1

C-13 ZERO-PHASING SEISMIC DATA WITHOUT

WELLS – REVISITED
1 2
MO SAFAR and NORBERT VAN DE COEVERING
1
14 Timline Green, Bracknell RG12 2QP, UK
2
CGG, Vantage West, Great West Road, Brentford TW8 9GG, UK

Summary

A new method is developed for estimating the seismic wavelet from seismic data acquired in offshore
areas without wells. Our method is essentially an improvement of the one proposed by Foster et al
(Geophysics, 1997), which involved combining two independent methods in order to increase the
reliability of the estimated wavelet.
One important requirement of the proposed method is that the near field signatures generated by the
air guns forming the source array are recorded during the survey so that the depths and the geometry
of the air guns, the reflection coefficient of the sea surface and the far field signature of the seismic
source can be determined (Safar, 1998, 1999). Another requirement is a Q value for modeling the
earth’s absorption, which is determined from the seismic data.
The main feature of the proposed method is that we do not need the assumption that the seismic
wavelet is minimum phase. The method is tested using 3D migrated data acquired around a logged
well in the North Sea. The accuracy of the proposed method is evaluated by comparing the estimated
seismic wavelet with those derived from the statistical and well to seismic matching methods.

Introduction

Recently Foster et al (1997) proposed a method for estimating the seismic wavelet for the purpose of
zero-phasing seismic data without wells. It consists of a combination of two independent methods for
deriving the seismic wavelet, namely the physical modeling and the statistical extractions. The main
objective is to increase the reliability of the estimated seismic wavelet. This is because one method
should compensate for the shortcomings of the other method.
The new methodology proposed in this paper differs from that discussed by Foster et al (1997): the
physical modeling is combined with only one part of the statistical method, which does not require the
assumption that the seismic wavelet is minimum phase. Another important difference is that the far
field signature, which is crucial for the physical modeling method, is determined form the near field
signatures recorded during the survey.

Acquisition and Processing

In 1995 a 3D survey of UKCS Quad 15 was carried out for Enterprise Oil by Geco-Prakla using the
Geco Gamma vessel, which was fitted with the Trisor source signature estimator (SSE) system. The
Trisor SSE system is capable of calculating the far field signature for each shot from the recorded
pressure field in the vicinity of each element forming the seismic source array. This array was dual air-
gun array – Flip/Flop (Port array and Starboard array). Each array consisted of two non-identical sub
arrays with each sub array having six elements. The element nearest to the vessel consisted of 3
identical air-gun clusters whereas the remaining five elements were single air guns.

EAGE 64th Conference & Exhibition — Florence, Italy, 27 - 30 May 2002

 2

The 3D survey was acquired using six 3000m streamers separated by 100m crossline with each
streamer having 240 groups with separation of 12.5m between two adjacent groups. The six
streamers were placed at a depth of 6.5 m.
The initial part of the processing sequence, which was carried out on board the Geco Gamma vessel,
consisted of resampling the data from 2 ms to 4ms after applying a zero phase anti-alias filter (5 Hz/18
dB – 90 Hz/72 dB). Subsequently the following steps were applied: a spatial anti-alias filter and trace
drop; source designature to minimum phase using a “library” far field signature, scaling T to the power
1.9, 2 zone predictive deconvolution before stack 20/300ms and 24/300ms.

Seismic Wavelet Estimation

The estimation of the seismic wavelet is essentially obtained using the physical modeling method,
which requires a far field signature of the seismic source and a Q value for modeling the earth’s
absorption.
The determination of the far field signature requires knowledge of the depths of the air-guns, the air-
gun-hydrophone geometry and the reflection coefficient of the sea surface, which are determined
(Safar, 1999) from the recorded near field signatures when the air-guns were fired individually. The
effective depths of the port and starboard arrays, obtained from the amplitude spectra of the far field
signatures synthesized from the measured near field signatures using the depths of the air-guns, were
6.5m and 6.25m respectively. However, the depth of the arrays specified in the contract was 5m.
Fortunately, the “library” for field signature used for source designature to minimum phase was
obtained at a depth of 6m instead of 5m.
The approach used by Foster et al (1997) for determining the average value of Q consisted of
matching the seismic wavelet obtained by physical modeling to the seismic wavelet derived by the
statistical method, which assumes that the seismic wavelet is minimum phase. In fact, a much better
approach for determining the Q value is to match the high frequency part of the amplitude spectrum of
the seismic wavelet derived by physical modeling to that of the statistically derived seismic wavelet
and consequently avoid the minimum phase assumption. We use the same approach used by Foster
et al (1997) for deriving the amplitude spectrum of the statistically derived seismic wavelet from the
seismic data. This consisted of using the multiple coherency function (White, 1972, 1973 and personal
communication) for separating noise from coherent signal and then smoothing for removing any
geological information. Fig 1 shows the statistically derived average signal spectrum using 20 x 20
traces around a well over the time windows of 1000 – 1500ms and 1700 – 2100 ms. It can be seen
from Fig.1 that the signal spectrum for the time window 1700 – 2100 ms, which is the target zone, has
a peak at a frequency of 14Hz indicating a very high earth absorption. A Q value of 40 is obtained
when matching the high frequency part of the signal spectrum shown in Fig.1 for the time window of
1700 – 2100 ms to that of the seismic wavelet derived by the physical modeling method.

Discussion

In order to check the accuracy of the estimation of the seismic wavelet using the proposed method,
the seismic wavelet is also derived using the statistical and the well to seismic matching (White, 1980,
Walden and White, 1984 and see also Van de Coevering, 2000 for overview) methods.
The estimation of the seismic wavelet using well to seismic matching essentially involves the
determination of the impulse response of the transfer function of the matching operator, which is
required so that the broadband synthetic seismogram and a segment of the processed seismic data
are matched. Unlike the statistical method, the well to seismic matching does not require the
assumptions that the seismic wavelet is minimum phase and the reflection sequence is white. This
means that the polarity and the onset time of the estimated seismic wavelet are the same as those of
the seismic wavelet in the seismic trace segment. Fig.2 shows three seismic wavelets derived by the
three methods. It can be seen from Fig.2 that the seismic wavelet estimated by the statistical method
differs significantly form those derived by the new method and the well to seismic matching method
(White, 1980 and White and Walden, 1984). The important difference is that about 20 ms of the front
of the statistically derived seismic wavelet is missing. This may explain the reason for the mistie, which
is usually found between the synthetic seismogram, generated using the statistically estimated seismic
wavelet, and the seismic data. It is worth pointing out that the missing part of the statistically derived
seismic wavelet was also observed by Porsani and Ursin (TLE, 2000) when comparing the statistically
 3

derived seismic wavelet with that derived using mixed phase statistical method (Geophysics, 1998 and
Geophysical Prospecting 1999).
A comparison between the seismic wavelet estimated by our proposed method and that estimated by
the well to seismic matching method displayed in Fig.2, shows that the early part of the well to seismic
matching derived seismic wavelet is in fact processing noise (the early part is that part which occurs
before the first trough). This is because the first peak of the seismic wavelet is not physically plausible.
The onset times and the shapes of the first two lobes of the estimated seismic wavelets agree fairly
well. However, there is a significant difference between the tails of the estimated seismic wavelets. A
possible explanation for this difference is that less attention is paid (White, 1980) in estimating the
seismic wavelet’s tail than that paid in estimating the main lobes. It can be deduced from Fig.2 that the
seismic wavelet is not minimum phase. This means that the attempt during processing to convert the
seismic wavelet to a minimum phase wavelet was unsuccessful. On the other hand, the new proposed
method for estimating the seismic wavelet makes not only the attempt to convert it into minimum
phase but also the current idiotic specifications for the source performance unnecessary.

Conclusion

The new method developed in this paper for the purpose of zero-phasing seismic data acquired in
areas without wells is believed to be more reliable in reducing uncertainty in the seismic wavelet
estimation than that proposed by Foster et al (Geophysics 1997). This is because the statistical
method, which is an essential part of their proposed method, is shown to be unreliable.

Acknowledgements

The authors wish to thank Enterprise for permission to use their dataset. They also thank R. White and
K. Nunn for their comments. N. Van de Coevering would like to thank Peter Harris and Henning
Hoeber for developing all the necessary methods and tools applied to the real data example, and in
particular Henning Hoeber for all his valuable comments. Also thanks to CGG for the possibility of
publishing this paper.

References

Foster M.S., Nunn K.R., Lewis S.A. and Reynolds D.J., 1997, Zero-phasing seismic data without wells
in offshore West Africa: reducing uncertainty and variability of the wavelet: Geophysics, 62, 352-
361.
Porsani M.J. and Ursin B., 2000, Mixed-phase deconvolution and wavelet estimation: The Leading
Edge, 19, 76-79.
Porsani M.J. and Ursin B., 1998, Mixed-phase deconvolution: Geophysics, 63, 637-647.
Porsani M.J. and Ursin B., 2000, Estimation of an optimal mixed-phase inverse filter: Geophysical
Prospecting, 4, 663-676.
Safar M., 1998, A new method of marine seismic source signature estimation from near field
measurements: 60th EAGE Conference Leipzig, Extended Abstracts, Paper 2-37.
Safar M., 1999, On the improvement and quality control of the Trisor source signature estimator: 61st
EAGE Conference Helsinki, Extended Abstracts, Paper 6-10.
Van de Coevering N.M.L., 2000, Wavelet estimation and zero-phasing by integrating seismic and well-
based methods, 62nd EAGE Conference Glasgow, Extended Abstracts, Poster P-123.
Walden A.T. and White R.E., 1984, On errors of fit and accuracy in matching synthetic seismograms
and seismic traces: Geophysical Prospecting, 32, 871-891.
White R.E., 1972, Determination of a minimum phase wavelet from its energy spectrum: BP Report
EPR / R1173.
White R.E., 1973, The estimation of signal spectra and related quantities by means of the multiple
coherence function: Geophysical Prospecting, 21, 660-703.
White R.E., 1980, Partial coherence matching of synthetic seismograms with seismic traces:
Geophysical Prospecting, 28, 333-358.

EAGE 64th Conference & Exhibition — Florence, Italy, 27 - 30 May 2002

 4

dB

frequency (Hz)

Fig 1: Average signal spectrum of 20 by 20 seismic traces around the well location for a time window
of 100-1500 ms (black) and 1700-2100 ms (red).

time (ms)

Fig 2: Comparison of the wavelets derived using different methods: (TOP) the matching operator,
(MIDDLE) the modeled wavelet and (BOTTOM) the autocorrelation wavelet.