PETROLEUM EXPLORATION AND DEVELOPMENT

Volume 36, Issue 4, August 2009

Online English edition of the Chinese language journal

Cite this article as: PETROL. EXPLOR. DEVELOP., 2009, 36(4): 448–455. RESEARCH PAPER

Application of high-resolution sequence stratigraphy in

improving the precision of reservoir prediction
Li Jinsong1,2,*, Zheng Xiaodong2, Gao Zhiyong2, Li Yonggen2, Liu Ying2, Li Xingli2
1. China University of Geosciences, Beijing 100083, China;
2. PetroChina Research Institute of Petroleum Exploration & Development, Beijing 100083, China

Abstract: On the basis of high-resolution sequence stratigraphy, this paper establishes a sequence stratigraphic frame, and predicts
reservoirs by analyzing velocity field, seismic inversion, and seismic attributes. In the case study of the Jurassic Toutunhe Formation
braided river delta reservoir of Shinan 21 well block, middle of the Junggar Basin, space variant velocity field inversion and reservoir
prediction were carried out based on the sequence stratigraphic frame and the reservoir prediction accuracy was improved. Comparing
the prediction results with the real drilling shows that, when the burial depth of bed of interest is 2 500 m, the maximum prediction error
of depth is only 10 m, and the average relative prediction error is 0.16%; for appraisal wells with an average effective sandstone of 13 m,
the maximum prediction error of sandstone thickness is only 5 m. The structure mapping and reservoir prediction constrained by the
sequence stratigraphic frame achieved success.

Key words: sequence stratigraphic frame; reservoir prediction; Junggar Basin; braided river delta

As the exploration targets for hydrocarbons are changing
from structural pools to various types of lithological ones[1-4],
the foundation geology has gone through noticeable changes.
As represented by high-resolution sequence stratigraphy,
sedimentology has progressively become the primary
theoretic basis of exploration for lithologic pools; on the
basis of the classification and set-up of sequence
stratigraphic frame for the study zone, the study on velocity
field, seismic inversion, and analysis of seismic attributes
would be more precise, consequently improving the
reservoir prediction precision in a effective manner. Based
on the case study of Jurassic Toutunhe Formation braided
river delta reservoir of Shinan 21 well field, middle of
Junggar Basin, this paper illustrates the role of
high-resolution sequence stratigraphy in improving the
precision of reservoir prediction, hopefully to provide new
thought in the exploration for lithologic pools in the
future[5].
1 Geological overview of study area
Perched on the center of Guerban Tonggute Desert in the
middle part of Junggar Basin, the Shinan area was
dominated by the braided river delta sedimentation in the
mid-Jurassic era. Under the control of the NE fault slope
break belt, the sand in the branch channels of the delta front
extended in strips parallel to or diagonally intersected with
the structural lines, which was favorable for the forming of
lithologic traps[6]. Featuring a low-energy braided river
lithosomic body, the target formation, i.e., Jurassic
Toutunhe Formation (J2t), could be classified (from top to
bottom) into 3 sand members: J2t1, J2t2 and J2t3. Concretely,
J2t1 and J2t3 are principally of mudstone, while J2t2 features
the primary oil pay zone, consisting of dark-gray medium-
and fine-grain sandstone, inequigranular sandstone, and a
small quantity of arenaceous conglomerate. Overall, J2t2
member, according to the spontaneous potential log,
presents the marked feature of “wrap-up of sandstone in
mudstone”, where extensive sand develops laterally,
indicating a satisfactory condition for the forming of
lithologic pools. As demonstrated by drilling, J2t2 reservoir
in Shinan 21 well field is principally controlled by lithologic
boundary, except for its southern part which is under the
control of oil-water boundary.
2 Single well classification and comparison of
high-resolution sequence stratigraphy
2.1 Single well classification of short base-level cycle
In the study area, sequence classification is conducted to
each well according to the periodic retrograding-prograding
variation as reflected by its log, as well as the sequence
interface identification marks, including mudstone color

Received date: 15 December 2008; Revised date: 25 April 2009.

* Corresponding author. E-mail: ljs@petrochina.com.cn
Copyright © 2009, Research Institute of Petroleum Exploration and Development, PetroChina. Published by Elsevier BV. All rights reserved.
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455
Fig. 1 Cross-well seismic interpretation profile of high-resolution sequence stratigraphy for Well Shi 108 - Well Shi 116
(reflecting the sedimentary environment), channel scour
surface, abrupt change and contact of the log (corresponding
to the left-over overburden sediments)[7-10]; it is believed that,
due to the high value of gamma ray logging inside Toutunhe
Formation, the top of mudstone layer represents the flooding
surface of the medium base-level cycle (the fourth order
sequence), below which is the prograding superimposition
pattern and above which is the retrograding pattern. Further,
3 - 4 short base-level cycle sequences (the fifth order
sequence) may be divided, namely, SSQ1 through SSQ4
from bottom to top; concretely, as for Well Shinan 3 located
in the east of the well field, 4 such sequences (i.e., SSQ1 -
SSQ4) could be observed, and, in other wells, only 3 ones
(i.e., SSQ1 - SSQ3) are available.
For the short base-level cycle sequence SSQ1, the lower
interface (i.e., SB1) features an unconformity surface
between Toutunhe Formation and Xishanyao Formation,
which, geographically corresponding to the Yanshan
episode 1 tectonic movement, represents a shifting surface
between fluvial and delta facies. Below such an interface,
there is one set of variegated reddish mudstone, and, above
it, there is one combination of sandstone and conglomerate;
its overall logging response indicates a transitional surface
between rhythm and counter-rhythm, while the seismic
response is mainly represented by conformity or the
corresponding parallel unconformities.
In the same sequence SSQ1, below the upper interface
(i.e., SB2) are pelitic siltstone, mixed with dark-gray,
grayish-black mudstone with non-uniform thickness,
calcareous mudstone, and black coal seam; while above it is
grayish fine-grain sandstone, featuring a lithologic
sudden-change surface. Whereas the seismic response
presents successive wave trough reflection, the lateral
seismic follow-up is difficult.
In the short base-level cycle sequence SSQ2, below the
upper interface (i.e., SB3) is predominantly grayish
fine-grain sandstone, while above it are mainly the brownish
pelitic siltstone and grayish siltstone, and the seismic
response is principally represented by parallel reflection,
irrespective of being above or below the interface.
In the short base-level cycle sequence SSQ3, the upper
interface SB4 is a non-conformity surface between
Qingshuihe Formation and Toutunhe Formation, which,
geographically corresponding to Yanshan episode 2 tectonic
movement, features a long-term erosion surface.
Irrespective of being above or below the interface, the
seismic response is represented by angular-unconformity
contact relation. Fig. 1 provides the joint seismic
interpretation profile of high-resolution sequence
stratigraphy for Well Shi 108 – Well Shinan 21 – Well Shi
106 – Well Shi 113 – Well Shi 121 – Well Shi 116.
2.2 Set-up of isochronal stratigraphic frame for short
base-level cycle
Based on the single well classification of short base-level
cycle for the study area, isochronal comparison is carried
out, one after another (the uppermost one first), to the short
cycle sequences falling within the semi rising cycle of the
medium base-level, starting from the mid-term flooding
surface developed within Toutunhe Formation, and ending
at the bottom of SB1; likewise, comparison in the opposite
direction (i.e., the lowest one first) is conducted to the short
cycle sequences falling within the semi falling cycle,
starting from the flooding surface and ending at the ceiling
of SB4 (or SB5 in the case of Well Shinan 3).
On the well-connecting line along the WE structural
strike, the spreading of braided channels inside the study
area was quite limited in the sedimentation stage of the short
base-level cycle SSQ1; Ji 004, Shi 108, Shi 112 and Shinan
3 wells feature the sedimentation of channels in braided
river delta, where fine siltstone dominates; while Ji 007,
Shinan 21, Shi 106 and Shi 109 wells are represented by the
dark-mudstone sedimentation of flood-plain (Fig. 2). In the
time span of sedimentation of the short base-level cycle
SSQ2, the branch channels of Shinan 21 well field attained
the climax of its great prosperity, nearly extending
throughout the entire zone, at which time the sedimentary
sand was mainly represented by channel sand and
channel-bar sand. In the sedimentation stage of SSQ3, the
braided branch channels in Shinan 21 well field changed its
route towards Shinan 3 well field, due to which the braided
branch channels merely developed in Shinan 3 well field,
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455

Fig. 2 Illustration of isochronal stratigraphic frame of Toutunhe Formation for Ji 004 - Shinan 3 wells

Fig. 3 Illustration of isochronal stratigraphic frame of Toutunhe Formation for Shi 114 - Shi 107 wells

while the rest wells were principally represented by the
dark-mudstone sedimentation of inter-braided channel
flood-plain. SSQ4 was retained only in Shinan 3 well field,
which, similarly, was represented by the sedimentation of
braided branch channel.
On the well-connecting line along the SN structural strike,
Shi 114 through Shi 107 wells were represented by the
sedimentation of braided-river delta flood-plain in the
sedimentary time span of the short base-level cycle SSQ1;
Shi 114 and Shi 107 well fields were dominated by the
brownish mudstone featuring the oxidation environment of
flood-plain, while the remaining well fields by the dark
mudstone featuring the weak reducing environment (Fig. 3).
During the sedimentation of the short base-level cycle SSQ2,
the braided branch channels of braided-river delta plain
nearly extended across the entire zone, and the
dark-mudstone sedimentation of inter-braided channel
flood-plain developed in a handful of well fields (including
Shinan 22, and Shi 107) in the northern part of Shinan 21
well field. In such a stage, the sedimentary sand was mainly
represented by the channel sand and the channel-bar sand,
which, together with the lateral flood-plain dark-mudstone,
could form the sand lateral pinch-out lithologic oil/gas pools;
being of certain heterogeneity; the intra-channel sand
reservoirs would also constitute lithologic traps. During the
sedimentation of SSQ3, Shi 114, Shinan 21 and Shi 107
wells were apparently represented by the extensive
sedimentation of dark-, brownish mudstone of the
flood-plain facies.
On the basis of well-connecting isochronal stratigraphic
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455
Fig. 4 Average velocity field illustration for SB3, with the sequence frame constraint incorporated (a) and not incorporated (b)
Fig. 5 Structural map for SB3 after well depth correction
conducted to 9 wells
frame, and by leveraging the 3D seismic interpretation,
full-zone tracking has been winded up (280 km2) for 3
sequences (i.e., SB1, SB3, and SB4), and the 3D sequence
stratigraphic frame has been thus set up. Due to the
isochronal sedimentation within each sequence, any
sedimentation body falling between SB1 and SB3 or
between SB3 and SB4 may be studied separately via seismic
techniques on the basis of high-resolution sequence
stratigraphic researches; relative to the reservoir prediction
based on the conventional horizon interpretation (SB1 and
SB4 to be interpreted only), the control of high-resolution
sequence stratigraphic frame could effectively improve the
precision in reservoir prediction.
3 Structural variable speed imaging technique
under the constraint of sequence frame
Fig. 4a (constraint of sequence frame incorporated) and b
Fig. 6 Structural map for SB3 after well depth correction
conducted to 20 wells
(no constraint of sequence frame incorporated) show the
average velocity field of SB3, as resulted from the inversion
based on velocity spectrum data[11]; as indicated therein, the
average velocity field as constrained by sequence frame
presents a more marked regularity, while the random noise
would exist when no such constraint is imposed. It indicates
that, with the constraint of sequence stratigraphic frame, the
average velocity field could overcome partial impacts
arising from the seismic data itself and artificial
interpretation, consequently reflecting, in a more objective
manner, the velocity field characteristics of the target
interval, in addition to improving the precision of structural
variable speed imaging.
Figs. 5 and 6 are the structural graphs after depth
correction is conducted to 9 wells and 20 wells, respectively;
evidently, the structural patterns are of high consistency.
After the structural graph is plotted for SB3 on the basis of
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455

Table 1 Depth prediction and error for each non-drilled well from the structural map in various rounds

Penetration Depth fetched from structural map / m Prediction error / m

Well ID
depth / m 9 wells 12 wells 15 wells 18 wells 9 wells 12 wells 15 wells 18 wells
Shi 112 2 531.000 2 526.760 4.243
Shi 113 2 504.000 2 507.250 - 3.254
Shi 114 2 581.000 2 573.230 7.773
Shi 115 2 442.500 2 443.200 2 446.610 - 0.704 - 4.101
Shi 116 2 457.000 2 457.840 2 459.560 - 0.836 - 2.560
Shi 117 2 435.000 2 438.450 2 439.470 - 3.457 - 4.469
Shi 118 2 502.500 2 494.150 2 494.990 2 492.670 8.354 7.510 9.829
Shi 119 2 458.000 2 451.830 2 453.810 2 451.340 6.169 4.194 6.657
Shi 120 2 510.000 2 505.590 2 506.280 2 504.150 4.411 3.718 5.843
Shi 121 2 504.500 2 504.820 2 506.370 2 503.950 2 504.550 - 0.321 - 1.878 0.550 - 0.053
Shi 122 2 430.000 2 427.780 2 429.860 2 429.860 2 430.110 2.217 0.140 0.131 - 0.106

Fig. 7 Waveform classification of formations between sequence interfaces

the velocity field resulted from sequence frame constraining,
the drilling history is simulated for the exploration appraisal
wells; firstly, well-site depth correction is carried out to the
9 appraisal wells in the first batch of drilling (namely, Ji 004,
Ji 007, Shinan 3, Shinan 22, Shinan 21, Shi 106, Shi 107,
Shi 108, and Shi 109 wells); thereafter, the well-site depths
of the other 11 wells are fetched for the comparison against
the penetration depth of each well, so as to acquire the
predicted depth and prediction error of each well; such a
process is then iterated for 12, 15, and 18 appraisal wells
respectively, so as to get the predicted depth and prediction
error of each of the other 8, 5, and 2 wells (see Table 1).
According to Table 1, the maximum error in predicted depth
is merely 10 m (target formation: about 2 500 m in depth),
with the relative prediction error averaged at 0.16%. That is,
a higher precision would be attained for the structural
variable speed imaging on the basis of average velocity field
resulted from the inversion of velocity spectrum data under
the constraint of sequence frame.
4 Reservoir prediction techniques under
constraint of sequence frame[12-14]
Fig. 7 shows the comparison between the waveform
classification graphs on the basis of conventional horizon
interpretation and the waveform classification graphs of
inter-sequence interface seismic response; it shall be noted
that the formation between sequence interfaces of SB1 and
SB3 includes the target reservoir. As indicated by the figure,
the classification result of seismic waveforms between
conventional horizon interpretation interfaces is nearly
equivalent to the superimposition of the classification result
between SB3 and SB4 interfaces with that between SB1 and
SB3 interfaces. That is, the set-up of high-resolution
sequence stratigraphic frame could differentiate relatively
small sedimentary bodies, so that the seismic attribute
analysis could be conducted specifically to a certain small
sedimentary body; as a result, the reservoir prediction
precision could be improved in the application of seismic
attribute analysis technique.
Fig. 8 shows the logging constraint inversion profile
through Well Shinan 21 plotted on the basis of conventional
horizon interpretation results, which could be compared
with the counterpart introducing high-resolution sequence
stratigraphic frame (Fig. 9). According to the two inversion
profiles, the overall reservoir development characteristics
are consistent, with the latter presenting a more clear-cut
depiction of the reservoir lateral variation particulars. After
the introduction of the high-resolution sequence
stratigraphic frame, the initial geological model approaches
much closer to the actual sand spreading (both transverse
and longitudinal), thus improving the reservoir prediction
precision in the application of seismic inversion.
Figs. 10 and 11 show the thickness of sandstone, which is
resulted from seismic inversion under the constraint of
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455
Fig. 8 Conventional horizon interpretation results through Well Shinan 21
Fig. 9 Seismic inversion profile through Well Shinan 21, as constrained by sequence frame
sequence stratigraphic frame, for 9 wells and 20 wells,
respectively. Obviously, the sand patterns in the two figures
merely present low inconsistency; however, difference
exists with the thickness of sand bodies. The drilling history
of the exploration and appraisal wells is simulated as
follows: firstly, carrying out logging restrained inversion to
the 9 appraisal wells of the first batch of drilling (namely, Ji
004, Ji 007, Shinan 3, Shinan 22, Shinan 21, Shi 106l, Shi
107, Shi 108, and Shi 109 wells). On the resultant sandstone
thickness maps, the predicted sandstone thickness for the
other 11 wells is fetched, which is compared with the actual
value of each well to calculate the prediction error
concerning the sandstone thickness for respective well
(Table 2); then, such a process is iterated for the cases of 12,
15, and 18 appraisal wells, so as to obtain the prediction
errors of the other 8 wells, 5 wells, and 2 wells, respectively.
According to the table, given the target layer with the burial
depth of about 2 500 m and the appraisal well with effective
sandstone thickness averaged at 13 m, the maximum
prediction error in sandstone thickness is about 5 m.
Provided that the Shi 118, Shi 119, Shi 120, and Shi 121
wells are regarded as a sample set, the average relative error
is 28.87%, no matter whether 9 wells or 12 wells are used
for the prediction; when 15 wells are used for the prediction,
the average relative error of the sample set arrives at 26.44%.
In other words, even when the number of wells used for the
prediction varies largely, the improvement in prediction
precision is not evident, indicating that, whereas the
high-resolution sequence stratigraphic frame has established
a rational sedimentation comparison relation, seismic
inversion could, with a less number of known wells, still
attain the prediction precision equivalent to the case with a
bigger number of known wells. Evidently, with the
sequence frame being imposed, effective improvement
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455

Fig. 10 Sandstone thickness predicted by seismic inversion with 9 wells constrained by sequence frame

Table 2 Depth prediction and error for each non-drilled well from the structural map in various rounds

9 wells 12 wells 15 wells 18 wells

Well Actual
Predicted Prediction error Predicted Prediction error Predicted Prediction error Predicted Prediction error
Name thickness / m
thickness / m /m thickness / m /m thickness / m /m thickness / m /m
Shi 112 21.000 18.000 3.000
Shi 113 14.000 13.000 1.000
Shi 114 12.000 11.700 0.300
Shi 115 0.000 1.000 -1.000 1.000 -1.000
Shi 116 20.000 15.000 5.000 15.000 5.000
Shi 117 19.000 18.300 0.700 18.300 0.700
Shi 118 5.000 6.200 -1.200 6.200 -1.200 6.400 -1.400
Shi 119 2.000 1.000 1.000 1.000 1.000 1.100 0.900
Shi 120 14.000 12.000 2.000 12.000 2.000 12.500 1.500
Shi 121 19.500 14.200 5.300 14.200 5.300 15.200 4.300 16.300 3.200

could be seen with the precision of both seismic attribute sedimentary reservoir of Jurassic Toutunhe Formation,
analysis and seismic inversion. Shinan 21 well field, Junggar Basin, as well as the analysis
on error, the set-up of high-resolution sequence stratigraphic
5 Conclusions frame would play a significant role in improving the
As indicated by its application in the braided-river precision in seismic prediction, since the structural variable
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455
Fig. 11 Sandstone thickness predicted by seismic inversion with 20 wells constrained by sequence frame
speed imaging under the constraint of sequence frame,
together with the reservoir prediction technique under such
constraint, could be highly effective in achieving this goal.
The study area elaborated in this paper concerns a cluster
of exploration and appraisal wells (up to 20 wells), and
abundant data are available for the classification of short
base-level cycles. Nevertheless, the scheme provided herein
could also be followed by the lithologic reservoir
exploration in the zones with small number of wells.
Whereas the objects of hydrocarbon exploration become
increasingly subtle and complicated, a better answer for
improving reservoir prediction precision would inevitably
rest with the combination of sequence stratigraphy with the
reservoir prediction techniques.
References
[1] Hu Jianyi. The symposium of subtle oil reservoirs exploration in
China. Harbin: Heilongjiang Science and Technology Press,
1984.
[2] Xiang Kui, Bao Zhidong, Zhuang Wenshan. Petroleum geology
and exploration of beach sandbar in Junggar Basin, NW China:
A case from the Shawan Formation in Well Pai 2. Petroleum
Exploration and Development, 2008, 35(2): 195-200.
[3] Pan Yuanlin. Subtle oil reservoirs in China. Beijing: Geological
Publishing House, 1998.
[4] Jia Chengzao, Chi Yingliu. The symposium of the 3th subtle oil
reservoirs international conference. Beijing: Petroleum Industry
Press, 2004.
[5] Jia Chengzao, Zhao Wenzhi, Zou Caineng, et al. Two key
technologies about exploration of stratigraphic/lithological
reservoirs. Petroleum Exploration and Development, 2004,
31(3): 3-9.
[6] Guan Shuwei, Li Benliang, Hou Lianhua, et al. New
hydrocarbon exploration areas in footwall covered structures in
northwestern margin of Junggar Basin. Petroleum Exploration
and Development, 2008, 35(1): 17-22.
[7] Cross T A, Lessenger M A. Sediment volume partitioning:
 Li Jinsong et al. / Petroleum Exploration and Development, 2009, 36(4): 448–455
Rational for stratigraphic model evaluation and high resolution
stratigraphic correlation. Gradstein F M, Sandvik K O, Milton N
J, et al. Sequence stratigraphy concepts and applications. Oslo:
NPF Special Publication, 1998. 171-195.
[8] Cross T A, Baker M R. Applications of high-resolution
sequence- stratigraphy to reservoir analysis. Eschard R, Doligez
B. Subsurface reservoir characterization from outcrop
observations. Paris: The 7th IFP Exploration and Production
Research Conference, 1993.
[9] Deng Hongwen. A new school of thought in sequence
stratigraphic studies in U.S.: High resolution sequence
stratigraphy. Oil & Gas Geology, 1995, 16(2):89-97.
[10] Deng Hongwen, Wang Hongliang. High resolution sequence
stratigraphy: Theory and application. Beijing: Geological
Publishing House, 2002.
[11] Zhao Bangliu. Application of multi-component seismic
exploration in the exploration and production of lithologic gas
reservoirs. Petroleum Exploration and Development, 2008,
35(4): 397-409, 423.
[12] Li Qingzhong. The road to precision exploration. Beijing:
Petroleum Industry Press, 2003.
[13] Liu Wenlin. The seismic technology of oilfield development.
Beijing: Petroleum Industry Press, 1996.
[14] Yao Pengchang, Gan Lideng. Application and restriction of
seismic inversion. Petroleum Exploration and Development,
2000, 27(2): 53-56.