Sei sulla pagina 1di 4

FOCUS ARTICLE

Coordinated
Coordinated
by Robby
Holt
Rob Holt

How much can I stretch and squeeze?


David Cho
Q E Y E L A BS C A N A DA

David Nordin
S I T K A E X PLO R AT I O N

Well to seismic ties is a fundamental step in seismic inter-


pretation. It relates subsurface measurements
obtained at a wellbore measured in depth and seismic data measured
The well to seismic ties can vary in quality and any mis-ties are typically
fine-tuned by stretching and squeezing the well logs to adjust the
time-depth relationship. Numerous justifications can be made regarding
in time. A time-depth relationship is typically computed by integrating the validity of stretching and squeezing. These include measurement
the slowness function measured at a wellbore. Mis-ties are often errors in the well logs and dispersion effects arising from differences in
present and adjustments to the time-depth relationship are typically the measurement frequencies, where the seismic bandwidth is typically
fine-tuned by stretching and squeezing of the well logs. However, the on the order of 100 Hz and logging frequencies are in the kilohertz
validity of stretching and squeezing is sometimes questioned and the range. Although these are valid reasons for manually adjusting the
question of how much stretching and squeezing is acceptable remains time-depth relationship, many practitioners still struggle with the
a topic of debate. idea of applying these seemingly arbitrary corrections. Moreover, the
question of how much stretching and squeezing is justifiable presents
In this study, we investigate the idea of using a Backus averaged well another major hurdle in our quest for an improved well to seismic tie.
log that accounts for dispersion resulting from layer induced long
wave elastic attenuation to compute a time-depth relationship. We In the following, we review the theory as presented by Liner (2014)
review the theory to compute a local Q function that results in both to compute a local Q function arising from layer induced long wave
positive and negative Q values, which imply that a mandatory stretch elastic attenuation. Subsequently, we show that the well to seismic tie
and squeeze is required when dispersion is present. We illustrate with can be improved by using a Backus averaged well log to compute the
an example from Western Canada where the presence of low velocity time-depth relationship.
coals results in large errors in the well to seismic tie. We show that
these mis-ties can be reduced by using a Backus averaged well log
Local Q attenuation
to compute the time-depth relationship, which accounts for velocity
differences between the logging and seismic frequencies. In addition, The propagation of seismic waves in a horizontally layered earth results
the ratio of the logging and seismic velocities provides an indication of in an apparent attenuation of the seismic energy that is constant Q
how much stretching and squeezing is justifiable. in nature (Liner, 2014). This attenuation can be related to dispersion
through the exact relationship given by Kjartansson (1979),

Introduction ⎛ f ⎞
γ
1 ⎛ 1 ⎞
V ( f ) = V0 ⎜⎜ ⎟⎟ ; γ = tan −1 ⎜⎜ ⎟⎟   (1)
f
⎝ 0 ⎠ π ⎝ Q ⎠
Well to seismic ties are fundamental in seismic interpretation as it
provides the link between subsurface properties measured directly where V(f) is the velocity corresponding to frequency f, V0 is the
at a wellbore and the remotely sensed data obtained through seismic reference velocity corresponding to reference frequency f0 and Q
sounding. Well logs measured in depth and seismic data measured is the quality factor. Equation 1 suggests that a finite Q will result in
in time require a time-depth relationship to facilitate the conver- different frequencies travelling at different velocities. Therefore, with
sion between the two different measurement domains. This could be the assumption that we have a constant Q earth, the velocities corre-
achieved through a checkshot obtained by vertical seismic profiling sponding to the logging frequencies and seismic frequencies are
(VSP) or the integration of the slowness function obtained by acoustic different and are related through equation 1.
logging, where differences between the two methods were discussed
by Gretener (1961). As checkshots from a VSP are not always avail- Furthermore, Liner (2014) argues that Backus averaging (1962) can be
able, a time-depth relationship is typically computed using the sonic used to determine the associated dispersion. Because the seismic
measurements obtained through well logging. frequencies are much lower than logging frequencies, the rapidly

54 C SEG RECORDER DECEMBER 2 014


varying substructure at smaller length scales from Western Canada where the presence of low velocity coals results in large errors in the well
relative to the seismic wavelength will not be to seismic tie. Our target interval is the Glauconite and Ostracod formations in the Manville
seen by the seismic waves. Backus averaging group, which is overlain by the Top Manville and Medicine River coals. These low velocity
of the well log velocities then yields the coal zones generate a number of complex wave propagation phenomena including multiple
velocity of the effective medium as seen generation, transmission filtering (Coulombe and Bird, 1996) and traveltime effects that result
by the seismic waves. Equation 1 can be in exploration challenges in the underlying formations. In this study, we are interested in the
rearranged to solve for Q as traveltime effects and the associated time-depth errors in the well to seismic tie.

⎛ π log[V V0 ] ⎞ Figure 1 shows the well to seismic tie using a time-depth relationship computed by integrating
Q = cot⎜⎜ ⎟⎟   (2)
⎝ log[V V0 ] + log[λ0 λ ] ⎠ the well log slowness. The panels from left to right show the acoustic impedance log, seismic
section with synthetic insert and the wavelet with its corresponding spectral representations.
where λ is the logging depth interval and λ 0 The coal zones begin just after 1.9 s and are identified by the low acoustic impedance intervals.
is the Backus averaging length. V and V0 then Note the poor tie in the vicinity of the coals where we have a correlation coefficient of 0.59 in
represent the velocity as measured by well the time window 1.7 s to 2.1 s. The synthetic seismogram would require a large stretch around
logging and the Backus averaged well log, the coals to visually correct for the traveltime differences.
respectively. In equation 2, the wavelength is
used instead of frequency, following Liner’s
argument that the wavelength is the more
natural domain of Backus averaging. Equation
2 yields a Q function that describes the local
attenuation behavior. A key consequence
of Liner’s results is that in addition to the
more familiar positive Q, where high frequen-
cies travel faster than low frequencies, local
negative Q is also possible, which implies
that high frequencies travel slower than low
frequencies. In the context of traveltime
differences between logging and seismic
frequencies, this implies that zones of positive
Q require stretching and zones of negative Q
require squeezing of the well logs.

Now consider the two-way traveltime given by

t = 2∫ s(z )dz   (3)


 
where s(z) represents the slowness as a
function of depth, z. As mentioned above, t
is typically computed using well log measure-
Figure 1. Well to seismic tie using a time-depth relationship computed by integrating the well log slowness.
ments and therefore, represents the traveltime
From left to right: Acoustic impedance log, seismic section with synthetic insert and the wavelet with its
associated with the logging frequencies. corresponding spectral representations. Note the poor tie around the coals identified by the low acoustic
However, for our well to seismic ties, we are impedance intervals beginning at approximately 1.9 s.
interested in the traveltime associated with
the seismic frequencies. Therefore, the s(z) Next, we investigate the local Q attenuation concept as presented by Liner (2014) for
used to compute the traveltime function computing a time-depth relationship. Here, our depth sampling interval for the well logs is
should be the Backus averaged slowness. 0.1524 m and we use an averaging length of 90 m, which corresponds to a maximum frequency
of approximately 70 Hz using the relationship, fmax=Vmax/λ min. Figure 2 shows the original
logging and Backus averaged velocities and the corresponding local attenuation function
Example defined as the inverse of Q . The local attenuation curve exhibits both positive and negative
To demonstrate the effects of using a Backus values, and as mentioned above, results in high frequencies (logging) travelling both faster
averaged well log to compute the time-depth and slower than low frequencies (seismic). Consequently, stretching and squeezing is required
relationship, we illustrate with an example for a proper calibration of the time-depth relationship between the well log and seismic. In

Continued on Page 56

DECEMBER 2 014 C SEG RECORDER 55


FOCUS ARTICLE

Continued from Page 55

Figure 2, it can be seen that at the coal intervals (measured depth of


~2700 m), the largest attenuation and hence dispersion is observed.
Consequently, a large stretch is required to properly calibrate the
time-depth relationship.

Figure 3 shows the well to seismic tie using a time-depth relationship


computed by integrating the Backus averaged slowness. The Backus
averaged time-depth relationship was loaded as a pseudo-checkshot
to allow for a direct comparison with the original well log derived
time-depth relationship. The panels from left to right show the acoustic
impedance log, pseudo-checkshot stretch factor, well log velocity and
pseudo-checkshot corrected velocity, seismic section with synthetic
insert and the wavelet with its corresponding spectral representations.
Note the improved tie relative to Figure 1 in the vicinity of the coals
where we now have a correlation coefficient of 0.79 in the time window
1.7 s to 2.1 s. The pseudo-checkshot stretch factor represents the ratio,
V/V0 or equivalently, t0 /t, the traveltime ratio of the seismic to logging
frequencies. At the coal intervals, the stretch factor is largest with a
value of approximately 1.5, which indicates that the seismic frequen-
cies require a traveltime that is 1.5 times longer than that of the logging
frequencies. In terms of the velocities as seen by the logging and seismic
frequencies, this implies a difference on the order of thousands of m/s.

Figure 2. Left: Original logging velocities (blue) and Backus averaged velocities
(red). Right: Local attenuation function, 1/Q.

Figure 3. Well to seismic tie using a time-depth relationship computed by integrating the Backus averaged slowness. From left to right: Acoustic impedance log,
pseudo-checkshot stretch factor, well log velocity and pseudo-checkshot corrected velocity, seismic section with synthetic insert and the wavelet with its corresponding
spectral representations. Note the improved tie relative to Figure 1 around the coals identified by the low acoustic impedance intervals beginning at approximately 1.9 s.

56 C SEG RECORDER DECEMBER 2 014


Conclusions
The question of the validity of stretching and squeezing and how much is accept-
David
able for well to seismic ties have always been a topic of debate. In this study, Cho
we show that in the presence of layer induced long wave elastic attenuation as
presented by Liner (2014), stretching and squeezing is mandatory as both positive
and negative local Q values are possible. By using a Backus averaged well log •• holds a B.Sc. in Physics and a Ph.D. in
Geophysics from the University of Calgary. His
to compute the time-depth relationship, an improved well to seismic tie can be
work experience includes various positions at
achieved. Furthermore, the ratio V/V0 provides an indication of how much stretching
and squeezing is acceptable. both operators and service companies, where
he specialized in quantitative interpretation
The method as presented provides a first pass at adjusting the time-depth relation- studies. He is currently General Manager and
ship to account for dispersion as discussed. However, further adjustments will
Lead Geophysicist at Qeye Labs Canada. His
typically be required as the chosen Backus averaging length only provides the
technical interests include anisotropic elasticity,
upper bound for the seismic frequencies and does not account for dispersion
within the seismic bandwidth. Therefore, this provides an approximation for the fracture characterization, rock physics, time
effective velocity as seen by the seismic waves. Furthermore, logging errors can still series analysis and inverse theory.
be present that will propagate through the time-depth calculations. Therefore, a
manual stretch and squeeze is still required to fine-tune the time-depth relationship
but should be guided by the above analysis to obtain a reasonable result.

Acknowledgements
We thank Sitka Exploration and Qeye Labs for supporting this work and David
WesternGeco for permission to show the data. Andrew Graham, Greg Cameron
and Marcello Orizzonte are acknowledged for helpful discussions. Nordin
•• started his geophysical career at Amoco in 1985
and moved to Crestar when it was spun off in
References 1991. After 6 years as an interpreter at Crestar,
Backus, G., 1962, Long-wave elastic anisotropy produced by horizontal layering: Journal of
Geophysical Research, 67, no. 11, 4427-4440.
he joined the management group, where he

Coulombe, C. A., and Bird, D. N., 1996, Transmission filtering by high-amplitude reflection
stayed until Crestar was bought by Gulf in the
coefficients: Theory, practice, and processing considerations: The Leading Edge, 15, no. 9, 1037-1042. fall of 2000. Through an interesting and involved
Gretener, P. E. F., 1961, An analysis of the observed time discrepancies between continuous and process, he decided to step sideways and start
conventional well velocity surveys: Geophysics, 26, no. 1, 1-11.
a search business where he spent 5 years in
Kjartansson, E., 1979, Constant Q-wave propagation and attenuation: Journal of Geophysical
Research: Solid Earth, 84, no. B9, 4737-4748. “working bliss”, spending his days building
Liner, C. L., 2014, Long-wave elastic attenuation produced by horizontal layering: The Leading relationships with employers and candidates
Edge, 33, no. 6, 634-638.
throughout the oil and gas business.

When the challenge went out of search, he


decided to return to the dream of being the
sole geophysicist in a small company. He
started consulting in 2006 and eventually joined
one of his clients, Skana Exploration, as their
geophysicist. Skana sold in Oct 2011 and he
is now reprising the role with the successor
company, Sitka Exploration, drilling horizontals
in the Mannville W5.

DECEMBER 2 014 C SEG RECORDER 57

Potrebbero piacerti anche