Prediction of Marine Thin Shale Gas Reservoir With
Prediction of Marine Thin Shale Gas Reservoir With
Article
Prediction of Marine Thin Shale Gas Reservoir with Seismic
Phase-Controlled Nonlinear Stochastic Inversion
Qingming Xie 1,2, * , Yanming Wu 1 , Qian Huang 1 , Yunbing Hu 1 , Xiaoliang Hu 1 , Xiaozai Guo 2 , Dongming Jia 2
and Bin Wu 3
Abstract: Due to the complicated interference sources and low signal-to-noise ratio of seismic records,
conventional seismic inversion methods are difficult to accurately identify shale gas reservoirs
with a thickness of less than 10 m. This presents great challenges to shale gas exploration and
development in China. Seismic phase-controlled nonlinear stochastic inversion (SPCNSI) is related to
the heterogeneity of underground media. With the constraints of the stratigraphic sequence or seismic
facies models, the minimum value between the seismic model and seismic record can be solved
through iterative processes. Based on the solved acoustic velocity in formation, the constraints for
SPCNSI can be formed with the matching relationship between target layers and different sequences
in three-dimensional space. The prediction resolution of an unconventional reservoir can be effectively
improved by combining logging, geological and seismic information. The method is suitable for
predicting thin shale gas reservoirs in complex geological structures. In this study, SPCNSI was
developed to predict the thin-layer marine shale gas reservoir in the southeast of Chongqing; the
horizontal and vertical distribution characteristics were proven in the Longmaxi Formation and the
Wufeng Formation; thin layers with a thickness of less than 7 m were also discovered. According to
Citation: Xie, Q.; Wu, Y.; Huang, Q.;
the results, three sets of potential development areas were determined by the inversion method, and
Hu, Y.; Hu, X.; Guo, X.; Jia, D.; Wu, B.
Prediction of Marine Thin Shale Gas
the accuracy and reliability of the method were verified by logging and productivity testing.
Reservoir with Seismic
Phase-Controlled Nonlinear Keywords: nonlinear stochastic inversion; reservoir prediction; seismic facies; shale gas; thin layer
Stochastic Inversion. Processes 2023,
11, 2301. https://doi.org/10.3390/
pr11082301
1. Introduction
Academic Editors: Shangwen Zhou,
Hongyan Wang and Jian Xiong Shale gas is a kind of clean energy. The recoverable resources of shale gas in China
amount to 21.8 × 1012 m3 , ranking first in the world. For increasing the supply of
Received: 14 June 2023 clean energy and reducing carbon emissions, annual natural gas production will exceed
Revised: 19 July 2023
2300 × 108 m3 by the end of 2025, according to the plan of “China’s 14th Five Year Plan
Accepted: 22 July 2023
for Modern Energy System”. With the orderly promotion of shale gas exploration and
Published: 1 August 2023
development, the Fuling shale gas field, the main battlefield of shale gas development in
China, has cumulatively produced more than 532 × 108 m3 as of December 2022, which
has effectively decreased China’s dependence on external sources of natural gas.
Copyright: © 2023 by the authors.
Seismic exploration is one of the key technologies for shale gas development. The
Licensee MDPI, Basel, Switzerland. seismic inversion of shale gas reservoirs is mainly divided into three ways: acoustic
This article is an open access article impedance inversion based on seismic data, model-based logging attribute inversion, and
distributed under the terms and geostatistics-based stochastic simulation and stochastic inversion [1,2]. Acoustic impedance
conditions of the Creative Commons and velocity inversion is one of the most effective ways for identifying underground
Attribution (CC BY) license (https:// lithologies, and has become an important technology for reservoir inversion and the
creativecommons.org/licenses/by/ prediction of shale gas. In practical applications, acoustic impedance inversion based on
4.0/). seismic data mainly include recursive inversion [3–5], constrained sparse pulse inversion [6],
reservoir characteristic attribute inversion [7–9] and frequency division inversion [10]. The
commonly used constrained sparse pulse inversion is less affected by the initial model,
and can accurately reflect the lateral changes of the reservoir; however, the resolution
is low. Geostatistical inversion introduces the idea of stochastic simulation to seismic
inversion, and uses stochastic simulation to derive the attribute data of bodies for reservoir
prediction [11,12]. A wavelet-based multiple-point geostatistical simulation (WAVESIM)
algorithm was proposed to reduce the uncertainty in predicting the spatial arrangements
of litho-facies away from wells or control points, which integrates the physical properties
of litho-facies, and geophysical data within a multiple-point geostatistical framework [13].
Random simulations based on geological statistics can better reflect the heterogeneity
of a reservoir, and are less affected by the initial model of the seismic wave velocity.
Zhang et al. [14] comprehensively utilized seismic, geological, logging data and geological
knowledge to perform wave impedance random inversion and lithology simulations on
the spatial distribution of reservoirs, and predicted the distribution patterns. Li et al. [15]
proposed a block constrained generalized wave impedance inversion method based on
the analytical solution of the acoustic equation, which extracts stable velocity and density
parameters by partially stacking profiles to invert the generalized acoustic impedance that
varies with the incident angle. Block constraints are introduced within the framework
of Bayesian theory to obtain stable and well-defined inversion results. Based on seismic
sedimentology, the method of grouping the optimal azimuth and offset stacking window to
partially stack OVT gathers effectively identifies channel sand bodies and makes reservoir
predictions that cannot be predicted using partial full stacks [16].
Many studies have reported the application of nonlinear inversion methods in solv-
ing wave equations. The geological structure-guided hybrid Markov Chain Monte Carlo
(MCMC) and Bayesian linearized inversion (BLI) methodologies for seismic inversion are
implemented to address the problem of 2D discontinuous seismic profiles, because the
spatial coupling of model parameters is not considered [17]. Yang [18,19] proposed the
nonlinear chaotic inversion for seismic traces by the phasing and quantitative description
of state changes of successive linearization iterations. Combined with nonlinear optimiza-
tion [20] and a random model, Huang et al. [21] proposed nonlinear random inversion
controlled by the seismic phase, which had high vertical resolution and relatively accurate
prediction results with the advantages of both broadband constrained inversion [22] and
model inversion [23]. The method was also used to track biological reefs and shoals by
establishing a recognition model with the seismic facies and seismic profiles [24]. Integrat-
ing sedimentary and seismic attributes, in addition to well logging, Zhao [25] established
an initial model based on facies controlled stochastic optimization seismic inversion to
achieve optimal matching between a forward model and observation results. The distribu-
tion characteristics of shale reservoirs, typically the Longmaxi Formation in China, was
predicted by the seismic facies-controlled nonlinear random inversion with the post-stack
seismic data [26].
In addition, phase-controlled inversion with seismic waveform is used to characterize
the distribution characteristics of thin reservoirs in the Yin’e Basin, integrating clustering
recognition and frequency domain curve reconstruction [27]. Zhang et al. [28] applied
seismic phase-controlled nonlinear stochastic inversion to carry out carbonate reservoir
prediction in the Amu Darya Basin by picking up the velocity, amplitude and frequency of
3D seismic waves with geological and logging information. Combining Bayesian classifica-
tion with pre-stack simultaneous inversion, facies-based Bayesian simultaneous inversion
was developed to divide different facies based on multi-elastic parameters, and increase
the stability of the density inversion in the north section of the No.5 fault zone in Shunbei
area, Tarim Basin, China [29]. Gao et al. [30] used the method for reservoir prediction
of subsalt microbial carbonate reservoirs to improve the accuracy of seismic prediction.
Wang et al. [31] studied the sand body boundary and quantitative reservoir prediction using
waveform clustering seismic phase analysis, and described the depositional characteristics
and thickness of the reservoir in the planar and vertical directions.
Processes 2023, 11, x FOR PEER REVIEW 3 of 16
Processes 2023, 11, 2301 prediction. Wang et al. [31] studied the sand body boundary and quantitative reservoir 3 of 15
prediction using waveform clustering seismic phase analysis, and described the deposi-
tional characteristics and thickness of the reservoir in the planar and vertical directions.
In this research, the problems of a low signal-to-noise ratio of seismic records and
In this research, the problems of a low signal-to-noise ratio of seismic records and
predicting thin shale gas reservoirs with a thickness less than 10 m in southeast Chong-
predicting thin shale gas reservoirs with a thickness less than 10 m in southeast Chongqing
qing were addressed. A seismic phase-controlled nonlinear stochastic inversion was ap-
were addressed. A seismic phase-controlled nonlinear stochastic inversion was applied
plied to the Southeast Qianjiang (S-Q) area in Chongqing, China. The horizontal and ver-
to the Southeast Qianjiang (S-Q) area in Chongqing, China. The horizontal and vertical
tical distribution characteristics were proven in the Longmaxi Formation and the Wufeng
distribution characteristics were proven in the Longmaxi Formation and the Wufeng
Formation;
Formation; thin
thin layers
layers with
with aa thickness
thickness of
of less
less than
than 77 m
m were also discovered.
were also discovered.
2.
2. Overview
Overview of of the
the Study
Study Area
Area
2.1. Geological
Geological Structure
Structure and
and Stratigraphic Distribution
The study area
area (S-Q)
(S-Q)isislocated
locatedininthe southeast
the southeast of of
Chongqing
Chongqing in China (Figure
in China 1). The
(Figure 1).
area is a typical karst landscape with large topographic undulations and
The area is a typical karst landscape with large topographic undulations and a steep for- a steep formation
dip angle,
mation dipas shown
angle, in Figure
as shown 2. The topographic
in Figure map inmap
2. The topographic the S-Q area
in the S-Qwas
areaderived using
was derived
the Google
using Earth Engine
the Google (GEE). The
Earth Engine highest
(GEE). elevationelevation
The highest reaches 1900 m, the
reaches 1900lowest
m, theelevation
lowest
is 270 m. isThe
elevation 270structure belongs belongs
m. The structure to the Qiyao
to theMountain
Qiyao Mountainconvex convex
fold belt and
fold Qianjiang
belt and Qi-
depression
anjiang fold belt,fold
depression located
belt, in the southeast
located of the Chongqing–Xiangxi
in the southeast isolated isolated
of the Chongqing–Xiangxi channel
type alluvial
channel fold belt.
type alluvial foldThe anticlinorium
belt. and syncline
The anticlinorium are in are
and syncline parallel with each
in parallel withother,
each
and the direction of the fold axis roughly extends in a northeast to
other, and the direction of the fold axis roughly extends in a northeast to southwest southwest direction;
direc-
the formation
tion; dip dip
the formation angle is about
angle is about35 35
degrees
degreesto to
7575degrees,
degrees,with
withaasteep
steep dip
dip and basic
and basic
symmetry on both flanks. The internal faults in the anticlinorium and
symmetry on both flanks. The internal faults in the anticlinorium and syncline are rela-syncline are relatively
developed,
tively and the
developed, and diptheangle in theincore
dip angle thefaults is steep
core faults (about
is steep 57 degrees
(about to 75to
57 degrees degrees)
75 de-
with the characteristics of a high stratigraphic uplift, strong tectonic compression,
grees) with the characteristics of a high stratigraphic uplift, strong tectonic compression, and high
and steep structural development.
and high and steep structural development.
Figure
Figure 1.
1. Location
Location and
and geological
geological structure
structure of
of S-Q
S-Q well
well area.
area. The
The S-Q
S-Q area
area is
is located
located in
in the
the southeast
southeast
of Chongqing in China, as shown in the red box.
of Chongqing in China, as shown in the red box.
The Yanshan Movement was the main period of tectonic formation in this area, devel-
oping reverse faults along the structure, with faults exhibiting a back-thrust pattern. The
sedimentary environment has deep water bodies, mainly developed as shelf deposits and
shore deposits, which are favorable to the development of dark mud shale of the Upper
Ordovician Wufeng Formation and Lower Silurian Longmaxi Formation [32]. It is the main
reservoir with high shale gas production in China. The exposed stratigraphy is from the
Triassic to Aurignacian; Silurian siltstone and siltstone shale are in a large area; the Lower
Permian limestone is exposed in some areas. From new to old, the stratum exposes the
Processes 2023, 11, 2301 4 of 15
Figure 2.
Figure Topographic map
2. Topographic map in
in S-Q
S-Q area
area was
was derived
derived using
using the
the Google
Google Earth
Earth Engine
Engine (GEE).
(GEE). The
The area
area
is aa typical
is typical karst landscape with large topographic undulations and steep formation dip angle.
The Yanshan
The thicknessMovement
of the Lower wasSilurian Longmaxi
the main periodFormation
of tectonicranges fromin
formation 65–110 m in de-
this area, the
area. The upper part is dark gray, gray-black calcareous, with sandy shale
veloping reverse faults along the structure, with faults exhibiting a back-thrust pa ern. and gray sandy
shale,
The and the lower
sedimentary part is dark
environment hassandy
deep shale,
water with carbonaceous
bodies, shale and
mainly developed carbonaceous
as shelf deposits
mudstone, which form the main reservoir of shale gas. The overlying
and shore deposits, which are favorable to the development of dark mud shale stratum is of
thethe
Lower
Up-
per Ordovician Wufeng Formation and Lower Silurian Longmaxi Formation [32]. It isand
Silurian Xin Tan Formation; the lithology is a set of greenish-gray, yellowish-gray the
blackish-gray
main reservoirmudstones
with high shalewith gas
carbonaceous
productionshales, andThe
in China. theexposed
bottom is a contact between
stratigraphy is from
the Triassic
the gray finetosiltstone
Aurignacian;and calcareous shale ofand
Silurian siltstone the siltstone
Longmaxi Formation.
shale The underlain
are in a large area; the
stratum is the Wufeng Formation, with a set of grayish-black carbonaceous and siliceous
Lower Permian limestone is exposed in some areas. From new to old, the stratum exposes
shales, which are thinly laminated and topped by the Garidonian weathering surface.
the lower Permian limestone, Silurian Luojiaoping formation siltstone and shale,
The thickness of high-quality shale of the Wufeng Group is less than 10 m; it is the main
Longmaxi Formation siltstone and silty shale.
gas-producing layer of shale gas in the area.
The thickness of the Lower Silurian Longmaxi Formation ranges from 65–110 m in
the area. The upper
2.2. Characteristics of part is dark gray,
the Sedimentary gray-black calcareous, with sandy shale and gray
Phase
sandy shale, and the lower part is dark sandy shale, with carbonaceous shale and carbo-
Integrating the data collected from test cores, well drilling and logging in the area,
naceous mudstone, which form the main reservoir of shale gas. The overlying stratum is
the sedimentary phases are carbonate shoals, terraces, occluded basins (lagoons), shallow
the Lower Silurian Xin Tan Formation; the lithology is a set of greenish-gray, yellowish-
marine shelves and foreshore slopes from the bottom to the top of the stratum, as shown
gray and blackish-gray mudstones with carbonaceous shales, and the bo om is a contact
in Figure 3. The phase shows the water body-changed process from the Ordovician to
between the gray fine siltstone and calcareous shale of the Longmaxi Formation. The un-
the Silurian. The water body took place by turning from shallow to deep. The typical
derlain stratum is the Wufeng Formation, with a set of grayish-black carbonaceous and
characteristics of the sedimentary phases of each stratum are briefly described as follows:
siliceous shales, which are thinly laminated and topped by the Garidonian weathering
• Sedimentation of Shallow shoal carbonate terrace (Baota Formation–Linxiang Formation)
surface. The thickness of high-quality shale of the Wufeng Group is less than 10 m; it is
The gas-producing
the main bottom of the Baota layer ofGroup
shale is
gasa in
thin
theinterbedded
area. mudstone and tuff, and the
upward transition is a laminated tuff containing a large amount of biogenic shells or
bioclasts,
2.2. with good
Characteristics rounding,
of the reflecting
Sedimentary Phase a shallow terrace edge environment with strong
hydrodynamic force. The middle and upper parts of the Baota Formation are dominated by
Integrating the data collected from test cores, well drilling and logging in the area,
verrucose tuffs and tortoise tuffs, which are products of shallow water and an arid climate,
the sedimentary phases are carbonate shoals, terraces, occluded basins (lagoons), shallow
with stable low-angle horizontal laminations reflecting low water energy and a depositional
marine shelves and foreshore slopes from the bo om to the top of the stratum, as shown
environment of carbonate terraces.
in Figure 3. The phase shows the water body-changed process from the Ordovician to the
Silurian. The water body took place by turning from shallow to deep. The typical charac-
teristics of the sedimentary phases of each stratum are briefly described as follows:
in the study area. The bo om of the Baota Formation is a thin interlayer of mudstone and
limestone, transitioning upwards into layered limestone, containing a large amount of bi-
ological crusts or debris; it is well-rounded, and reflects the shallow environment of the
platform edge with strong hydrodynamic forces. The upper and middle parts of the Baota
Formation are mainly composed of nodular limestone and turtle crack limestone, which
Processes 2023, 11, 2301 are products of shallow water and an arid climate. The stable low-angle horizontal5 of 15
bed-
ding reflects low water energy, and the sedimentary environment is a carbonate platform.
Figure3.3.Analysis
Figure Analysisofofsedimentary
sedimentaryphase
phaseand
andstratigraphic
stratigraphicsequence
sequencewith
withS-Q-2
S-Q-2well
welllogging.
logging.MD,
MD,
measure depth; FMI, formation micro-scanner image; GR, natural gamma-ray logging;
measure depth; FMI, formation micro-scanner image; GR, natural gamma-ray logging; DLL, dual DLL, dual
laterallogging;
lateral logging;DT,
DT,acoustic
acousticdifference
differencetime
timelogging.
logging.
TheSedimentation
Ordovician of closed sea
is mainly basins and
composed lagoonslimestone
of nodular (Wufeng andFormation–Longmaxi
turtle crack limestoneFor-
mation)
in the study area. The bottom of the Baota Formation is a thin interlayer of mudstone and
limestone, transitioning
The upper Ordovicianupwards
Wufeng into layered limestone,
Formation and lowercontaining a large amount
Silurian Longmaxi Formationof
biological crusts or debris; it is well-rounded, and reflects the shallow environment
have well-developed foliation, containing slightly gray ma er, and is relatively dense. The of the
platform
top of theedge with Formation
Wufeng strong hydrodynamic forces.
and Longmaxi The upper
Formation and middle
generally partscarbon,
contains of the Baota
which
Formation are mainly composed of nodular limestone and turtle crack
is mainly composed of carbon shale. Most layers of the Longmaxi Formation contain limestone, which
less
are products
sand, and theoflithology
shallow water and anshale.
is siliceous arid climate.
Common The stable low-angle
graptolite horizontal
fossils and bedding
pyrite patches or
reflects low water energy, and the sedimentary environment is a carbonate platform.
• Sedimentation of closed sea basins and lagoons (Wufeng Formation–Longmaxi Formation)
The upper Ordovician Wufeng Formation and lower Silurian Longmaxi Formation
have well-developed foliation, containing slightly gray matter, and is relatively dense. The
top of the Wufeng Formation and Longmaxi Formation generally contains carbon, which
is mainly composed of carbon shale. Most layers of the Longmaxi Formation contain less
sand, and the lithology is siliceous shale. Common graptolite fossils and pyrite patches or
strips with horizontal bedding development reflect the retention reduction environment,
with weak hydrodynamic strength and slow sedimentation speed, forming a lagoon or
closed basin environment.
• Sedimentation of shallow shelf-frontal slope (Xintan Formation)
The bottom of the Xintan Formation is mainly composed of calcareous shale deposits;
horizontal bedding and cross bedding at a certain angle are developed. No reduction miner-
als such as pyrite are found, reflecting shallow water carbonate a sedimentary environment
of shallow water, medium hydrodynamic force and normal redox environment-shallow
sea shelf deposits. The upper and middle parts of the Xintan Formation are thick shale
deposits, with frequent interaction between sliding deformation structures caused by grav-
Processes 2023, 11, 2301 6 of 15
ity flow and the horizontal bedding of still water deposits, reflecting the carbonate slope
environment, i.e., the slope zone between shallow and deep sea.
Figure 4.
Figure 4. Petrophysical
Petrophysical characteristics
characteristicsand
anddistribution
distributionof
ofshale
shalein inS-Q
S-Qwell
wellarea.
area.(a):
(a): P-wave
P-wave velocity
velocity
(m/s); (b): S-wave 3);3 (d): natural gamma (API).
(m/s); S-wave velocity
velocity (m/s);
(m/s);(c):
(c):rock
rockdensity
density(g/cm
(g/cm ); (d): natural gamma (API).
It includes carbonaceous shale and sandy shale, which can be divided into three
3. Methodology
sections according
A seismic to their
phase is thegas-bearing
reflection ofproperties. Among
the sedimentary them,
phase in the
the vertical
seismic depths of
data, and
727 m to 778 m (thickness of 51 m) and 778 m to 792 m (thickness of 14 m)
any kind of seismic phase has specific seismic reflection characteristics that correspond to are low
gas content, while
the appropriate the depthphase.
sedimentary of 792Based
m toon802
themgeometry,
(thicknessinterrelationship
of 10 m) are enriched gas-
and internal
bearing layers. Various parameters are influenced by the condition of the gas
structure of the seismic phase, the position in the regional structure and the phase inter- content.
preted by the well data, the corresponding sedimentary phase can be initially determined.
The seismic phase-controlled nonlinear stochastic inversion performs a nonlinear optimi-
zation of the seismic records. Combining the individual problem into a joint inversion, the
degrees of freedom of the inversion can be reduced under the control of the seismic phase
model. The wave impedance inversion is transformed into an inversion of the formation
velocity, and the minimum value of the least square solution between the seismic model
Processes 2023, 11, 2301 7 of 15
Gas-bearing reservoirs have the characteristics of high gamma, low density, and low
acoustic wave velocity.
The P-wave velocity of the gas-bearing reservoir is from 3700 m/s to 4100 m/s, and
the S-wave velocity is 2300 m/s to 2500 m/s. The rock density is between 1.94–2.03 g/cm3 .
On the contrary, the velocity of P-waves and S-waves on non-gas bearing reservoirs are
both high, ranging from 4300–4900 m/s and 2500–2800 m/s, respectively. The rock density
is between 2.04–2.15 g/cm3 , and the natural gamma is greater than 160 API.
In summary, the gas saturation, petrophysical characteristics and marine sedimentary
environment in the study area show that the shale reservoir has good economic value.
However, the reservoir in the area is developed in thin interbedded layers, and the reser-
voir thickness changes greatly. As a result, exploration of the reservoir’s distribution
characteristics requires an accurate seismic inversion method.
3. Methodology
A seismic phase is the reflection of the sedimentary phase in the seismic data, and any
kind of seismic phase has specific seismic reflection characteristics that correspond to the
appropriate sedimentary phase. Based on the geometry, interrelationship and internal struc-
ture of the seismic phase, the position in the regional structure and the phase interpreted
by the well data, the corresponding sedimentary phase can be initially determined. The
seismic phase-controlled nonlinear stochastic inversion performs a nonlinear optimization
of the seismic records. Combining the individual problem into a joint inversion, the degrees
of freedom of the inversion can be reduced under the control of the seismic phase model.
The wave impedance inversion is transformed into an inversion of the formation velocity,
and the minimum value of the least square solution between the seismic model and actual
seismic records is obtained. The special process can be expressed as follows:
n −1 2
f (V ) = ∑ Si∆ − Di → min (1)
i =0
where V is the acoustic velocity in the formation, Si∆ is the synthesized seismic record using
the velocity model of the formation medium, Di is the actual seismic record and i is the
sampling point number of the seismic record.
According to the theory of the generalized linear inversion [33], Equation (1) can
be expanded at the initial model response Si with Taylor’s formula. To improve the
inversion accuracy of the thin-layer seismic record, only the quadratic terms are retained;
the Formula (1) is simplified as follows:
!
n −1 n −1
∆ ∂Si 1 ∂ 2 Si
Si − Di = Si − Di + ∑ ∆Vk + ∆Vk ∑ 2
(2)
k =0
∂Vk 2 k =0
∂Vk ∂Vj
where Si is the synthetic seismic record corresponding to the initial velocity model, and
∆V is the parameter perturbation in the model. Making the first-order derivative of the
perturbation term ∆V to zero, the Formula (2) may be written as follows:
2
n −1 ∂(Si∆ − Di ) n −1 ∂S∆
∑ = 2 ∑ (Si∆ − Di ) i = 0 (3)
i =0
∂∆Vj i =0
∂∆Vj
n −1 n −1
A= ∑ ∂Si
∂Vj ∑
∂Si
∂Vk
i =0 k =0
n −1 n −1
∂ 2 Si
B= ∑ ( S i − Di ) ∑ ∂Vj ∂Vk (6)
i =0 k =0
n −1 n o
C= ∑ ∂Si
(Si − Di ) ∂V
j
i =0
The actual acoustic velocity V in the formation is obtained by iterating Equation (6)
with the model parameter perturbation ∆V by Equation (7).
Figure5.5.The
Figure Theworkflow
workflowfor
forseismic
seismicphase-controlled
phase-controllednonlinear
nonlinearstochastic
stochasticinversion
inversion(SPCNSI).
(SPCNSI).
Basedon
Based onthe
thesolved
solvedacoustic
acousticvelocity
velocitywith
withiteration
iteration processing,
processing,the
the sequence
sequence bound-
bound-
ariesof
aries ofthe
thestrata
stratawere
wereinterpreted
interpretedon onthe
theseismic
seismicprofile
profilecompared
comparedwith withthe
thesequences
sequences
dividedby
divided bydrilling,
drilling,logging
loggingand
andother
otherdata.
data.The
Theconstraints
constraintsfor
forseismic
seismicphase-controlled
phase-controlled
inversioncan
inversion canbebeformed
formedwith
withthe
thematching
matching relationship
relationship between
between target
target layers
layers andand differ-
different
sequences in three-dimensional
ent sequences space.
in three-dimensional space.
In this research, the seismic acquisition instrument adopted Sercel’s 428 digital seis-
mometer, with a seismic channel gain of 18 dB and a sampling rate of 1 ms; the recording
length was 5 s. The main acquisition parameters are shown in Table 1.
In this research, the seismic acquisition instrument adopted Sercel’s 428 digital seis-
mometer, with a seismic channel gain of 18 dB and a sampling rate of 1 ms; the recording
length was 5 s. The main acquisition parameters are shown in Table 1.
4. Results
4.1. 3D Seismic Exploration for Shale Gas Reservoirs
The seismic records were severely affected by the poor conditions for seismic excitation
and reception, due to the large number of exposed limestones and elevation differences in
the surface terrain of the area. The implementation of techniques such as selecting physical
points, optimizing observation systems and optimizing excitation and reception parameters
ensured the quality of the 3D seismic acquisition data.
From the survey line profile, it can be seen that the underground structure of the
study area is relatively flat, with obvious reflection characteristics of the Silurian Long-
maxi Formation and the Upper Ordovician Wufeng Formation, and obvious wave group
relationships. The internal reflection structure of both formations is parallel, with good
continuity of the same phase records, indicating that the formations were in a stable and
low-energy environment during their sedimentary period, as shown in Figure 6. After
the well-to-seismic calibration, it is believed that the continuously weak parallel seismic
reflection corresponds to the sedimentation of a lagoon (closed basin). In comparing the
synthetic records and seismic records next to the well, the high-quality shale is located
between the trough and peak (the top boundary is trough reflection, and the bottom is
strong reflection). The main characteristics of each layer are as follows:
(1) Lower Silurian Xiaoheba Formation (S1xh): thickness approximately 200 m; light
yellow-grey and chartreuse silty hydromica shale and hydromica shale interbedded
with different thicknesses, and local lenticular limestone and bands are occasionally
found; the bottom is light gray medium-thick to thick-layered siltstone, and part of it
is calcareous siltstone or calcareous fine-grained sandstone.
(2) Lower Silurian Xintan Formation section 2 (S1x2): approximately 250 m thick; the
grayish green and chartreuse silty hydromica shale is mainly mixed with hydromica
shale, and the top and bottom are light gray medium-thick to thick bedded argilla-
ceous siltstone.
(3) The first section of the Lower Silurian Xintan Formation (S1x1): approximately 180 m
thick; gray, green-gray hydromica shale, silty shale, mixed with gray medium-thick to
thick bedded argillaceous siltstone. The local shale contains a small amount of carbon
and integrates and contacts with the underlying strata.
(4) Lower Silurian Longmaxi Formation (S1l): approximately 50 m thick; black carbona-
ceous silty hydromica shale, carbonaceous hydromica shale, with light yellow-gray
thin to medium-thick bedded quartz siltstone in the middle and upper part. Contains
a large number of graptolite fossils, and integrates and contacts with the underly-
ing strata.
(5) Upper Ordovician Wufeng Formation (O3w): approximately 10 m thick; black thin-
layer carbonaceous siliceous shale, occasionally mixed with light gray thin-layered
dolomitic limestone; integrates and contacts with the underlying strata.
(6) Upper Ordovician Linxiang Formation (O3l): thickness of about 10 m; black medium-
thick layered nodular argillaceous limestone; integrates and contacts with the under-
lying strata.
(5) Upper Ordovician Wufeng Formation (O3w): approximately 10 m thick; black thin-
layer carbonaceous siliceous shale, occasionally mixed with light gray thin-layered dol-
omitic limestone; integrates and contacts with the underlying strata.
(6) Upper Ordovician Linxiang Formation (O3l): thickness of about 10 m; black medium-
Processes 2023, 11, 2301 thick layered nodular argillaceous limestone; integrates and contacts with the 10 of 15
underly-
ing strata.
(7) Middle Ordovician Baota Formation (O2b): gray to dark gray thin to medium-thick
(7) layered
Middledry
Ordovician Baota Formation
cracked limestone, (O2b): mixed
occasionally gray towith
darkmesoscale
gray thin to medium-thick
limestone.
layered dry cracked limestone, occasionally mixed with mesoscale limestone.
Figure 6. Horizon calibration and sequence boundary tracking of shale gas reservoirs in S-Q area.
CDP represents the different common depth points in the same survey line for 3D seismic exploration.
The red lines represent the seismic records synthesized from logging data.
area (Figure 8); this reflects the spatial distribution characteristics of reservoirs in both the
horizontal and vertical directions. The distribution range of shale reservoirs is wide, with
of
theless than 9 m
obvious lowonly appearvelocities
P-wave locally. The reservoir of the3500–4700
(approximately Wufeng Formation is characterized
m/s). Combined with geo-
by a stable distribution and wide range, with a maximum thickness of 12 m
logical parameters such as the gas content, reservoir thickness and rock porosity in this and an average
thickness of about oil
area, the original 7 m.inReservoirs
place of the with a thickness
shale greater
gas in this area than 10 m210
reached are×concentrated
108 m3. in
the north of the study area, while thicknesses less than 4 m are mainly distributed near the
The shale gas reservoir of the Longmaxi Formation is distributed stably in the whole
faults. The inversion results are consistent with the logging interpretation of the shale gas
area, with less thickness variation, as shown in Figure 9. The maximum thickness is about
reservoir on wells Q1 and Q2.
18 m, and the minimum thickness is about 7 m. Reservoirs with a thickness greater than
In addition, the results of drilling, logging and gas content testing from the S-Q-6
15 mshow
well are mainly developed
that there are threeinshale
the S-Q-2 well areawith
gas reservoirs and in the north
a total of the
thickness ofarea.
77.8 mThicknesses
in the
of less than
Longmaxi 9 m onlyand
Formation appear locally.
Wufeng The reservoir
Formation of theofWufeng
in the north the study Formation
area. Theisvertical
character-
ized byare
depths a stable
730.4 mdistribution
to 784 m, 786andm towide
795 mrange, withma to
and 801.3 maximum
816.5 m. Thethickness of reservoir
shale gas 12 m and an
average
with the thickness of about
best gas content is 7 m.Wufeng
the Reservoirs with a thickness
Formation (801.3–816.5greater thana10
m), with m are concen-
thickness of
trated
15.2 m,inanthe north of
average theorganic
total study area,
carbonwhile
(TOC)thicknesses
content of less than
3.8%, a 4porosity
m are mainly
of aboutdistributed
4.1%
nearathe
and gasfaults.
content The 3.1 m3 /t. After
of inversion results are
the consistent
reservoir with the logging
stimulation, the dailyinterpretation
gas production of the
is × 10
1.38gas 4 3
m per on day. TheQ1 accuracy and reliability of this method w also verified in
shale reservoir wells and Q2.
predictions of the marine thin shale gas reservoir.
Figure
Figure7.7. Well
Well logging constrained inversion of formation velocity profile.
profile. The
The “Line”
“Line” represents
represents the
the
different survey
different survey line
line for
for 3D
3D seismic
seismic exploration.
exploration. Different
Different color blocks represent different
different P–wave
P–wave
velocitiesand
velocities andthe
the red
red line
line is
is the
the GR
GR logging
logging data.
data. (a),
(a), inversion
inversion results
results from
from the
theseismic
seismic“CDP”
“CDP”data;
data;
(b) inversion results from the seismic “Line”
(b) inversion results from the seismic “Line” data. data.
Figure 7. Well logging constrained inversion of formation velocity profile. The “Line” represents the
different survey line for 3D seismic exploration. Different color blocks represent different P–wave
Processes 2023, 11, 2301 12 of 15
velocities and the red line is the GR logging data. (a), inversion results from the seismic “CDP” data;
(b) inversion results from the seismic “Line” data.
Figure9.9.Thickness
Figure Thicknessprediction
predictionofofhigh-quality
high-qualityshale gas
shale reservoir
gas in in
reservoir thethe
Longmaxi Formation
Longmaxi andand
Formation
Wufeng Formation.
Wufeng Formation.
In addition, the results of drilling, logging and gas content testing from the S-Q-6
well show that there are three shale gas reservoirs with a total thickness of 77.8 m in the
Longmaxi Formation and Wufeng Formation in the north of the study area. The vertical
depths are 730.4 m to 784 m, 786 m to 795 m and 801.3 m to 816.5 m. The shale gas reservoir
with the best gas content is the Wufeng Formation (801.3–816.5 m), with a thickness of
15.2 m, an average total organic carbon (TOC) content of 3.8%, a porosity of about 4.1%
Processes 2023, 11, 2301 13 of 15
5. Discussion
As for predicting the thin shale gas reservoir, the SPCNSI method solves for the
minimum value between the seismic model and seismic record, with the constraints of
stratigraphic sequence or seismic facies through iterative processes. It could effectively
improve the vertical and longitudinal resolution. The method is suitable for the prediction
of thin shale gas reservoirs in complex geological structures. However, the method needs
a lot of data (i.e., logging, petrophysical and geology), as the inversion constraints condition
differs from the acoustic impedance inversion and geostatistics-based stochastic inversion.
It is limited for the new exploration area due to a lack of the necessary data. On the
other hand, we only implemented SPCNSI programming with Microsoft Visual Studio
C++ (version 2019, Microsoft Corporation, Redmond, WA, USA) [35], and the inversion
results were imported into Landmark [36] software (version 2016, Halliburton Company,
Odessa, TX, USA) for further reservoir analysis and interpretation. Unfortunately, we did
not integrate these methods into a complete software package until now. However, we will
address this problem in the next study.
In recent years, machine learning has become an effective method for understanding
geophysics and geomorphic properties [37]. Due to its excellent ability to solve nonlinear
problems, it has become a popular topic in seismic exploration. It has been widely devel-
oped to predict seismic attributes and formation structures [38–41]. Objectively, reservoir
heterogeneity leads to multiple solutions and uncertainty in solving petroleum geological
problems, and it is difficult to obtain “textbook” tag data for machine learning. Due to the
strong specialty and particularity of oil E&D data, the general artificial intelligence (AI)
algorithm cannot be used directly. However, AI technology will certainly provide new
momentum for scientific breakthroughs in the whole industrial chain of oil and gas [42].
The multi-type unstructured data (i.e., logging, geology, seismology and petrophysics)
could be applied to machine learning models for reservoir predictions of thin interbed
shale gas.
6. Conclusions
Seismic phase-controlled nonlinear stochastic inversion (SPCNSI) can effectively im-
prove the resolution of oil and gas reservoir predictions. In this research, we solved the
difficulty of predicting thin layer shale gas reservoirs under marine sedimentation condi-
tions. Three sets of reservoirs with a thickness of less than 20 m were identified, and can be
traced throughout the study area. Two favorable areas for shale gas have been determined
in the S-Q area, in terms of the inversion results. The horizontal and vertical distribution
characteristics of the reservoir have been proven. The following understandings have
been obtained:
(1) The characteristics of low formation acoustic velocity in high-quality shale gas reser-
voirs is obviously in southeast Chongqing, in China.
(2) The inversion results of SPCNSI can effectively improve the vertical resolution of
seismic profiles, and can also identify thin reservoirs below 7 m in thickness.
(3) The high-quality shale gas reservoirs (Longmaxi Formation and Wufeng Formation)
are stably distributed in the study area, with little change in thickness. The average
thickness of the Longmaxi Formation is about 12 m, and the reservoirs with a thickness
greater than 10 m in the Wufeng Formation are concentrated in the north of the study
area, with an average thickness of about 7 m.
(4) Combined with drilling, logging and geology, the sequence and sedimentary phases from
seismic profiles can be interactively calibrated, and the results of SPCNSI can accurately
describe the sedimentary characteristics of thin interbeds of shale gas reservoirs.
Processes 2023, 11, 2301 14 of 15
Author Contributions: Conceptualization, Q.X.; methodology, Q.X. and X.H.; investigation, Q.X.
and Y.W.; resources, X.G. and D.J.; data curation, X.H. and Q.H.; writing—original draft preparation,
Q.X. and Q.H.; writing—review and editing, Q.X., Q.H. and X.H.; supervision, Q.X., Y.W. and B.W.;
project administration, Q.X., Y.W. and Y.H. All authors have read and agreed to the published version
of the manuscript.
Funding: We are grateful to be supported by the Key Project of Science and Technology Research
program of Chongqing Education Commission of China (No. KJZD-K202203103), and the Chongqing
Administration of Science and Technology (No. CSTB2022NSCQ-MSX1433).
Data Availability Statement: Unable to obtain data due to privacy restrictions.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Li, Q.Z. On strategy of seismic restricted inversion. Oil Geophys. Prospect. 1998, 33, 423–438.
2. Mou, Y.G.; Pei, Z.L. Seismic Numerical Simulation of 3D Complex Media; Petroleum Industry Press: Beijing, China, 2015.
3. Pei, S.; Yin, X.Y.; Zong, Z.Y.; Li, K. High resolution fixed-point seismic inversion. Interpretation 2021, 9, 1–54. [CrossRef]
4. Cheng, L.; Wang, S.X.; Wei, W.W.; Gao, H.Y.; Yu, Z.H.; Zhu, C.J.; Xiu, J.L. Structural geosteering constrained multi-trace sparse
reflectivity inversion based on mixed norms. J. Appl. Geophys. 2020, 172, 103880. [CrossRef]
5. Zheng, Y.; Liu, Y.S.; Xu, T.; Li, Z.Y. Elastic correlative least-squares reverse time migration based on wave mode decomposition.
Processes 2022, 10, 288. [CrossRef]
6. Dai, S.H.; Chen, Z.G.; Yu, J.B.; Xie, J.L.; Liu, Y.Q. Application of the fluid mobility attribute technique in the reservoir characteriza-
tion of KG Oilfield, Algeria. Pet. Explor. Dev. 2010, 37, 573–578.
7. Parra, J.O.; Hackert, C. Wave attenuation attributes as flow unit indicators. Lead. Edge 2002, 21, 564–572.
8. Klimentos, T. Attenuation of P- and S-waves as a method of distinguishing gas and condensate from oil and water. Geophysics
1995, 60, 447–458. [CrossRef]
9. Sui, S.L.; Tang, J.; Jiang, Y.B.; Jing, L.C. Technical characteristics and applicable conditions analysis on seismic inversion traditional
methods. Pet. Geol. Recover. Effic. 2012, 19, 38–41.
10. Trapp, M.; Nestorovic, T. Non-gradient full waveform inversion approaches for exploration during mechanized tunneling applied
to surrogate laboratory measurements. Tunn. Undergr. Space Technol. 2022, 2, 120.
11. Sun, S.M.; Peng, S.M. Inersion of geostatistics based on simulated annealing algorithm. Oil Geophys. Prospect. 2007, 42, 38–43.
12. Dong, W.B.; Li, Y.G.; Gui, Z.X.; Zhou, L. Theory and application of geostatistical inversion: A facies-constrained MCMC algorithm.
Processes 2023, 11, 1335. [CrossRef]
13. Mohammed, M.A.; Snehamoy, C.; Roohollah, A. Stochastic inversion combining seismic data, facies properties, and advanced
multiple-point geostatistics. J. Appl. Geophys. 2023, 213, 105026.
14. Zhang, B. Application of probabilistic simulation technique for reservoir prediction in Taiyangdao-Kuihuadao area. Spec. Oil Gas
Reserv. 2008, 15, 46–49.
15. Li, Y.Q.; Huo, Z.Z.; Li, J.Y.; Chen, X.H.; Zhang, J.; Geng, W.H. Generalized impedance blocky inversion based on analytic solution
to wave equation. Oil Geophys. Prospect. 2020, 55, 1073–1083.
16. Li, H.X. A Seismic reservoir prediction based on OVT domain. Spec. Oil Gas Reserv. 2019, 26, 52–57.
17. Zhang, J.; Li, J.G.; Chen, X.H.; Li, Y.Q. Geological structure-guided hybrid MCMC and Bayesian linearized inversion methodology.
J. Pet. Sci. Eng. 2021, 199, 108296. [CrossRef]
18. Yang, W.C. Nonlinear chaotic inversion of seismic traces: I. Theory and numerical experiments. ACTA Geophys. Sin. 1993, 36,
222–232.
19. Yang, W.C. Nonlinear chaotic inversion of seismic traces: II. Lyapunov exponents and attractors. ACTA Geophys. Sin. 1993, 36,
376–387.
20. Jiang, L.; Huang, H.D.; Wei, X.C.; He, Z.H. Non-linear constrained inversion of seismic data. Oil Geophys. Prospect. 2003, 38,
435–438.
21. Huang, H.D.; Luo, Q.; Fu, Y.; Wang, B.H. Study and application of seismic phase-controlled non-linear random inversion. Oil
Geophys. Prospect. 2007, 42, 694–698.
22. Sambridge, M. Geophysical inversion with a neighbourhood algorithm—I. Searching a parameter space. Geophys. J. Int. 1999, 138,
479–494. [CrossRef]
23. Sambridge, M. Geophysical inversion with a neighbourhood algorithm--II. Appraisng the ensemble. Geophys. J. Int. 1999, 138,
727–746. [CrossRef]
24. Huang, H.D.; Cao, X.H.; Luo, Q. An application of seismic sedimentology in predicting organic reefs and banks: A case study on
the Jiannan-Longjuba region of the eastern Sichuan fold belt. ACTA Pertrolei Sin. 2011, 32, 629–636.
25. Zhao, W.P.; Fan, T.E.; Yang, L. Facies constrained stochastic optimization seismic inversion and application. Adv. Geosci. 2018, 8,
484–491.
Processes 2023, 11, 2301 15 of 15
26. Fu, C. Application of seismic nonlinear random inversion to shale gas reservoir prediction in northwestern Hunan. Chin. J. Eng.
Geophys. 2020, 17, 393–398.
27. Guo, X.; Li, F.L.; Zhang, X.F.; Yin, X.B.; Wang, B. Application of seismic waveform facies-controlled inversion in the cretaceous
thin reservoir prediction in Tiancao Sag. Spec. Oil Gas Reserv. 2019, 26, 38–43.
28. Zhang, Z.W.; Wang, C.S.; Lin, Y.P.; Zhou, T.W.; Huang, H.D. Applications of facies-controlled nonlinear random inversion for
carbonate reservoir prediction in Block A of Amu-darya Basin. Oil Geophys. Prospect. 2011, 46, 304–310.
29. Cao, F.; Lu, Z.Q.; Liu, W.F. Facies-based Bayesian simultaneous inversion technology and its application: A case study of the
north section of No.5 fault zone in Shunbei area, Tarim Basin, China. J. Nat. Gas Geosci. 2022, 7, 347–357.
30. Gao, J.; Huang, H.D.; Ji, M.; Yuan, S.Q. Seismic phase-controlled nonlinear inversion of a carbonate reservoir. Geophys. Prospect.
Pet. 2020, 59, 396–403.
31. Wang, D.K.; Lu, Y.Y.; Li, L.H.; Lin, L.M.; Shi, H.; Du, J.; Xie, Y.G. Prediction of tight sandstone reservoirs in marine coal strata
based on seismic facies controlled inversion. Sci. Technol. Eng. 2020, 20, 12349–12354.
32. He, X.P.; He, G.S.; Gao, Y.Q.; Zhang, P.X.; Lu, S.F.; Wan, J.Y. Geological characteristics and enrichment laws of normal-pressure
shale gas in the basin-margin transition zone of SE Chongqing. Nat. Gas Ind. 2018, 38, 333–346.
33. Cooke, D.A.; Schneider, W.A. Generalized linear inversion of reflection seismic data. Geophysics 1983, 48, 665. [CrossRef]
34. Li, S. Study on Seismic Inversion Method of Complex Reservoir-Taking Jiyang Depression as an Example. Ph. D. Thesis, China
University of Petroleum, Beijing, China, 2020.
35. Microsoft Corporation. Microsoft Visual Studio C++. Available online: https://learn.microsoft.com/zh-cn/cpp/overview/
visual-cpp-in-visual-studio (accessed on 15 March 2020).
36. American Halliburton Company. Landmark. Available online: https://www.halliburton.com/en/software (accessed on 18 July
2021).
37. Jamali, A.A.; Zarekia, S.; Randhir, T.O. Risk assessment of sand dune disaster in relation to geomorphic properties and vulnerabil-
ity in the Saduq-Yazd Erg. Appl. Ecol. Environ. Res. 2018, 16, 579–590. [CrossRef]
38. Gao, Z.Q.; Yang, W.; Tian, Y.J.; Li, C.; Jiang, X.D.; Gao, J.H.; Xu, Z.B. Global optimization with deep-learning-based acceleration
surrogate for large-scale seismic acoustic-impedance inversion. Geophysics 2022, 87, R35–R51. [CrossRef]
39. Yan, X.S.; Zhang, M.Z.; Wu, Q.H. Big-data-driven pre-stack seismic intelligent inversion. Inf. Sci. 2021, 549, 34–52. [CrossRef]
40. Wang, Z.T.; Shen, G.Q.; Wang, Y.M.; Meng, X.J. Model driven high-precision prestack seismic inversion. Geophys. Prospect. Pet.
2020, 59, 927–935.
41. Ajah, N.J.; Dosunmu, A.; Akaolisa, C.Z.; Dagogo, T. Analysis of elastic geomechanical properties derived from well Log and
seismic data, using artificial intelligence (ANN): A case study of “AJAH” field offshore Niger Delta. J. Appl. Geol. Geophys. 2020,
8, 19–27.
42. Kuang, L.C.; Liu, H.; Ren, Y.L.; Luo, K.; Shi, M.Y.; Su, J.; Li, X. Application and development trend of artificial intelligence in
petroleum exploration and development. Pet. Explor. Dev. 2021, 48, 1–14. [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.