An Amplitude-normalized Pseudo Well-log Construction Method and its Application on AVO Inversion in a Well-absent Marine Area
An Amplitude-normalized Pseudo Well-log Construction Method and its Application on AVO Inversion in a Well-absent Marine Area
Chunyan FAN 1
Yan SONG 1
Yuanyin ZHANG 0 1
Zhenxue JIANG 1
0 Petroleum Exploration & Production Research Institute , SINOPEC, Beijing , China
1 China University of Petroleum - Beijing , Beijing , China
A b s t r a c t AVO inversion is hard to be efficiently applied in unexploited fields due to the insufficiency of well information. For the sake of AVO inversion in a well-absent area, the most conventional method is to construct pseudo well-logs by defining seismic processing velocity as the P-velocity and computing S-velocity and density using empirical formulas, yet the resolution of the corresponding earth models and final inverted results could be extremely low, and a rough formula could destroy the inversion thoroughly. To overcome this problem, an amplitudenormalized pseudo well-log construction method that reconstructs pseudo well-logs in accordance with computed P-wave reflection amplitudes and nearby drilling data is proposed in this paper. It enhances the inversion resolution efficiently with respect to the real elastic parameter relationships, so that the corresponding AVO inversion results are reasonably improved. In summary, the proposed method is successfully applied in the AVO inversion of a well-absent marine area, and could be valuable in the early phase, particularly of the offshore hydrocarbon exploration.
well-absent pre-stack seismic AVO inversion; marine exploration; reservoir appraisal; pseudo well-logs reconstruction
The pre-stack seismic AVO inversion is better for reservoir prediction than
post-stack methods owing to its preservation of the AVO information and
more elastic parameters (Connolly 1999, Zhang et al. 2011b), but it cannot
be efficiently conducted if the well information is insufficient to accurately
scale the wavelets and build constrained elastic models. Because the
constrained models essentially represent initial predictions to construct
inversion, and can lead to a better resolvability and link between the seismic data
and the actual lithology (e.g., from logging data and geological mapping),
the inversion result of predicted sediment properties is actually a set of
elastic volumes that deviate as little as possible from the initial predictions,
while at the same time they are modeling the real geology as closely as
possible (Hampson et al. 2001, 2005).
For a convenient application of AVO inversion method in a well-absent
area, the most conventional way is to construct some pseudo well-logs by
defining interval velocities obtained from 3D seismic processing stage,
initially as the P-velocities, and then compute S-velocities and densities using
empirical formulas. However, by this method the resolutions of
corresponding earth models and final results directly inverted from seismic processing
velocities could be extremely low, and using a rough formula for elastic
parameters prediction could further destroy the inversion (Zhang et al. 2011b,
Fan et al. 2013).
In this paper, in order to increase the resolution of well-absent AVO
inversion results and hence retrieve reliable elastic parameters for reservoir
description, we propose a new pseudo well-log construction method and apply
it in an undrilled marine study area, where quite promising but not
wellstudied reef and turbidity sand reservoirs need to be appraised for further
2. GEOLOGICAL SETTING AND DATA SETS
As shown in Fig. 1, the study field is located in the northwestern segment of
South China Sea. It lies offshore southern Hainan trending
northwestsoutheast with an areal extent of around 200 km2; the nearest drilling site is
about 60 km away from the study area, marked as well Y-1 on the index
map. The water depth gradually increases from 1.2 s (around 600 m) in the
northwest to 1.6 s (around 800 m) in the southeast, whereas a seabed pit can
be distinctly seen in the southern segment.
The pre- and post-stack 3D seismic data quality is quite good for seismic
stratigraphic interpretation and AVO analysis. The bin size of CMP gathers
is 12.5 × 25 m, with a full fold of 140. As shown in Fig. 2, the seismic traces
are characterized by a frequency content from 8 to 75 Hz with the
People's Republic of China
Fig. 1. Location of the study field, about 60 km away from Well Y-1 (a). Its seafloor
map (b) shows that the water depth gradually increases from 1.2 s (around 600 m) in
the northwest to 1.6 s (around 800 m) in the southeast, whereas an obvious seabed
pit exists in the southern segment.
Potential reef reservoir
Fig. 2. The typical seismic section following line AB in Fig. 1b. Two sets of
potential reservoirs, separately marked by yellow and green arrowheads, need to be finely
described for further exploration.
nant frequency of 34 Hz. The main stratigraphy summarized from the
nearest drilling area is listed in Table 1. It indicates that Meishan (T40 to T50)
and Lingshui (T60 to T70) fm are the two target formations, prone to
develop reservoirs in carbonate reef and turbidity sand units, respectively.
T a b l e 1
Stratigraphy summarized from the nearest drilling location
The top and base units of two potential reservoir intervals can be clearly
identified from the post-stack seismic reflection data by their very strong
reflection amplitudes (Fig. 2); the time thickness distributions of Meishan and
Lingshui formations (Fig. 3) have roughly revealed that both of the two sets
of reservoirs lie in an area of large extent. Unfortunately, AVO inversion
cannot be efficiently applied to retrieve valuable elastic parameters for
Fig. 3. The time thickness maps of Meishan (T40-T50) (a), and Lingshui (T60-T70)
fm (b). WA, WB, WC, and WD are the four pseudo wells.
voir description since there is no well currently drilled in the study area. The
drilling cost and risks in a marine environment, especially in deep waters,
are actually a few times higher than that of land exploration. Therefore, it is
crucial not only to perform a detailed reservoir assessment or cautious
welldeployment but also to carry on an AVO study without well-logging data.
On the contrary, the lack of the logging data in the early exploration phase
also restricts the reservoir prediction accuracy. This common contradiction is
essentially the key problem in the study area.
Different from the conventional method by directly defining seismic
processing velocity as the P-velocity and computing S-velocity and density using
empirical formulas, the new method proposed in this paper reconstructs, at
first, the pseudo well-logs incorporated with computed P-wave reflection
amplitudes and an actual well-log data from the nearest drilling site. It then
conducts pre-stack simultaneous AVO/AVA inversion for a high-quality
description of elastic reservoir parameters. The P-wave data are computed
from CRP (common reflection point) gathers by AVO inversion. As shown
in Fig. 4, this method consists of three major approaches, including the
initial pseudo well-logs construction, reconstruction, and AVO inversion.
Vp- , Vp-Vs Relationship,
Initial pseudo logs
Reconstructed pseudo logs
R( 1) Logs
Pre-stack Simultaneous Inversion
Fig. 4. The well-absent AVO inversion workflow based on amplitude-normalized
pseudo well-logs construction method. The original data are marked by gray shades.
The Vp- and Vp-Vs relationships, as well as P-impedance distribution, are achieved
from nearby drilling data. The P-wave data computed from CRP gathers by AVO
inversion have a higher resolution, and are more beneficial for reservoir prediction
and pseudo well-logs reconstruction.
3.1 Initial pseudo well-logs construction
The first step, constructing P-velocity for a pseudo well from interval
velocity, is probably the same as in the conventional method. Although the
interval velocities defined in 3D seismic processing stage could be carefully and
densely picked, its dominant frequency (related to seismic velocity picking,
often less than 5 Hz) is still inferior to that of seismic data (around 20-40 Hz),
and far below that of logging data (often > 1000 Hz). Thus, the
corresponding final inverted results are liable to have low resolutions and incorrect
Besides, as mentioned above, using a rough formula for elastic
parameters prediction could destroy inversion thoroughly (Zhang et al. 2011b, Fan
et al. 2013). Therefore, comparing to empirical formulas (for instance,
Castagna’s mud-rock line and Greenberg-Castagna’s Vp-Vs relationship, etc.;
Lamb et al. 1992), it is preferable to use the Vp-Vs and Vp- relationships
summarized from nearby drilling data to compute S-velocity and density,
3.2 Pseudo well-logs reconstruction
To reasonably improve the inversion resolution, the amplitude-preserved
seismic data, drilling information and seismic processing velocity are
integrated to reconstruct pseudo well-logs. The basic methodology is that the
magnitudes and signs of seismic reflections are correlative with those of the
subsurface layers if the stacked seismic data are amplitude-preserved. The
reflectivity variations are also related to the changes of P-impedance (PI).
The pseudo PI well-logs can therefore be reconstructed by converting
fullstack seismic reflection to elastic parameters following the PI tendency.
3.2.1 P-wave data computation
As the full-stacked data have made a reluctant compromise to data resolution
sacrificing due to the contamination of AVO effects no matter how much the
SNR degree could be enhanced in the stacking process (Zhang et al. 2013),
we compute the P-wave data from CRP gathers by AVO inversion to make
a corroborating substitution, based on Eq. 1 (Gidlow 1992, Sun 1999):
2 Vs2 sin2
where Rp and Rs are theoretical zero-offset P-wave reflectivity and S-wave
and Vp, Vs, and are the average P-wave velocity, S-wave velocity, and
density of the reflection boundary, respectively. is the average of incidence
and transmission angles, Rpp( ) is the elastic reflectivity with ray-path of
incident angle, / is the density gradient, and Vs/Vp is the S-to-P-wave
For solving the non-linear Eq. 1, the P-wave result is comparatively
stable since the coefficient of Rp is usually bigger than that of Rs or / , and
has nothing to do with Vp/Vs (Zhang et al. 2013). If we use seismic
amplitudes of CRP gathers to substitute Rpp( ), then the computed corresponding
Rp results are the zero offset P-wave reflection data. Comparing to
conventional full-stack data, the computed P-wave data have a higher resolution
and are more beneficial for reservoir prediction (Zhang et al. 2013),
inevitably for pseudo well-logs reconstruction.
3.2.2 P-impedance tendency computation
The P-impedance tendency (PI0) is computed by the multiplication between
seismic interval velocity and density, while density is computed from the
statistical relationship of the nearest drilling data instead of any empirical
3.2.3 Pseudo well-logs reconstruction
The amplitude-normalized pseudo-well log reconstruction equation for each
sample in each trace can be listed as follows:
where PI is the reconstructed P-impedance, PI0 is the P-impedance tendency,
and x is the reflection amplitude for P-wave data in that sample. [A, B] and
[C, D] are the distribution ranges of computed P-wave reflection amplitudes
and recorded P-impedance values from the nearby drilling data, respectively.
The density and S-impedance of pseudo well-logs can be computed from
the statistical relationships from nearby drill well logs too.
3.3 AVO inversion conduction
With these reconstructed pseudo well-logs, the pre-stack simultaneous
AVO/AVA inversion can be conducted to achieve a great number of elastic
parameters such as PI, SI (S-impedance), Vp/Vs, Poisson ratio, , and ,
etc., for reservoir prediction. The reconstructed curves have higher dominant
frequencies (same as that of seismic data), which could be definitely helpful
for scaling wavelets and building initial models. The final AVO inverted
results could therefore be improved.
4. COMPARISON OF RESULTS
Figure 5 shows the interval velocity profile defined in seismic processing
stage of study area in the same line as in Fig. 2. It has been carefully and
densely picked, with an interval of 100 × 100 m. Besides, the amplitude of
CRP gathers analyzed in this study is well-preserved and qualified for
P-wave data computation via AVO inversion because of the careful seismic
data processing, including signal to noise ratio (SNR) enhancement,
consistency and vertical resolution improvement, amplitude-preserved pre-stack
migration, as well as data conditioning (Zhang et al. 2011a, b).
As shown in Fig. 6, comparing with conventional full-stack data
(Fig. 6a), the inverted P-wave data from CRP gathers by AVO inversion
(Fig. 6b) have a higher resolution with more clear and continuous events.
The dominant frequency of inverted P-wave data is 39 Hz, while the
counterpart of stack data is only 34 Hz on the whole section. This enhancement is
even higher in some target zones, and crucial for pseudo well-logs
Figure 7 shows the comparison of seismic (a), synthetic (b), constructed
P-impedance (PI) (c), and extracted wavelet (d) by conventional method for
pseudo-well WC. The constructed pseudo PI log inevitably has a very low
Fig. 5. The seismic processing velocity profile in the same line as in Fig. 2.
Fig. 6. Comparison between full-stack section (a) and inverted P-wave section (b)
across the same profile as in Fig. 2. After the removal of AVO effects, seismic
reflection events are more clear and continuous on the inverted P-wave section, and
the dominant frequency is enhanced by 5 Hz compared with full-stack data across
the whole section.
frequency bandwidth and definitely lost many details, thus the extracted
wavelet energy has to be increased to compensate for the tiny reflection
variations. Consequently, it is failed to produce a qualified correction
between synthetic and seismic data. The extraction of reflection coefficients in
a Seismic b Synthetic
Fig. 7. The comparison of seismic (a), synthetic (b), constructed pseudo
P-impedance (PI) (c), and extracted wavelet (d) by conventional method for pseudo-well WC.
The constructed pseudo PI well-log with an extremely low dominant frequency has
inevitably smoothed many details and irrationally boosted up the energy of extracted
wavelets, which subsequently cause a bad-correlation between synthetic and seismic
a Seismic b Synthetic
Fig. 8. The comparison of seismic (a), synthetic (b), reconstructed pseudo
P-impedance (PI) (c), and extracted wavelet (d) by our proposed method for pseudo-well
WC. The dominant frequency of pseudo well-log is higher than that of conventional
counterpart (Fig. 7c).
Fig. 9. Comparison
between inverted PI
and / results across
the same line as in
Fig. 2 based on
conventional (a and b)
and proposed (c and
d) methods. The
conventional result is much
less than that of
Note that the
conventional PI result is
highly correlated to
velocity (Fig. 5), and
both conventional PI
and / results show
distributions of reservoirs,
while the new results
are more correlative
to seismic reflections
and are more
beneficial for reservoir
the conventional pseudo well-logs directly from interval velocities is
obviously not applicable. It is not difficult to forecast that the corresponding in
vert resolution will not be qualified, and the computed results can hardly end
up with good correlations to reservoir parameters. On the contrary, the
reconstructed curve by the method we proposed has a higher dominant
frequency (same as that of seismic data), as shown The extracted wavelet has
produced a higher correlation rate made by correlating synthetic
seismograms of reconstructed reflection coefficients and original seismic signals
(Fig. 8a, b), which is quite essential to improve the inversion results.
A further comparison will be illustrated in Fig. 9.
With these pseudo well-logs, the pre-stack simultaneous AVO/AVA
inversion can be conducted to achieve a great number of elastic parameters,
such as PI, SI (S-impedance), Vp/Vs, Poisson ratio, , and , etc., for
reservoir prediction. In particular, both the two sets of pseudo well-logs
constructed from the conventional and proposed methods are employed to
conduct pre-stack simultaneous AVO/AVA inversion. Figure 9 shows a
comparison between the two sets of inverted PI and / results across the same
line based on conventional (a and b) and new (c and d) methods,
respectively. Obviously, the resolution of conventional result is much lower than that
of proposed competitor, as indicated by these illustrations. In particular, the
conventional PI result is highly correlated to seismic processing velocity
(Fig. 5), and both conventional PI and / results have very limited value
distributions. On the other hand, the new results by the proposed method are
more correlative to seismic reflections and are definitely more beneficial for
5. APPLICATION IN RESERVOIR PREDICTION
Specifically, as shown in Fig. 10b, the cross-plot of / and PI, that is
usually regarded as the best reservoir indicator, is employed for reef and sand
reservoir discrimination by separately tracking appropriate data following
the yellow and green polygons. The two polygons are similar to that of the
nearest drilling place. Thus, the predicted reef (yellow) and sand reservoir
(green) can be described, respectively, and elaborately plotted in Fig. 10a.
The time thickness maps of predicted reef (Fig. 11b) and sand reservoirs
(Fig. 11d) could also be further mapped by summing these qualified samples
in a 3D area within two sets of formations, respectively. The two sets of
reservoirs have been better delineated by the comparison with their
corresponding formation thickness maps (Fig. 11a and c), and they are useful for further
exploration of the reservoirs.
Platform marginal reef reservoir
Turbidity sand reservoir
Fig. 10. The cross-plot of inverted PI and / attributes (b), and the predicted
reservoir distributions in the same line as in Fig. 2 (a).
Fig. 11. The time thickness maps of Meishan (T40-T50) (a), and Lingshui
(T60T70) fm (c), the predicted reef (b), and turbidity sand (d) reservoir distributions in
the two formations, respectively.
It is a common method to construct pseudo well-logs for AVO inversion in
unexploited fields that are short of wells, especially in marine cases. But
most of conventional methods by directly defining P-velocities from seismic
processing and computing the S-velocity and density using empirical
equations are not qualified to scale the extracted source wavelets and build
constrained models. They tend to produce very low resolution AVO inversion
results. On the contrary, the method we proposed, that reconstructs pseudo
well-logs in accordance with reflection amplitudes of computed pure P-wave
data from CRP gathers (AVO) and drilling information from a close area, is
proven to be more beneficial for AVO inversion and more suitable for
defining the reservoir prediction parameters. However, it should be noted, in
particular, that the CRP data used in this method should have high quality.
A c k n o w l e d g m e n t s . The authors thank National Basic Research
Program of China (Grant No. 2011CB201103), the National Science and
Technology Major Project (Grant No. 2011ZX05004003), the Science
Foundation of China University of Petroleum-Beijing (KYJJ2012-05-03), and
Chinese Scholarship Council for financial support.
R e f e r e n c e s
Connolly , P. ( 1999 ), Elastic impedance , The Leading Edge 18 , 4 , 438 - 452 , DOI: 10.1190/1.1438307.
Fan , C. , Y. Song , Z. Jiang , Y. Zhang , and M. Sun ( 2013 ), Reservoir prediction in a well-absent area - a case study in a marine area . In: 75th EAGE Conference and Exhibition incorporating SPE EUROPEC 2013 , 10 - 13 June 2013, London, UK, Extended abstracts, DOI: 10.3997/ 2214 - 4609 .20130877.
Gidlow , P.M. , G.C. Smith, and P.J. Vail ( 1992 ), Hydrocarbon detection using fluid factor traces: A case history . In: Joint SEG/EAEG Summer Research Workshop on “How Useful is Amplitude-Versus-Offset (AVO) Analysis?”, Expanded abstracts , 78 - 89 .
Hampson , D.P. , J.S. Schuelke , and J.A. Quirein ( 2001 ), Use of multiattribute transforms to predict log properties from seismic data , Geophysics 66 , 1 , 220 - 236 , DOI: 10.1190/1.1444899.
Hampson , D.P. , B.H. Russell , and B. Bankhead ( 2005 ), Simultaneous inversion of pre-stack seismic data . In: 75th Annual International Meeting SEG, Expanded abstracts, Society of Exploration Geophysicists , 1633 - 1637 .
Lamb , W.J. , X.H. Zhu , G.A. McMechan , M.L. Greenberg , and J.P. Castagna ( 1992 ), Elastic wave propagation in composite media , Geophysics 57 , 9 , 1155 - 1165 , DOI: 10.1190/1.1443329.
Sun , Z. ( 1999 ), Seismic methods for heavy oil reservoir monitoring and characterization, Ph.D. Thesis , University of Calgary, Calgary, Canada.
Zhang , Y. , Z. Sun , C. Fan , and H. Bai ( 2011a ), Data conditioning for pre-stack inversion - a case study from Xingma area, Liaohe Oil field, China . In: 73rd EAGE Conference and Exhibition incorporating SPE EUROPEC 2011 , 23 - 26 May 2011 , Vienna, Austria, Extended abstracts, DOI: 10.3997/ 2214 - 4609 .20149593.
Zhang , Y. , Z. Sam , H. Yang , H. Wang , J. Han , H. Gao , C. Luo , and B. Jing (2011b), Pre-stack inversion for caved carbonate reservoir prediction: A case study from Tarim Basin , China, Pet. Sci. 8 , 4, 415 - 421 , DOI: 10.1007/s12182- 011 - 0159 -4.
Zhang , Y. , Z. Sun , and C. Fan ( 2013 ), An iterative AVO inversion workflow for S-wave improvement . In: 75th EAGE Conference and Exhibition incorporating SPE EUROPEC 2013 , 10 - 13 June 2013, London, UK, Extended abstracts, DOI: 10.3997/ 2214 - 4609 .20130268.