velocity are multiplied by the factor G (Eq. (6))
4. Modeling results
h i
2
G ¼ exp −ð−0:305=ΝÞ  ðΝ−iÞ 0≤i ≤Ν ; ð6Þ
4.1. Model I results

where N denotes the number grid of absorbing boundary. Using the reverse time migration method and corresponding bound-
In addition, absorbing boundary condition is also tackled in tunnel ary conditions proposed above, we fulfill full wave field simulation and
space. Using m × n grid, the damping factor in tunnel space is shown reverse time migration imaging by FORTRAN. Full wave field simulation
in Eq. (7) based on tunnel space are achieved in that these records include direct
P wave, Rayleigh wave, reflected P wave, reflected P–S converted wave,
8 h i
2
>
> exp −ð−0:305=ðn=2ÞÞ  ðn=2 þ zÞ 0 ≤x ≤ ðm−n=2Þ−z reflected S–P converted wave, and reflected S wave (Fig. 3a, b). Fig. 4a
>
< h i −n=2≤ z≤0;
2 0 ≤x ≤ ðm−n=2Þ þ z and b shows the P wave and S wave reverse time migration imaging
G¼ exp −ð−0:305=ðn=2ÞÞ  ðn=2−zÞ 0 ≤z≤ n=2; −ði−m þ n=2Þ ≤ z
>
> h i
>
: exp −ð−0:305=ðn=2ÞÞ  ðm−xÞ 2 m−n=2≤ x ≤m; profile corresponding to Fig. 3a and b, respectively, Because of the tun-
≤ x−m þ n=2
nel space limitation, the extension of the imaging profile is limited in
ð7Þ that the observation system layout cannot be set extensively like surface
observation system. Nevertheless, Fig. 4a and b can still indicate a con-
centrated and convergent energy group at the depth of 65 m, and it is
3. Survey layout and modeling consistent with the position of the interface. This demonstrates the ef-
fectiveness of the reverse time migration method to fulfill geological in-
3.1. Survey layout terface imaging ahead of tunnel. By contrasting Fig. 4a and b, we can see
that the energy of the S wave migration profile (Fig. 4b) is stronger than
Two survey layouts (Fig. 1) are designed to investigate the effects on that of P wave migration profile (Fig. 4a). This illustrates that common-
reverse time migration imaging results from different survey layouts source gather with shots on the side wall is better for S wave imaging.
systems. One of these two systems is a common-receiver gather with We can see that the energy of P wave in Fig. 3c and d is stronger than
shots on the side walls, in which the source is excited at Source1 and that in Fig. 3a and b while the energy of S wave in Fig. 3c and d is weaker
geophones are located along the side wall. The other is a common- than that in Fig. 3a and b. This is because when shooting on the side wall,
source gather with shots on the tunnel face, in which the source is excit- Rayleigh wave converts to S wave at the intersection of side wall and
ed at Source2 and geophones are located along the side wall. tunnel face, then travels back to receivers. This travel time approximates
two-way S wave travel time. So this in-phase stacking strengthens the
3.2. Modeling energy of reflected S wave. When it comes to shooting on the tunnel
face, the energy of reflected S wave is weaker because there is only
In order to investigate the influence of survey layout and interface on two-way S wave. Fig. 4c and d shows the P wave and S wave pre-
reverse time migration imaging, 3 theoretical models are designed and stack reverse time migration imaging profile, respectively. The energy

Fig. 2. Theoretical models. (a) An interface with a dipping angle of 90°; (b) an interface with a dipping angle of 75°; (c) an interface with a dipping angle of 60°; (d) a three-layer wedge
model; (e) a two-layer model with an interface with a dipping angle of 60° and considers the tunnel as a homogeneous medium with the same elastic features of surrounded layers. The
asterisks represent the sources.
F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113 109

Table 1 that in Fig. 5d. There is another interface on the left of the previously
Parameters of theoretical models. Vp and Vs are P- and S-wave velocities in m/s, respective- discussed interface, which is considered as the result of reflected
ly. ρ is density in kg/m3.
P wave reverse time migration. Its forward position is caused by calcu-
Layer number Vp (m/s) Vs (m/s) ρ (kg/m3) lating travel time with the velocity of S wave. This shows, from one as-
① 340.0 0.0 100.0 pect, that lower inclinations are better for P wave imaging.
② 2000.0 1100.0 1500.0
③ 2400.0 1350.0 1650.0 4.2. Model II results
④ 2800.0 1600.0 1800.0

Fig. 7a and b shows the X and Z components of complicated Model II


of P wave migration in Fig. 4c is stronger than that in Fig. 4a while the generated by a point source at (100,153), respectively. In Fig. 2, a large
energy of S wave migration in Fig. 4d is weaker than that in Fig. 4b, interface can be found with a dip angle of 60°intersecting the axis of
while all of them can indicate the position of the interface. the tunnel at 165 m and a small interface with a dip angle of 90°
Through the analyses above, it is proved that the proposed pre-stack intersecting the axis of the tunnel at 177 m. Fig. 7c and d shows the P
reverse time migration imaging method can provide the correct image wave and S wave reverse time migration imaging profile, respectively.
of geological interfaces with large dipping angles. For interfaces with Fig. 7c indicates two interfaces, of which one is inclined while the
large dipping angles, common-source gather with shots on the tunnel other is a vertical interface located at 77 m ahead, in front of the tunnel.
face is better for reflection P wave imaging while common-receiver It matches the model status. S wave reverse time migration imaging
gather with shots on the side walls can give out a better S-wave imaging (Fig. 7d) includes P wave imaging interference since lower inclinations
result. In the following text, our discussion is based on common-source are better for P wave imaging. Fortunately, this interference can be elim-
gather with shots on the tunnel face. inated by contrasting with P wave reverse time imaging profile (Fig. 7c).
Fig. 5a and b shows the synthetic horizontal (X) and vertical (Z) The analysis reveals that the migration method proposed in this paper
components records for Model I-2 with a point source at (100,153), re- can also image the complicated geology model in tunnels effectively.
spectively. Fig. 5c and d shows the P wave and S wave pre-stack reverse
time migration imaging profile of ModelI-2, respectively. The energy of 4.3. Model III results
P wave migration in Fig. 5c is stronger and more concentrated than that
in Fig. 4c while the energy of S wave migration in Fig. 5d is weaker than Since the tunnel space is relatively limited, all of previous research
that in Fig. 4d. hasn't taken into account the existence of tunnel. Instead, they consider
Fig. 6a and b shows the synthetic horizontal (X) and vertical (Z) it as a homogeneous medium, leading to interferential artifacts of mi-
components records for Model I-3 with a point source at (100,153), re- gration mirrors when imaging dipping interfaces (Chang et al., 2006;
spectively. Fig. 6c and d shows the P wave and S wave pre-stack reverse Gong et al., 2010). We will take P wave imaging to illustrate this phe-
time migration imaging profile of Model I-3, respectively. The energy of nomenon in the following.
P wave migration in Fig. 6c is stronger and more concentrated than that Fig. 8a and b shows the X component and reverse time migration
in Fig. 5c while the energy of S wave migration in Fig. 6d is weaker than imaging profile generated by a point source at (100,153). Without

Fig. 3. Synthetic records of Model I-1. The records are simulated by a finite difference method. (a) and (b) show the synthetic horizontal (X) and vertical (Z) components records of Model I-
1 with a point source on the side wall at (50,153), respectively. (c) and (d) show the synthetic horizontal (X) and vertical (Z) components records for Model I-1 with a point source in the
side wall at (100,153), respectively. The cell size is 250 × 300 and the space-step length is 1 m. The recording length is 250 ms with a sampling interval of 0.01 ms. Labels P and R represent
direct P wave and Rayleigh wave, respectively; PP, PS, SP and SS represent reflected P wave, reflected P-S converted wave, reflected S-P converted wave, and reflected S wave, respectively.
110 F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113

Fig. 4. Reverse time migration imaging of Model I-1. Where (a), (b), (c) and (d) are the reverse time migration imaging results corresponding to (a), (b), (c) and (d) in Fig. 3, respectively.
The dotted line indicates the tunnel axis; the solid line denotes the interface position.

considering tunnel space, the relatively simple records only include di- from reality. No consideration of tunnel space also causes the problem
rect P wave, reflected P wave (PP), and reflected P–S converted wave that the wave field records become symmetrical and the travel time is
(PS) as shown in Fig. 8a. Comparing with Fig. 6a, since the records centered in source point symmetrically as shown in Fig. 8a. This
here do not contain wave field, the results are relatively further different problem leads to a concentrated group, namely artifact, appearing

Fig. 5. Simulated records and reverse time migration imaging profile of Model I-2. (a) The horizontal component (X) record. (b) The vertical component (Z) record. (c) and (d) are the
reverse time migration imaging results corresponding to (a) and (b), respectively.
F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113 111

Fig. 6. Simulated records and reverse time migration imaging profile of Model I-3. (a) The horizontal component (X) record. (b) The vertical component (Z) record. (c) and (d) are the
reverse time migration imaging results corresponding to (a) and (b), respectively.

in the mirror position as the tunnel for axis (Fig. 8b). Obviously, it is conclude that tunnel-based reverse time migration imaging can not
not reasonable to interpret this artifact as interface. In Fig. 6c, this only realize interface imaging correctly but also effectively suppress
shortage is overcome by taking into account tunnel space. So we can mirror artifact.

Fig. 7. Simulated records and reverse time migration imaging profile of Model II. (a) The horizontal component (X) record. (b) The vertical component (Z) record. (c) and (d) are the re-
verse time migration imaging results corresponding to (a) and (b), respectively.
112 F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113

Fig. 8. Simulated records and reverse time migration imaging profile of Model III. (a) The horizontal component (X) record. (b) is the reverse time migration imaging result corresponding
to (a). Label P represents direct P wave. PP and PS represent reflected P wave, reflected P–S converted wave, respectively. Reflection indicates the actual reflected interface; the Artifact
indicates mirror artifact.

5. Case study 6. Conclusions

To verify practicability of the migration method proposed in this Taking into account the tunnel space, more realistic tunnel geologi-
paper, field data acquired from one highway in China is used for reverse cal models are developed by using the first-order coupled elastic
time migration. This area belongs to sedimentary region and layered equations of particle velocity and stress, and high order staggered grid
shale is dominated. The rock mass quality in this working area is poor. finite-difference algorithm to realize numerical simulation of seismic
Fig. 9 shows the sketch map of the tunnel observation system. full-wave fields in tunnel space. Then for simulated records, we utilize
The data shown in Fig. 10a are the original P wave component record reverse time operator based on non-conversion wave equation with
with shot point at the 15th receiver. There are four very developed decoupled P- and S-waves, and excitation time imaging condition to
events in this record (label A, B, C and D, Fig. 10a). At the same time, achieve reliable two dimensional (2D) reverse time migration imaging
we take the P-wave velocity in the surface layer (about 2000 m/s) sur- in tunnel space effectively. Besides, we verify and demonstrate avail-
veyed in the actual test based on the tunnel space as P-wave velocity ability of this method in tunnel field data.
in full-space reverse time migration; after migration and high frequency Since conventional imaging approaches do not consider the impact
noise suppression, we get the profile (Fig. 10b) which indicates four of tunnel space in imaging, wave-field records accorded with reality
events with strong energy (label A, B, C and D, Fig. 10a). are not able to be obtained. Even worse, when it comes to dipping
By calculating the theoretical model, we can see that, under the con-
dition of achieving the velocity in the surface layer, if we use the uni-
form velocity to do migration in the entire space, each energy arc
achieved is corresponded to an interface of physical property, so it can
determine the number and zone of interface. In addition, the position
and dip of first physical interface are relatively accurate, but other inter-
faces of physical properties have some differences in the position and
dip. However, if we can obtain the accurate velocity structure, the mi-
gration method proposed in this paper is able to achieve accurate re-
sults, such as the imaging result of model II in the paper.
According to this, we can do the following semi-quantitative inter-
pretation. In view of the centralized location of energy and strength of
energy, we can see that in Fig. 1(c), there are two relatively strong ener-
gy arcs and the long extension of interface at B and D; hence, we can de-
duce the presence of two relatively large-scale geological interfaces
existing in front of the tunnel face and oblique to the tunnel axis. Energy
arcs A and C are weak and have lower concentration level, which indi-
cates the presence of two small-scale geological interfaces.

Fig. 9. Sketch map of the tunnel observation system. 32 simple component vertical re- Fig. 10. Practical record and reverse time migration imaging. (a) Practical seismic wave
ceivers were set alone the side wall. The receiver intervals are both 1.5 m. The explosive record. The recording length is 400 ms with a sampling interval of 0.0625 ms. (b) Practical
source depth is 1.5 m. reverse time migration imaging.
F. Cheng et al. / Journal of Applied Geophysics 104 (2014) 106–113 113

interface imaging, mirrored artifacts will come out to affect interpreta- Cerjan, C., Kosloff, D., Kosloff, R., Reshef, M., 1985. A nonreflecting boundary condition for
discrete acoustic and elastic wave equations. Geophysics 50 (4), 705–708.
tion. Tunnel-based reverse time migration imaging can not only realize Chang, X., Liu, Y., Gui, Z., 2006. Zero-offset reverse time migration for prediction ahead of
interface imaging correctly, but also effectively suppress mirror artifact. tunnel face. Chin. J. Geophys. 49 (5), 1482–1488.
The proposed pre-stack reverse time migration imaging method can Dickmann, T., Sander, B., 1996. Drivage concurrent tunnel seismic prediction. Felsbau-
Rock and Soil Engineering 14, 406–411.
fulfill the correct image of models including geological interfaces with Gong, X., Han, L., Niu, J., Zhang, X., Wang, D., Du, L., 2010. Combined migration velocity
vertical or large dipping angles (≈90°) ahead of tunnel. When consider- model-building and its application in tunnel seismic prediction. Appl. Geophys. 7
ing tunnel space, common-source gather with shots near the tunnel (3), 265–271.
Inazaki, T., Isahai, H., Kawamura, S., Kurahashi, T., Hayashi, H., 1999. Stepwise application
face is better for reflection P wave imaging while common-receiver
of horizontal seismic profiling for tunnel prediction ahead of the face. Lead. Edge 18
gather with shots on the side walls is better for S wave imaging. For (12), 1429–1431.
common-receiver gather with shots on the tunnel face, the energy of Liu, J., Cheng, F., Fan, C., Cao, J., 2012. Two-dimensional numerical simulation of tunnel-
based seismic full-wave fields. Chin. J. Geotech. Eng. 34 (9), 1705–1711.
P wave imaging increases while the energy of S wave imaging decreases
Lüth, S., Giese, R., Otto, P., Krüger, K., Mielitz, S., Bohlen, T., Dickmann, T., 2008. Seismic in-
as the dip angle of geology interfaces decreases. Namely, interface with vestigations of the Piora Basin using S-wave conversions at the tunnel face of the
large dip angle is beneficial to the image of S wave while one with small Piora adit (Gotthard Base Tunnel). Int. J. Rock Mech. Min. Sci. 45 (1), 86–93.
dip angle is beneficial to the image of P wave. Neil, D., Haramy, K., Hanson, D., Descour, J., 1999. Tomography to evaluate site conditions
during tunneling. 3rd National Conference of the Geo-Institute. American Society of
Civil Engineers, Geotechnical Special Publication, 89, pp. 13–17.
Acknowledgment Petronio, L., Poletto, F., Schleifer, A., 2007. Interface prediction ahead of the excavation
front by the tunnel-seismic-while-drilling (TSWD) method. Geophysics 72 (4),
G39–G44.
We thank anonymous reviewers and editors for their detailed Rechlin, A.J., Lüth, S., Giese, R., 2009. OnSITE: integrated seismic imaging and interpreta-
and constructive comments and editing. This research is supported tion for tunnel excavation. Proceedings of the International Conference on Rock Joints
and Jointed Rock Masses, pp. 1–7.
by the National Natural Science Foundation of China (NSFC), under
Schneider, W.A., Ranzinger, K.A., Balch, A.H., Kruse, C., 1992. A dynamic programming ap-
Grant No. 41202223. proach to first arrival traveltime computation in media with arbitrarily distributed
velocities. Geophysics 57 (1), 39–50.
Stefan, J., Thomas, B., Denise, D., 2010. On the propagation characteristics of tunnel
References
surface-waves for seismic prediction. Geophys. Prospect. 58 (245), 256.
Stefan, J., Thomas, B., Kurzmann, A., 2011. Seismic prediction of geological structures
Bohlen, T., Lorang, U., Rabbel, W., Müller, C., Giese, R., Lüth, S., Jetschny, S., 2007. Rayleigh-
ahead of the tunnel using tunnel surface waves. Geophys. Prospect. 59 (934), 946.
to-shear wave conversion at the tunnel face—from 3D-FD modeling to ahead-of-drill
Tzavaras, J., Buske, S., Groß, K., Shapiro, S., 2012. Three-dimensional seismic imaging of
exploration. Geophysics 72 (6), T67–T79.
tunnels. Int. J. Rock Mech. Min. Sci. 49, 12–20.
Borm, G., Giese, R., Otto, P., Dickmann, T., Amberg, F., 2001. Integrated seismic imaging
Zeng, Z., 1994. Prediction ahead of the tunnel face by the seismic reflection methods.
system for geological prediction ahead in underground construction. In 2001 Rapid
Chin. J. Geophys. 37 (2), 218–230.
Excavation and Tunneling Conference, pp. 263–271.
Zhang, D., 2009. Study on cross-well seismic elastic wave reverse-time migration imag-
Borm, G., Giese, R., Klose, C., Mielitz, S., Otto, P., Bohlen, T., 2003. ISIS—integrated seismic
ing. PhD thesis, China University of Geoscience (Wuhan), unpublished results.
imaging system for the geologic prediction ahead in underground construction. 65th
Annual Conference and Exhibition, EAGE, Extended Abstracts.