Professional Documents
Culture Documents
C a r b SECTION:
o n a t e C
r aersb e
ona artce h
r e si ena r C
c hh iinn C
ah i n a
Figure 1. A section from area TZ45 in Tarim Basin. The spectra on the right are extracted from the three rectangular areas (from shallow to
deep). The event selected by the deepest rectangle is the top of carbonate which is the exploration target. If we choose an eighth of the wavelength as
the resolution, it is easy to calculate that the resolution is 9.45 m at 800 ms and 18.5 m at 2500 ms. At 4000 ms, where we are most interested,
we can only recognize a layer with a thickness of 41.65 m.
. (11)
Figure 3. Comparison of wavelets extracted from traditional Kirchhoff migration gather (a),(b),(c) and Q-migration gather (d), (e), (f ) at
depths of 2000 ms (a), (d), 3500 ms (b), (e) and 5000 ms (c), (f ). The wavelets of Q migration are “thinner” than those of traditional Kirchhoff
migration. That means increased resolution. And the waveforms, magnitudes and phases are stable from shallow to deep in the Q migration,
because it can recover the attenuation.
Figure 4. (a) The spectrum of the reflectivity series. (b)The reflectivity. Figure 5. (a) The reflectivity series. (b) The unattenuated synthetic
(c) The unattenuated synthetic trace. (d) The result of traditional trace. (c) The result of the traditional nonstationary deconvolution.
nonstationary deconvolution. (e) The result of the improved (d) The result of the improved deconvolution. (e) Comparison of the
deconvolution. (f ) The comparison of the wavelets extracted by the wavelets extracted by the traditional method and the new method. (f )
traditional method and the new method at 0.2 s. (g) The comparison Comparison of the correlation curves of the wavelets extracted by the
of the correlation curves of the wavelets extracted by the two methods. two methods.
trace, so we could obtain 51 different wavelets. By comparing ure 4c) is attenuated with a constant Q filter (Q = 100). This
these 51 wavelets and the model wavelet, we form a correlation represents the input trace (Figure 5b) before deconvolution.
curve. The higher degree of correlation between the obtained The number of the wavelets is the same as the above example.
wavelet and model wavelet, the more accurate the extracted The difference between the two methods of deconvolution can
wavelet. Figure 4e displays curves from the two methods, and be observed by comparing Figure 5c (the traditional method)
the correlation of the new method is higher, Figure 4d, a com- with Figure 5d (the new method). The resolution of the decon-
parison of the two wavelets at 0.2 s obtained by two methods, volution obtained by the new method is higher. Figure 5e, a
shows that wavelet obtained by the new method is closer to the comparison of the two wavelets at 0.2 s obtained by two meth-
model wavelet. The difference between the deconvolution re- ods, illustrates that the wavelet obtained by the new method
sults of the two methods can be observed by comparing Figure is closer to the attenuated model wavelet. Figure 5f displays
4f (the traditional method) with Figure 4g (the new method). curves resulting from the two methods, and correlation with
The resolution of the deconvolution result obtained by the new the result of the new method is higher.
method is higher. Figure 6a is a poststack seismic profile. The seismic profiles
In order to compare the effect of the two methods when processed by the traditional method (Figure 6b) and the new
applied to an attenuated trace, the synthetic seismic trace (Fig- method (Figure 6c), both identify large features but the meth-
February 2012 The Leading Edge 209
Downloaded 15 Feb 2012 to 41.224.248.253. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
C a r b o n a t e r e s e a r c h i n C h i n a
Figure 6. Comparison of the two methods when applied to real data. (a) Raw data. (b) Result of the traditional method. (c) Result of the
improved method.
od proposed here can identify the fault shown in the red circle, Margrave, G. F., 1998, Theory of nonstationary linear filtering in the
and the traditional method cannot. Furthermore, the resolu- Fourier domain with application to time-variant filtering: Geophys-
tion of the profile obtained by the proposed method is higher ics, 63, no. 1, 244–259, doi:10.1190/1.1444318.
than that of the other two. Margrave, G. F. and M. P. Lamoureux, 2002, Gabor deconvolution:
CREWES Annual Research Report, 13.
Conclusions Montana, C. A. and G. F. Margrave, 2005, Color correction in Gabor
deconvolution: CREWES research report, 17.
1) Q migration is based on Kirchhoff migration, and can com- Traynin, P., J. Liu, and J. M. Reilly, 2008, Amplitude and bandwidth
pensate for attenuation along every migration raypath. So recovery beneath gas zones using Kirchhoff prestack depth Q mi-
this method can exactly recover amplitude, phase and fre- gration: 76th Annual International Meeting, SEG, Expanded Ab-
quency, and widen bandwidth to eventually enhance seis- stracts, 2412–2416.
mic resolution. Wu, S., S. Z. Sun, and Z. Wang, 2011, Improved nonstationary decon-
2) The improved nonstationary decovolution is based on a volution: 73rd EAGE Conference & Exhibition incorporating SPE
joint time-frequency analysis, and works on non-white re- EUROPEC.
flectivity. It enhances seismic resolution by compensating Yu, Y., R. Lu, and M. Deal, 2002, Compensation for the effects of shal-
for amplitude and frequency loss without a Q model. low gas attenuation with viscoacoustic wave-equation migration:
72nd Annual International Meeting, SEG, Expanded Abstracts,
3) Q migration considers both amplitude and phase, and ac-
2062–2065.
curately preserves amplitudes. The improved nonstationary
deconvolution assumes a minimum-phase wavelet, but it
Acknowledgments: We thank Laboratory for Integration of Geology
works without a Q model. So it is robust.
and Geophysics, China University of Petroleum for the authoriza-
tion to present this work. We also appreciate that Exploration and
References
Development Research Institution in Tarim Oil Field provided the
Cheng, P. and G. F. Margrave, 2009, Color correction for Gabor de-
convolution and nonstationary phase rotation: 2009 CSPG CSEG data used in this research. We are grateful to Yu Zhang, Wenbo Sun,
CWLS Convention, 396–399. Lifeng Liu, and Lulu Cai for helpful discussions.
Keers, H., D. W. Vasco, and L. R. Johnson, 2001, Viscoacoustic cross-
well imaging using asymptotic waveforms: Geophysics, 66, no. 3, Corresponding author: szd@cup.edu.cn
861–870, doi:10.1190/1.1444975.