www.elsevier.com/locate/epsl
Earth and Planetary Science
Seismic structure of the Bohai Bay Basin, northern China:
Implications for basin evolution
Liang ZhaoT, Tianyu Zheng1
Institute of Geology and Geophysics, Chinese Academy of Sciences, Qijiahuozi, Deshengmenwai, Chaoyang District, P.O. Box 9825,
Beijing 100029, PR China
Received 1 July 2004; received in revised form 28 November 2004; accepted 20 December 2004
Available online 1 February 2005
Editor: R.D. van der Hilst
Abstract
As part of an extensive seismic research program, 33 portable broadband seismic stations were deployed along a line
crossing the Bohai Bay Basin, northern China. Three teleseismic events were selected to constrain the seismic structure along a
~280 km profile across the western edge of the basin. We determined the basin structure that described the observed shear
horizontal (SH) wave field. The synthetic SH wave was calculated using a finite difference (FD) method with its computational
domain localized in the basin area and input motions at the base of the model extrapolated from the displacement recorded at a
nearby hard-rock station. Synthetic seismographs calculated for the models match the observations well in both waveform and
travel time. Numerical tests indicate that the structural features of the preferred models are well resolved. The analysis of
relations between structures and stratigraphic units along the cross sections allows multiple deformational events in the basin to
be inferred. In conjunction with a profile across the southern edge that has been presented previously by Zhao et al. [14] [L.
Zhao, T.Y. Zheng, W.W. Xu, Modeling the Jiyang depression, Northern China, using a wave field extrapolation FD method and
waveform inversion, Bull. Seismol. Soc. Am. 94 (2004) 988–1001], the results reveal basin-wide extension with local inversion
features in the Bohai Bay basin.
D 2005 Elsevier B.V. All rights reserved.
Keywords: Cenozoic; SH wave modeling; basin seismology; sedimentary basin structure
0012-821X/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2004.12.028
T Corresponding author. Tel.: +86 10 62007354; fax: 86 10
62010846.
E-mail addresses: [email protected] (L. Zhao)8
[email protected] (T. Zheng).1 Tel.: +86 10 62363458; fax: +86 10 62010846.
1. Introduction
The Bohai Bay basin, located in the North China
block [1,2], is a large oil-producing basin that under-
went compound and complex tectonic events during
its formation and development. In the last two
decades, a number of papers (e.g., [3–9]) have been
published on the evolution of the basin.
Letters 231 (2005) 9–22
Fig. 1. (a) Location map of the studied area. The lines indicate three cross sections. AAV is modeled in this paper; BBV has been modeled by
Zhao et al. [14]; and RRV is a geological cross section modified from Ye et al. [5]. (b) Location map of stations (solid triangles) for cross section
AAV. The stars on the inset represent locations of seismic events, and the solid lines represent great circle paths.
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2210
Table 1
Event list
Event Origin, UT Latitude Longitude Depth
(km)
Mw
020303 3 Mar. 2002, 12:08:19 70.488 36.508 225 7.4
020325 25 Mar. 2002, 14:56:33 69.328 36.068 8 6.2
020412 12 Apr. 2002, 04:00:23 69.428 35.968 10 5.9
Epicentral location is given in degrees. Event data is from the USGS
on web.
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–22 11
The basic knowledge of the Bohai Bay basin is
mainly from surface geological investigation and
seismic exploration (e.g., [7]). During late Permian
time, the North China block and the South China
block collided. After the collision, Eastern China
experienced widespread tectono-thermal reactivation
during the Late Mesozoic and Cenozoic, as indi-
cated by emplacement of voluminous late Mesozoic
granites and extensive Cenozoic volcanism [10].
The rifting of the eastern China took place between
late Cretaceous and Oligocene [3,11]. As a result,
many basins, including the Bohai Bay basin, formed
across a vast area of eastern China. It is generally
accepted that the most important tectonic control on
extension was probably the subduction of the
Pacific plate in the eastern margin of Asia (e.g.,
[12,13]).
Some basic images of the Bohai Bay basin remain
ambiguous, including the tectonic character of the
basin-mountains edge and the deep faults cutting
through the basement of the basin, which are of great
importance in understanding the evolution of the
basin. Therefore, further investigation is necessary to
image the basin edge structure and the basement of the
basin.
An effective approach to image the basin edge
structure and the basement of the basin is to use the
seismic wave propagating through the basin. With
this target in mind for the last 3 yr, we participated
in the Northern China Interior Structure Project
(NCISP). This project deployed dense broadband
portable seismograph arrays across the Bohai Bay
basin and the surrounding uplifts. Some of the
objectives of this project are: (1) to improve our
understanding of the seismic and geological struc-
tures of the Bohai Bay basin, and regional tectonics;
(2) to study deep seismic structures beneath the
North China Plate.
In this paper we describe the seismic structure
along a new cross section crossing the western edge of
the basin obtained by waveform inversion of the
seismic data using a hybrid method that consists of
shear horizontal (SH) wave field extrapolation, finite
difference (FD) calculation and travel time and
waveform inversion [14]. The formation of the basin
was then interpreted based on analysis of relations
between structures and stratigraphic units inferred
from our preferred seismic model.
2. Seismic structure of cross section AAV
2.1. Seismic dataset
We model the basin edge structures from a cross
section AAV (Fig. 1a). This cross section orients
NNW–SSE direction and transects the western edge
of the Bohai Bay Basin. For section AAV (See Fig.
1b), two types of portable broadband seismograph
units were deployed in an approximately 280 km
long transecting line with an average station spacing
of ~10 km. Seismic data recorded at 29 stations were
selected for the modeling. 18 of these 29 stations
were deployed with REFTEK data loggers and
Guralp CMG-3ESP sensors (50 Hz–30 s). The other
instruments were CAS DS24-3 data loggers and
BKD-2 broadband sensors (40 Hz–20 s). We name
the stations differently according to their recording
sensors and data logger. Stations with the Guralp
system are named as station number+abbreviated
address, while those with the CAS system are named
as station number. Data were recorded at a rate of 40
samples per second for all the stations. Seismic data
from three seismic events were selected for modeling
the seismic structure along section AAV (shown in
the inset of Fig. 1b). All observations are almost in
an identical azimuth along the profile AAV. The
event magnitudes (Mw) are all z5.5. Table 1 lists the
event parameters. Event locations and origin times
are taken from the USGS catalog on the web (http://
www.neic.usgs.gov). Zero-phase band-pass filters are
applied with a typical band-pass frequency range of
0.05–4.00 Hz.
2.2. Methodology
We employ the technique introduced by Zhao et al.
[14] to model the shear wave velocity structure of the
Fig. 3. Observed tangential displacements for event 020325
Stations from 181DC to 187NT were deployed at hard-rock sites
outside the basin, while stations from 154XW to 180DM were
inside the basin. Note that the waveforms recorded at the hard-rock
sites are very similar, while those recorded inside the basin are
noticeably different.
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2212
western edge of the basin. Fig. 2 schematically
illustrates the principle of this method. The method
consists of forward synthetic FD calculations, wave-
form and travel time inversions [15]. We briefly
review the method here. Readers are referred to Zhao
et al. [14] and Wen [15] for details of the method.
For the in-plane propagation of the teleseismic SH
wave, we assume that the bbasin site responseQ is themost significant cause of the waveform difference
between stations outside and inside the basin, while
the path effects from the earthquake source to the base
of the basin are similar. The validity of this
assumption can be verified with the observations.
Fig. 3 shows that the tangential displacements
recorded at hard-rock stations (182DL to 187NT)
are in excellent agreement with each other, whereas
those recorded inside the basin (station from 154XW
to 181DC) are evidently different. Therefore, it is
reasonable to extrapolate the displacement recorded at
a nearby hard-rock station to the bottom of the FD
region as input motions of the FD calculation (e.g.,
P0, P1, P2 in Fig. 2). In the FD calculation, the basin
model vector is velocity as a function of position
m(x)=m(x, z), and a grid size of 0.15 km is adopted to
discretize the model.
Fig. 2. Schematic illustration of the principle of the SH wave field extrapolation FD method [14]. The basin is confined inside a rectangle, where
the FD method is applied. When a hard-rock site station P is sufficiently far away from the source, the SH-wave recorded at P can be
extrapolated to the bottom interface of the FD region (e.g., P0, P1, P2) as input motions in the FD calculations. The solid circles represent the
control depth points of the basin model, and the dashed lines schematically indicate FD grid lines.
.
Table 2
Average velocity assigned for the stratified sediments and the
basement rocks [14]
Strata Vp
(km/s)
Vs
(km/s
Quaternary (Q) 1.32 0.60
Neogene (N) 2.20 1.25
Paleogene (E) 2.90 1.70
Pre-Tertiary (P-T) 3.80 2.20
Basement rocks 5.50 3.20
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–22 13
A waveform and travel time inversion method is
applied to invert the basin structure (e.g., Ji et al.
[16]). The waveform misfit is defined as square norm
of the difference between the synthetic and recorded
seismograms
f sð Þ ¼ � 12Rp tð Þobs p t þ sð ÞsyndtR
p tð Þ2obs þ p t þ sð Þ2synh i
dt; ð1Þ
where p(t)syn and p(t)obs are the synthetic and
recorded seismograms, respectively, and s is the time
shift between the two.
The misfit function between observation and
synthetic is represent as [17]
E m xð Þð Þ ¼ Wt
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
N � 1
XNi¼2
dsi � dslð Þ2vuut
þWf
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
N
XNi¼1
f 2i dsiÞ;ð
vuut ð2Þ
where N is the number of stations, dsl is the retardedtime between synthetic and record at the origin point
of FD region, Wt and Wf are weights for travel time
residual and waveform fit, respectively. Based on our
data quality, we assign Wt as 0.3 and Wf as 0.7.
In the inversion, the basin models are parameter-
ized as isovelocity layers with linearly dipping seg-
ments represented with control depth points (marked
as solid circles in Fig. 2). Considering the spacing of
our stations and a reasonable inversion calculation
Fig. 4. The initial model constructed from approximate teleseismic travel time residuals with respect to IASP91 for event 020325. The SH trave
time residuals were represented by solid circles plotted at the locations of corresponding receivers, with their diameters proportioning to the
amplitude of the time residuals. The scales of the time residual are plotted at the bottom.
)
time, we adopt a horizontal spacing of 5 km between
control depth points in the initial inversion calcu-
lation. Then, we employ the conjugate gradient
algorithm [18] to search for the geometry of the strata
interfaces that minimizes the misfit defined in Eq. (2).
In the case that higher resolution is needed to
constrain some local structures, we add more control
points or adjust the spacing between control points to
represent the local structure. Since we adopt a grid
size of 0.15 km to discretize the model, the minimal
space between control depths could be reduced to
0.15 km if necessary.
2.3. Starting model
Following Zhao et al. [14], we divided our basin
model into four major isovelocity layers, correspond-
ing to the Quaternary, Neogene, Paleogene and Pre-
Tertiary strata, respectively. The average shear veloc-
ities (Table 2) of the layers are obtained from the
compressive wave velocities and the Vp/Vs ratios of the
l
Fig. 5. Best-fitting models for cross section AAV obtained based on waveform inversion of teleseismic data (top panels), and comparisons of
synthetic displacements (dashed traces) and seismic data (solid traces) (bottom panels) for three events: (a) event 020203, (b) event 020325 and
(c) event 020412. The best-fitting model for event 020203 has a shorter length than the other two, since we have only good records from station
181DC to 165JD for event 020203. Waveform is normalized by its maximum amplitude. The black solid circles indicate the depth control points
in describing the basin structure. Black solid triangles represent the stations used in the modeling, with the numbers above indicating the index
of stations. The input motions to the FD calculations are extrapolated based on the seismic data observed at a hard-rock site station 182DL
(located at distance 0 km in the model).
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2214
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–22 15
four strata averaged from the drilling core and sonic
data [14]. The shear wave velocity of the basement
rock is assigned to be 3.20 km/s. The station 182DL is
chosen as the hard-rock station whose tangential
displacement is used for wave field extrapolations.
An initial geometry of the basin (Fig. 4) is first
constructed based on the travel time difference
between the observed first arrival (from event
020325) and theoretical predictions based on IASP91
[19]. In the initial model, four layers are assigned with
same thickness, and the depth of the basin beneath the
stations satisfies the following equation
X4l¼1
hi
4vl� hi
m basement
¼ si; ð3Þ
where si is the travel time difference between the
observation and prediction for station no. i, hi is the
Fig. 6. (a) The preferred model for cross section AAV, and (b) a geologicallines represent faults, and the strata are illustrated at the bottom of the fig
total depth of the basin beneath station no. i, ml is theaverage shear wave velocity of the lth layer, and
mbasement is the average shear velocity of the basement
rocks.
2.4. Results
Beginning with the initial model, we model the
observed waveforms from the three events listed in
Table 1 and invert the best-fitting velocity structure
models along cross section AAV. The best-fitting
models obtained from inverting seismic data recorded
for the three events are shown in Fig. 5, and the
synthetics (dashed lines) from the best-fitting models
are compared with the seismic data (solid lines).
Overall, the relative timing and waveform match the
data well, except for several stations (for example,
station 163). We note that, although the input motions
cross section RRV [5] in the vicinity of AAV (see Fig. 1). The dashedure.
Fig. 7. Numerical test for a model without structural features
beneath stations from 175 to 172. (a) Smoothed model, which is
obtained by smoothing the interface undulance of the best-fitting
model for event 020325 in Fig. 5a with the fault beneath stations
from 175 to 172. (b) Comparison between data (solid traces) and
synthetics for the smoothed model (dashed lines). The main misfits
between them are marked with black arrows. (c) Comparison
between data (solid traces) and synthetics for the preferred mode
(dashed lines).
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2216
at the hard-rock site stations are quite different for the
three events, the models inverted from the seismic
data for the three earthquakes are very similar, which
confirms the validity of our results. A preferred model
was obtained along cross section AAV (Fig. 6a) by
averaging the models obtained from the three event
data. To evaluate our results, we compare our final
inverted model with a geological cross section RRV[5] (Fig. 6b, its location is shown in Fig. 1a) in the
vicinity of section AAV. The overall agreement
between them confirms the reliability of our results.
Note that, the preferred model and the geological
cross section RRV resemble the initial model in
general, implying that travel time residuals place
strong constraints in waveform inversion.
2.5. Resolution tests
Our teleseismic data have frequency content from
0.05 to 1 Hz. The resolution test of Zhao et al. [14]
has illustrated that this frequency content would allow
detection of a 1.0 km high and 10 km wide local rise.
Generally, such a local rise would increase waveform
misfits of 5–20% and travel time residuals of 0.2–0.5
s (defined in Eqs. (1) and (2)) for the receivers above
the local rise.
The preferred model reveals an image of large-
scale velocity structure, in which structural features
such as rises and depressions are resolved. To evaluate
the reliability of these structural features, we present a
series of numerical experiments.
The first resolution test focuses on the structural
features using the data from event 020325. To
illustrate we take a local rise beneath stations 175ZC
to 172 as an example. We assume a modified model,
in which the interface undulance (beneath station
175ZC to 172) was smoothed as shown in Fig. 7a.
Forward calculation using the smoothed model
produces significant discrepancy between the syn-
thetics and the data for the associated stations (Fig.
7b). Compared to the result from the preferred model
(Fig. 7c), the smoothed model produces time residual
increasing from �0.05 s to 0.75 s and waveform
misfit increasing by about 48% from 0.25 to 0.37 for
station 175. For station 174 the time residuals increase
from 0.05 s to 0.98 s and waveform misfit increases
by about 100% from 0.17 to 0.34. When we invert
basin models from the initial model constructed from
lFig. 8. Waveform sensitivity to different features in the inverted basin model for cross section AAV. Strata are added from (a) to (d) step by step.
Bottom panels show comparisons between the data (thick lines) and synthetic waveforms (dashed lines) generated from themodels above. The time
residual and waveform misfit values corresponding to each of the steps are marked above the waveforms. The solid triangles indicate receivers.
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–22 17
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2218
the travel time difference, the inversions converge to
the optimal model after 32 iterations. In contrast, if we
forced the structure beneath stations 175 to 172 to be a
smooth shape in inversions, we are not able to obtain
a better fitting result after 100 iterations. This means
we could not obtain a better result if we used the
model without the structural features, thereby it
illustrates the robustness of the existence of the
undulance in the strata boundaries beneath stations
175ZC to 172.
The second resolution test is to illustrate the physical
relationship between the model features and the
observed motions. We take the inverted model beneath
stations 182DL to 166ZF as an example.We start with a
thin surface stratum and add strata step by step. Fig. 8
shows comparison between synthetic waveforms
(dashed lines) and the observed data for event
020325 (listed in Table 1). The results illustrate that
different stratum affect synthetics differently. Fig. 8a
shows large discrepancy between synthetics (dashed
Fig. 9. Relationship between inferred basin structural features and faulting
motion of hanging wall and its normal faults interpretation, and (b) an upw
candidate interpretations: back-thrust faults (left lower panel) or normal
corresponding interpretation model. bNQ and bEQ indicate the strata, the dasof the motions between the blocks.
line) and data when only a thin surface layer is present.
Adding more strata (Fig. 8a–d) improves the fit to both
the timing and waveform (misfit values corresponding
to each step are marked in Fig. 8). The intermediate
strata have strong effects on the timing and waveform
of the synthetics, but have less effect on the reflected
phases (Fig. 8b,c). The shape of the basement interface,
which has a strong effect on the reflected phases, plays
an important role in fitting the waveform.
3. Interpretations and discussions
3.1. Inferred tectonic features
We infer the characteristics of the faults inside the
basin based on the preferred seismic velocity model
obtained from the waveform and travel time inver-
sions. Fig. 9 schematically illustrates our approach
and assumption. As can be seen from Fig. 9a, we can
in the basin for two types of faults: (a) a structure with downward
ard arc-shape interface with low undulance (upper panel) and its two
faults (right lower panel). The misfit values are marked below the
hed lines represent faults, and the solid arrows denote the directions
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–22 19
readily infer that a downward motion of hanging wall
probably indicates a normal fault associated with
extension. However, it seems that there are two
possible interpretations for the upward arc-shape
interface with low undulance in Fig. 9b (it corre-
sponds to the Jizhong depression in Fig. 6a): one is a
back-thrust structure and the other is a normal-faults
structure. To check which interpretation is better, we
compare the misfit values of the two different
interpretation models by using forward calculation.
The back-thrusts model produces an average time
residual of 0.48 s and waveform misfit of 0.30, while
the normal faults model produces an average time
residual of 0.55 s and waveform misfit of 0.34. The
result shows that synthetics from the back-thrusts
model match the observation better than from the
normal faults. Based on this, we prefer the back-thrust
faults interpretation.
Fig. 10. Geological interpretation inferred from the preferred models for cro
was modeled in Zhao et al. [14]. The dashed lines represent faults, and the s
The strata are illustrated at the bottom of the figure.
Fig. 10a shows a possible interpretation of the
structural features in the preferred models obtained for
cross sections AAV. To obtain a general image of the
basin structure, we also interpret a previously modeled
cross section BBV [14] (the location of BBV is shown
in Fig. 1a) transecting the southern edge of the basin.
Both cross sections were cut through by a series of
large normal faults, which suggest that extension was
the dominant effect in controlling the evolution of the
basin.
This extension has actually produced relatively
parallel normal faults with extremely low-lying
grabens, i.e., the extension has broken up the basin
into several sub-basins. For cross section AAV, themajor normal faults contain F1, F2, F5, F6 and F7.
Controlled by these major faults, the west edge of the
basin is divided tectonically into alternating grabens
and horsts corresponding to the Jizhong depression,
ss sections AAV (a) and BBV (b) (see Fig. 1). The cross section BBVolid arrows denote the directions of the motions between the blocks.
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2220
the Cangxian horst, the Huanghua depression and the
Chenning horst going from west to east (e.g., [2,5,6]).
For cross section BBV, the major faults contain F11 to
F15 that have smaller throws compared to the faults in
AAV. The major graben in BBV is named the Jiyang
depression. For these sub-basins, except the Jizhong
depression, the characteristic structures in the other
depressions exhibit a series of similar extensional
pattern (normal fault system, tilted blocks). Together
with the major faults, the smaller faults such as F3, F4
and so on likely controlled the development of the
sub-basins.
The normal fault F1 is the primary fault with the
greatest throw of about 35 km in our two cross
sections. It deformed all the sediment strata and the
basement, and it is coincident with the previously
proposed location of the Taihang Shan piedmont Fault
(e.g., [7]). Trending NNE–SSW with length over 500
km [5], this steep fault F1 forms the western boundary
of the basin and separates the basin from the Taihang
Shan uplift. It is probably a major detachment surface
and was active during the Cenozoic extension.
Extension is the general character of the basin.
However the two cross sections show certain differ-
ences. For instance, compared with the marginal faults
in the southern edge of the basin, the Taihang Shan
piedmont fault (F1) has steeper scarps and greater
throws. This spatial difference of the faults that belong
to different ages probably implies that the basin
underwent multiply rifts with varying extensional
amplitudes.
3.2. Timing of Cenozoic extension
The timing of deformation in the basin is con-
strained by the age of the units involved in the
deformation [20]. The major normal faults and the
angular unconformity between strata not only exhibit
characteristics of growth faults but also imply the
timing that the faults played a role in extension.
For example, normal faults F1 and F2 deformed the
Pre-Tertiary strata and the Paleogene strata. Moreover,
the Paleocene strata become thicker in the Jizhong
depression than in the Cangxian horst. These obser-
vations suggest that: (1) faults F1 and F2 were active
and controlled the sediments during early Paleocene;
(2) the Jizhong depression received sediments asso-
ciated with the extension of the basin. Since the oldest
syn-rift deposits belong to the Paleogene strata for
both of the edges, we can infer that the triggering time
of extension was not later than early Paleocene.
Extension continued during Neogene and/or Quater-
nary due to the fact that most of the normal faults cut
through the Neogene strata, and some of them, such as
F2 and F3, even cut through the Neogene strata to the
Quaternary strata. However, extensional deformation
occurred on a much smaller scale than the Paleogene
strata, with most faults either entirely inactive since
the early Neogene or showing only minor amounts of
Neogene extension in comparison with their early
Tertiary histories [2]. This indicated that the major
extension in the most of areas ended at the end of
Paleogene.
The Paleogene and Neogene strata form the major
reservoir rocks of the Bohai Bay basin; however, the
major syn-rift deposits in the different depressions
belong to different strata. In the piedmont zones of the
Taihang Shan (e.g., the Jizhong depression) and the
Luxi uplift (e.g., the Jiyang depression), the Paleogene
strata are thick. Yet, in the Huanghua depression, the
Paleogene strata become thinner while the Neogene
strata become thicker. According to thickness change
of the Paleogene strata and the Neogene strata, it can
be inferred that the depocenter migrated from the
flanks to the center resulting in that the sediments
became progressively younger. As rifting continued
and the basin widened, the locus of rifting changed
from the flanks to the center of the basin [3].
3.3. Possible local compressional structures
Two localized back-thrust faults (C1 and C2 in Fig.
10) and doming folds are recognized in the Jizhong
depression. The hanging walls back thrusted onto the
footwall due to the local tectonic shortening. The
cross section image (Fig. 10a) suggests a minimum of
22% shortening along this cross section, calculated by
comparing line length change between deformed and
undeformed sections. These reverse faults deformed
all sedimentary sequences and basement rocks,
possibly providing evidence that the compressional
deformation took place during Neogene and Quater-
nary [5]. It was also reported that Mid-Tertiary
thrusting exists within the western part of the Liaohe
Depression, the northern Bohai Bay basin [21]. It is
possible that some of the thrust and reverse faults are
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–22 21
inverted Tertiary normal faults. However, it is
important to note that such kind of compressional
pattern is only local and is not basin-wide. For
example, the compression is extensive in the Jizhong
depression the western edge, but is insignificant in the
Huanghua depression and the Jiyang depression.
4. Conclusions
Using an SH wave field extrapolation FD method
combined with travel time and waveform inversion,
the western edge shear velocity structures of the Bohai
Bay basin are constrained by teleseismic data from 29
portable seismic stations. Comparison between syn-
thetics and the observed data shows good agreements.
The overall similarity between our preferred model
and a nearby geological profile confirms the reliability
of our results. Distinct characters of basin-wide
extension and local inversion are recognized in our
cross sections. The most striking evidence of exten-
sion relate to a series of extensional pattern such as
normal fault system, horst and graben. The results
presented in this paper support the assumption that the
Bohai Bay basin was formed after early Cenozoic
extension and late Cenozoic thermal subsidence and
inversion (e.g., [2–7]). It appears that extension was
the main mode during the basin evolution, accom-
panied by local compression and some renewed
extensional movement along faults.
In addition, the method employed in this paper
provides an example to constrain the basin edge
structures by applying portable seismic observations.
Fundamental tectonic questions of basins such as
tectonic history of basin formation and development
can be addressed based on the basin edge structure
images. This method would probably find wider
applications with increasing number of stations
deployed inside basins, especially in the cases when
the detailed seismic reflection data is lacking.
Acknowledgements
The review of Qian Song improved our first
manuscript. We thank Dr. Lianxing Wen for his FD
calculation code, his constructive advice and most
helpful reviews. We acknowledge the participants of
the Broadband Seismic Array Laboratory, IGGCAS.
This research is supported by Chinese Academy of
Sciences (No. KZCX 1-07) and China Earthquake
Data Sharing (2003-DZGX-2004). We appreciate the
proofreading of Dr. M. Hill, the careful and insightful
reviews by Dr. R. D. van der Hilst, Dr. T. Hearn and
Dr. Eric Hetland.
References
[1] G.Y. Li, M.G. Lu, Atlas of China’s Petroliferous Basins, 2nd
ed., Petroleum Industrial Press, Beijing, 2002 (in Chinese).
[2] K.Z. Lu, J.F. Qi, Tectonic Model of Cenozoic Petroliferous
Basin Bohai Bay Province (in Chinese), Geological Publishing
House, Beijing, 1997.
[3] J.Y. Ren, K. Tamaki, S.T. Li, J.X. Zhang, Late Mesozoic and
Cenozoic rifting and its dynamic setting in Eastern China and
adjacent areas, Tectonophysics 344 (2002) 175–205.
[4] M.B. Allen, D.I.M. Macdonald, X. Zhao, S.J. Vincent, C.
Brouet-Menzies, Early Cenozoic two-phase extension and late
Cenozoic thermal subsidence and inversion of the Bohai Bain,
northern China, Mar. Pet. Geol. 14 (1997) 951–972.
[5] H. Ye, K.M. Shedlock, S.J. Hellinger, J.G. Sclater, The north
china basin: an example of a Cenozoic rifted intraplate basin,
Tectonics 4 (1985) 153–169.
[6] S.J. Hellinger, K.M. Shedlock, J.G. Sclater, H. Ye, The
Cenozoic evolution of the north China basin, Tectonics 4
(1985) 343–358.
[7] Z.Y. Zhao, B.F. Windley, Cenozoic tectonic extension and
inversion of the Jizhong Basin, Hebei, northern China,
Tectonophysics 185 (1990) 83–89.
[8] C. Wang, X. Zhang, Q. Wu, Z. Zhu, Seismic evidence of
detachment in north China basin (in Chinese), Chin. J.
Geophys. 37 (1994) 614–620.
[9] C. Wang, X. Zhang, Z. Lin, Q. Wu, Y. Zhang, Crustal structure
beneath the Xingtai earthquake area in North China and its
tectonic implications, Tectonophysics 274 (1997) 307–319.
[10] A. Yin, S.Y. Nie, Phanerozoic palinspastic reconstruction of
China and its neighboring regions, in: Yin An, T.M. Harrison
(Eds.), The Tectonic Evolution of Asia, Cambridge University
Press, 1996, pp. 442–485.
[11] X.Y. Ma, H.F. Liu, W.X. Wang, Y.P. Wang, Rifting and
extensional tectonics of Meso-Cenozoic in East China (in
Chinese), J. Geol. 57 (1983) 22–25.
[12] M.P. Watson, A.B. Hayward, D.N. Parkinson, Z. Zhang, Plate
tectonic evolution, basin development and petroleum source
rock deposition onshore China, Mar. Pet. Geol. 4 (1987)
205–225.
[13] Z.Y. Tian, P. Han, K.D. Xu, The Mesozoic–Cenozoic east
China rift system, Tectonophysics 208 (1992) 341–363.
[14] L. Zhao, T.Y. Zheng, W.W. Xu, Modeling the Jiyang
depression, Northern China, using a wave-field extrapolation
FD method and waveform inversion, Bull. Seismol. Soc. Am.
94 (2004) 988–1001.
L. Zhao, T. Zheng / Earth and Planetary Science Letters 231 (2005) 9–2222
[15] L.X. Wen, An SH hybrid method and shear velocity structures
in the lowermost mantle beneath the central Pacific and South
Atlantic Oceans, J. Geophys. Res. 107 (2002).
[16] C. Ji, D.V. Helmberger, D. Wald, Basin structure estimation by
waveform modeling: forward and inverse methods, Bull.
Seismol. Soc. Am. 90 (2000) 964–976.
[17] Y. Luo, G.T. Schuster, Wave-equation travel-time inversion,
Rev. Geophys. 56 (1991) 645–653.
[18] E. Polak, Computational Method in Optimization, Academic
Press, New York, 1971, pp. 44–65.
[19] B.L.N. Kennett, E.R. Engdahl, Traveltimes for global earth-
quake location and phase identification, Geophys. J. Int. 105
(1991) 429–465.
[20] B.J. Darby, B.D. Ritts, Mesozoic contractional deformation in
the middle of the Asian tectonic collage: the intraplate Western
Ordos fold-thrust belt, China, Earth Planet. Sci. Lett. 205
(2002) 13–24.
[21] T.H. Wang, Genetic types of thrust faults in Eastern China
petroliferous regions, Earth Sci. 13 (1988) 627–634.