Sie sind auf Seite 1von 13

Geophysical Prospecting, 2011, 59, 464476

doi: 10.1111/j.1365-2478.2010.00933.x

Seismic interferometry experiment in a shallow cased borehole


using a seismic vibrator source
Flavio Poletto , Lorenzo Petronio, Biancamaria Farina and Andrea Schleifer
Istituto Nazionale di Oceanografia e di Geofisica Sperimentale, OGS, Borgo Grotta Gigante n. 42/c, 34010 Sgonico (TS), Italy

Received February 2009, revision accepted October 2010

ABSTRACT
We present the results of a seismic interferometry experiment in a shallow cased
borehole. The experiment is an initial study for subsequent borehole seismic surveys
in an instrumented well site, where we plan to test other surface/borehole seismic
techniques. The purpose of this application is to improve the knowledge of the reflectivity sequence and to verify the potential of the seismic interferometry approach to
retrieve high-frequency signals in the single well geometry, overcoming the loss and
attenuation effects introduced by the overburden. We used a walkaway vertical seismic profile (VSP) geometry with a seismic vibrator to generate polarized vertical and
horizontal components along a surface seismic line and an array of 3C geophones
cemented outside the casing. The recorded traces are processed to obtain virtual
sources in the borehole and to simulate single-well gathers with a variable sourcereceiver offset in the vertical array. We compare the results obtained by processing
the field data with synthetic signals calculated by numerical simulation and analyse
the signal bandwidth and amplitude versus offset to evaluate near-field effects in the
virtual signals. The application provides direct and reflected signals with improved
bandwidth after vibrator signal deconvolution. Clear reflections are detected in the
virtual seismic sections in agreement with the geology and other surface and borehole
seismic data recorded with conventional seismic exploration techniques.
Key words: Overburden, Seismic interferometry, Vibroseis.

INTRODUCTION
Seismic interferometry uses the cross-correlation of recorded
traces to obtain virtual sources at the position of the receivers (Claerbout 1968; Bakulin and Calvert 2004; Calvert
2004). The method allows exploration geophysicists to simulate source points where the possibility to use real sources
is limited, as in the case of borehole geophysics. Several examples of interferometry for seismic exploration are shown
in the literature with vertical seismic profiling (VSP) data

This

paper is based on extended abstract P278 presented at the 70th


EAGE Conference & Exhibition Incorporating SPE EUROPEC 2008,
912 June 2008 in Rome, Italy.
E-mail: fpoletto@ogs.trieste.it

464

sets, where the seismic virtual source method provides successful results for the detection and separation of wavefields
(e.g., Bakulin et al. 2007; Mateeva et al. 2007). In these applications, important aspects are the available coverage as a
function of the distribution of the real exploration sources
illuminating the receivers and the source spacing required to
minimize and prevent spatial aliasing (Mehta et al. 2008).
We apply seismic interferometry to process the data recorded
by an array of 3C receivers fixed in a borehole of an instrumented test site facility. This survey is an introductory study to
evaluate borehole signals in the near-surface and to provide
reference signals in the well, which can be used to perform
acquisitions with other conventional techniques and, potentially, to experiment with borehole instrumentations. We use
a seismic vibrator at the surface as the exploration source and


C

2010 European Association of Geoscientists & Engineers

Seismic interferometry experiment in a shallow cased borehole 465

simulate the virtual sources at any recording depth, with the


aim to obtain high-resolution data. Albeit this study is substantially an application of the seismic interferometric method
to a multi-component walkaway VSP experiment, we focus
our attention on the typical geometry of single well imaging
(Hornby 1989; Chabot et al. 2001) in this work. The target of the single well imaging sonic techniques is to image
the near-borehole structure using full-waveform sonic data
(Hornby 1989) with a signal frequency of several thousand
Hz. In this experiment the signal bandwidth is confined to a
lower frequency range, so that the result is a low-frequency
approximation of sonic single-well-imaging wavefields. The
effort in the processing phase is to improve the seismic frequency content, exploiting the potential of the interferometry
method and to recover the high-frequency signal of the surface vibrator source between closely-spaced receivers. Virtual
single-well sections are obtained by gathering the traces with
virtual sources and receivers at very close positions. These signals are interpreted and can be considered as reference results
representative of single-well sections in the site area, in the
low-frequency approximation with respect to sonic logs and
with higher frequency content with respect to conventional
seismic.

ACQUISITION
In this work we use the seismic data specifically recorded for
the purposes of interferometry at the Osservatore di Geofisica
Sperimentale (OGS) test site (Poletto et al. 2008). This site
is located in the Toppo inter-mountain plain (north-eastern
Italy) on the external thrust-belt of the eastern Southern Alps
(Zanferrari, Poli and Rogledi 2002) where Jurassic carbon-

ate overlays Miocene sedimentary sequences. In the studied


area, quaternary alluvial sediments, mainly gravels, overly the
Miocene conglomerate (Montello conglomerate) formation.
At this geophysical site, three closely spaced wells were drilled
to depths of 280 m, 380 m and 420 m below ground level.
Near-surface overburden conditions, which affect seismic surveys in this area, are due to the presence of loose gravel in the
shallower part in conjunction with a deep water table (located
at a depth of approximately 120 m below ground level).
The field layout of this experiment (Fig. 1) consists of an instrumented well with 30 3C geophones cemented outside the
casing and a surface seismic source line crossing the well. Borehole sensors (borehole geophones 15 Hz natural frequency)
were installed between 35240 m below ground level with a
depth interval of 10 m in the shallower section (35145 m
depth) and 5 m in the deeper one (>145 m depth). A 2500pound seismic vibrator (minivib IVI) energized three series of
80 shot points polarized in the P- (vertical), SV- (horizontal
in-line) and SH-wave (horizontal cross-line) configurations.
The maximum horizontal offset from the wellhead was 150
m. The source intervals were 2.5 m below and 5 m above
50 m offset. These small source-sampling intervals were chosen to avoid aliasing effects and to improve the S/N ratio
(van Manen, Curtis and Robertsson 2006; Mehta et al. 2008).
We used a linear upsweep of 12 s duration, the theoretical
sweep ranging from 10400 Hz, with 5 sweeps for each source
position. Raw data consisting of borehole geophone data and
vibroseis pilot signals were recorded with a 1 ms sampling
rate. The pilot signal traces were the baseplate acceleration,
the reaction-mass acceleration, the ground force signal obtained by the weighted sum of the baseplate and reactionmass signals (Sallas 1984) and the theoretical sweep. After

Figure 1 Schematic layout of the single well interferometry test (not to scale).


C

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

466 F. Poletto et al.

Figure 2 Real vibroseis signals. P-wave energizations: pilot signal


(ground force) power spectra. Surface terrain and ground coupling
conditions vary along the line and affect the sweep-signal bandwidth.

in-field quality control on the pilot signals, the field data were
cross-correlated with the ground force signal obtained by the
weighted sum and stacked for each source position. However,
all the recorded raw field signals, consisting of uncorrelated
geophone and pilot traces, were available for subsequent reprocessing.
Figure 2 shows the power spectra of the groundforce pilot
signals recorded in the P-wave energizations. These spectra are
obtained by Fourier frequency transforming the stacked autocorrelations of the groundforce pilot signals for each source
position. The reference signal variability along the energization line depends on the different vibroseis ground-coupling
conditions, related to the presence of different near-surface
terrains.
Figure 3 shows example vibroseis cross-correlated signals
of P-, SV- and SH-wave energizations recorded by 3C downhole geophones. The data have a good S/N ratio. Tube waves
do not significantly affect this VSP data set, so no particular attention was given in the processing phase to avoid the
contribution of the tube-wave noise induced by near-offset
sources (Gaiser, Vasconcelos and Ramkhelawan 2009). The
weak amplitude of the borehole tube waves is interpreted as
due to the location of the borehole geophones outside the
casing.

THE PROCESSING METHOD


The processing method adopted in this work for the synthesis
of the virtual signals uses the standard interferometry by crosscorrelation algorithm


S Ai ()S Bi
(),
(1)
C AB () =
i


C

where S Ai and S Bi are the Fourier frequency transforms of the


preprocessed vibrator signals from the i-th field source position recorded by two receivers in A and B, respectively,
is the angular frequency and denotes complex conjugate.
The summation in the interferometry-by-correlation equation
(1) is extended over the offset domain (or P, SV and SH subdomains) of the surface vibrator sources (here we omit for
simplicity the multi-component tensorial notation). The signal C AB is a band-limited estimate of the Greens function
(Wapenaar and Fokkema 2006) between two receiver points
B and A. Even if interferometry by correlation can also be applied using the raw vibrator data (Halliday, Curtis and Kragh
2008), in this work we prefer using preprocessed data as input
to equation (1). This approach makes it possible to interpret
and analyse the signal wavefields in the seismograms preprocessed prior to interferometry (a similar analysis was done for
drill-bit signals by Poletto, Corubolo and Comelli (2010)). We
used two approaches to preprocess the vibrator data.

Interferometry using vibroseis correlation data


In the first approach, we used as input to equation (1) the
vibrator signals S Ai obtained using the conventional pilotcross-correlation approach
S Ai () = XAi ()Pi (),

(2)

where Pi is the groundforce pilot-sweep signal of the i-th


source point and XAi is the signal recorded in A (similar
reasoning holds for S Bi ). Here we omit for simplicity the
shot stacking index. Using equation (2), the interferometryby-cross correlation equation (1) can be rewritten as
C AB () =

XAi ()XBi ()Pi ()Pi ().

(3)

Later in this paper we show that this approach provides good


quality virtual seismic signals in the lower frequency bandwidth, say with maximum frequency up to 100140 Hz, for
the data set of this experiment. Assuming that the signal radiated from the i-th source point and recorded in A is the composition of the Greens function GAi from the source point i to
the receiver point A and of the true groundforce signal Vi of
the vibrator (Sallas 1984; Baeten and Ziolkowski 1990), we
have (neglecting the response of the recording instrumentation
and omitting tensorial notation for multi-components)
XAi () GAi ()Vi (),

(4)

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

Seismic interferometry experiment in a shallow cased borehole 467

Figure 3 Vibroseis cross-correlated signals. P-, SV- and SH-wave energizations recorded by 3C downhole geophones. A notch filter (50 Hz)
is applied to the data to remove AC power noise. Z, R and T represent the vertical, horizontal in-line and horizontal cross-line components,
respectively.

where the symbol means equal apart from a global scaling


factor (a similar equation holds for B). Equation (3) becomes
C AB ()

GAi ()GBi ()Vi ()Vi ()Pi ()Pi ().

(5)

In the representation by equation (5) we neglect the anti-causal


term C AB (Wapenaar and Fokkema 2006), not relevant for
the purposes of our analysis. Assuming that P() is a reliable approximation of V(), we obtain that equation (5)
ultimately contains a fourth power of the vibroseis source signal. Even if the theoretical source sweep was designed using a


C

theoretical linear sweep in the frequency bandwidth ranging


between 10400 Hz, the effective pilot signals were recorded
with a significant amplitude attenuation for frequencies higher
than 100140 Hz. Moreover the weighted groundforce pilot
signal contains distortions due to baseplate bending vibrations (Baeten and Ziolkowski 1990) and the source signal
contains frequency peaks due to amplification by the local
soil-vibrator system response. During acquisition a significant
effort was made to minimize these effects and to optimize the
spectrum of the emitted signal. However, this task was only
in part achieved by expending additional time to optimize the

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

468 F. Poletto et al.

in equation (8) improves the S/N in the virtual-signal results


obtained by using the vibroseis deconvolved signals. The approach is robust with respect to the phase fluctuations and
distortions after deconvolution as the averaged signature is

adaptation of the baseplate to the soil surface at each vibration


location. Figure 2 shows the power spectra of the groundforce
pilot signals acquired at different positions. These pilot-signal
power spectra contain undesired coloured effects, which are
amplified by the power (in this case second, according to equation (5)) and ultimately limit the bandwidth in the recovered
Greens function synthesized using the vibroseis correlation
data.

where

Interferometry using vibroseis deconvolution data

Qi () =

As interferometry provides a local response and preserves


high frequencies, we tested the method by using the vibroseis
deconvolution input traces, aimed at improving bandwidth
and at compensating for local effects at each source location
(Poletto et al. 2008; for a discussion on the role of energy
equipartitioning for interferometry see Snieder, Wapenaar and
Wegler 2007). In this approach, we use as input to equation
(1) the vibroseis signals deconvolved by the pilot signals (Brittle, Lines and Dey 2001)
D
() =
S Ai

XAi ()
,
Pi ()

(6)

where using some additional white noise to bias the pilot signal P() before the spectral division is beneficial. Using equation (6), equation (1) can be rewritten as
D
()
C AB

D
D
S Ai
()(S Bi
())

 XAi ()X ()
Bi
.
=

P
()P
()
i
i
i

(7)

Equation (7) shows that interferometry by the crosscorrelation of the vibroseis deconvolution signals is similar to
the conventional interferometry-by-deconvolution approach
(Vasconcelos and Snieder 2008a,b); with the difference that
equation (7) removes only the reference source signature and
does not remove propagation effects. Moreover, using equation (4) in equation (7) gives
D
C AB
()


i

GAi ()GBi ()

Vi ()Vi ()
.
Pi ()Pi ()

(8)

The key problem in vibroseis deconvolution is that the


weighted groundforce estimate deviates from the true signal
emitted in the formation at the baseplate (Sallas 1984; Baeten
and Ziolkowsky 1990; Mewhort, Bezdan and Jones 2002).
This leads to frequency distortions, which make vibroseis deconvolution difficult to be applied to the individual shots of
our experiment. However, we may observe in Fig. 2 that the
pilot power spectra are variable along the acquisition line.
For this reason the deconvolution response is different for
different source positions. Averaging the deconvolved terms


C

W () =

Qi ,

(9)

Vi ()Vi ()
.
Pi ()Pi ()

(10)

The deconvolved signature W() is a zero-phase signal for


construction. Other possible approaches to removing the
source-signature effects are to deconvolve the interferometryby-correlation result C AB given by equation (3) by the
averaged-energy deconvolution operator
DAV () = 
i

1
2 ,
Pi ()Pi ()

(11)

or, also, by using the fourth power of averaged synchronized


sweep signals | Pi ()|. If we assume that the source signature and the propagation effects are independent, i.e., assuming statistical independence for Qi and (GAi GBi ), using
equations (9) and (10) we can rewrite equation (8) as
D
() W ()
C AB

GAi ()GBi ().

(12)

The source-deconvolution approach in equations (8) and (12)


is similar to the semblance technique for performing optimal,
least-squares deconvolution of VSP data, in which the operator is estimated using moveout aligned traces (Haldorsen,
Miller and Walsh 1994). However, the interferometry vibroseis deconvolution approach using equations (8) and (12) is
rephased for construction and does not require estimating the
source signature from the seismic data.
Conventional interferometry-by-deconvolution methods
are also tested with the correlated vibrator data of equation (3), to remove the source signature and to improve the virtual signal signature. In the following we present only the results obtained with the standard approach of interferometryby-correlation (equation (1)) with the vibrator correlated
(equation (2)) and vibrator deconvolved (equation (6)) data
sets, which provided good-quality results for subsequent signal analysis. This analysis includes the processing of traces
of different components and data gathering to simulate the
geometry of a survey with a source/receiver downhole tool in
the well.

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

Seismic interferometry experiment in a shallow cased borehole 469

BOREHOLE DATA ANALYSIS


The real interferometry data are compared to the synthetic
data generated by a 2D elastic finite difference code, with a
regular spatial grid interval x = z = 1 m, representing the
geological model of the test site. These synthetic data were
computed to guide the interpretation of the redatumed wavefields. The numerical simulation of the signals recorded along
the casing does not include the borehole.
Figure 4 shows the synthetic data, P- and SV-waves, obtained with the vertical and the horizontal sources, respectively, located in the borehole at elevation 132 m. The amplitude versus-offset decay in the frequency bandwidth 60
140 Hz is shown in Fig. 5. The real P and SV interferometry data computed at the same source position (virtual source at 132 m elevation above sea level)
are shown in Fig. 6. The analysis of the amplitudes
in the field interferometry traces is performed by selecting the compressional and shear direct arrivals at different offsets of the deeper receivers from the virtual source
(Fig. 7). These results are obtained with all the real sources
(2.5 m and 5 m interval) and with an equi-spaced (regular)
subset of them (5 m). The real-signal amplitude results (Fig. 7)
are compared to the amplitudes of the synthetic data (Fig. 5).
The amplitudes of the data obtained with equi-spaced sources
are more in agreement with the trends showed by the synthetic data, where near-offset effects (in the 2D model approximation) are included. This result is in agreement with
the theory, for which the retrieval of the Greens function
by cross-correlation depends on the appropriate spatial distri-

Figure 5 Synthetic signals. P and SV direct signal amplitudes at average frequency 100 Hz versus source-receiver offset in the well. The
amplitude is relative to the amplitude of the signal at the source location.

bution of the real sources in terms of energy equipartioning


(Snieder et al. 2007).
In the single well imaging technique, a critical point is the
large amplitude (including near-offset effects) of the borehole
waves with respect to the magnitude of the investigated reflections. In this case, the data are not strongly contaminated by
tube waves probably because the sensors are installed outside
the cased borehole.
Due to source illumination conditions from the surface with
the stationary-phase region at the well head, the coverage is

Figure 4 Synthetic results obtained with the source at 132 m elevation in the well: a) P-waves and b) SV-waves.


C

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

470 F. Poletto et al.

Figure 6 Real vibrator interferometry results obtained with the virtual source at 132 m elevation in the well: a) P-waves and b) SV-waves.

Figure 7 Real-vibrator virtual signals. Comparison of direct signal


amplitudes at average frequency 100 Hz versus source-receiver offset
in the well. The solid line is obtained by using a regular vibrator
interval and the dashed line is obtained by using all the source points.
The amplitude is relative to the amplitude of the signal at the source
location.

appropriate for traces located below the reference receiver in


which the virtual source is simulated. The signal arrivals in
the borehole above the source are represented by the reciprocal virtual signals, observable in Figs 6 and 8 at negative
times at shallower receiver positions with respect to the reference trace used as the virtual source. Here reciprocal means
that if we interchange the order of two receiver traces in the


C

cross-correlation, i.e., interchange the reference trace in equation (1), we obtain the time reversed result (see, for comparison, the synthetic signals of Fig. 4).
The comparison between interferometric virtual-shot data
obtained by processing cross-correlated vibroseis data and deconvolved vibroseis data is shown in Fig. 8 for P-wave gathers.
The sweep-based deconvolution improves the bandwidth of
the signal produced by vibrators and the interferometry crosscorrelation and stacking improves the signal-to-noise ratio in
the virtual signals by attenuating the deconvolution noise and
distortions. The sweep-correlation and sweep-deconvolution
virtual results are compared in Fig. 8(a,b), respectively. The
average improvement in the spectrum of the interferometry by
correlation using the vibroseis deconvolution signal beyond
the observed sweep bandwidth (however, within the nominal
sweep bandwidth) can be appreciated in Fig. 9 (Poletto et al.
2008). Similar results of improving the signal bandwidth were
shown by Haldorsen and Borland (2008) for walkaway VSP
using the downgoing energy to estimate the deconvolution
operator.

SINGLE-WELL RESULTS
The interferometry processing is repeated for all the receivers
to obtain a virtual source for each sensor of the borehole array.
Subsequently, the data are selected and gathered in single-well
sections.
The single-well interferometry signals are gathered by Pand SV-components by extracting the traces vertical and

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

Seismic interferometry experiment in a shallow cased borehole 471

Figure 8 Interferometry by a) vibroseis correlation and by b) vibroseis deconvolution.

Figure 9 Amplitude spectra comparison: interferometry by vibroseis


correlation and by vibroseis deconvolution.

horizontal components of the receivers used as virtual


sources (Fig. 10). This gather simulates (approximates) a
single-well imaging section with zero offset between the virtual
source and receiver. In this case the seismic traces are stacked
autocorrelations, i.e., diagonal elements Ckk of equation (1), k
being the receiver index. A dipping event, corresponding to a
reflection from a layer encountered during subsequent drilling
(and confirmed by the well lithology results and log data), can
be recognized in both these seismic data sets.
Figure 11a shows the traces (P-components) extracted with
10 m offset from the virtual source to the (lower) receiver. The


C

data are filtered in the frequency bandwidth between 80300


Hz. The 10-m-offset single well section of Fig. 11a is compared to the zero-offset section (Fig. 11b) obtained by selecting the virtual-source traces used to calculate Fig. 11(a). We
can observe the difference in the amplitude of the interpreted
reflection at minimum elevation (maximum depth). This effect
is due to the fact that the amplitude of the reflection relative to
the amplitude of the normalized direct arrival, represented by
the event sub-aligned and aligned at zero time in Fig. 11(a,b),
respectively, has strong variations at short source-receiver offsets along the borehole (near-field effect).
These real data can be used, as an approximation, to process
the reflections and signals of the single-well imaging gathers for borehole seismic and acoustic purposes in the lowfrequency approximation.
Figure 12 shows the comparison between 10 m offset singlewell signals obtained by vibroseis cross-correlation and vibroseis deconvolution input data (P-component gathers) and the
corresponding amplitude spectra. A significant bandwidth improvement can be observed in all the traces. This improvement
is more evident in the shallower traces (those at higher elevations), because the vibroseis deconvolution compensates for
the source signature but not for the attenuation effects due to
propagation.
The single-well imaging data were processed to remove direct arrivals and enhance the upgoing wavefield to obtain the
reflectivity in the borehole. The wavefield separation was performed by the use of median and f-k (frequency-wavenumber)
filters. Because of the large amplitude of the direct arrivals
relative to the reflections, some residual dipping artefacts are

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

472 F. Poletto et al.

Figure 10 Virtual signals. Single-well imaging gathers (virtual zero offset) tool: a) P-waves and b) SV-waves BP filtered 20300 Hz. Arrows
indicate the reflections coming from a lithological interface located at about 10 m below sea level. For display purposes the polarity of the
signals is reversed.

Figure 11 Single-well imaging gathers (P-component). a) Simulation of a virtual signal of a 10 m offset tool compared to b) corresponding zero
offset signal. The arrows indicate the reflection coming from an interface located at about 10 m below sea level.

still present after removal of the direct signals. However, these


residual events can be distinguished from reflections in the final sections for the flatter moveout of the artefacts.
The interferometric data interpretation was performed by
comparison with surface reflection seismic, borehole seismic
and well logs data. Figure 13 shows the upgoing wavefield of
the P-wave gather compared with the lithological profile and
borehole sampling and the compressional velocity measured
by the sonic log. The geologic column results from the drill
cutting analysis and well log interpretation. The moveout associated with the virtual reflection events is in agreement with
the formation velocity and the reflection projections in depth
are compared with seismic data obtained by a surface reflection survey and by conventional borehole (VSP) acquisition,
respectively. The strong signal at about 10 m elevation below sea level is detected by the interferometric data as well
as by the conventional surface and borehole data. This signal


C

corresponds to an interface between the conglomerate and


not-well-cemented gravel inside the Montello conglomerate.
At this depth, an abrupt velocity decrease can be observed in
the sonic log data. The borehole interferometric result shows
other reflections confirmed by the sonic log velocity variations. The arrows in Fig. 13 indicate reflections at about
75 m and 100 m, respectively. Both the signals are in correspondence of compressional velocity variations in the sonic
log data.
As for P-wave signals, we perform a comparison between
lithology, sonic log and interferometric data for the SV-wave
configuration (Fig. 14). The P-wave reflections reported in
Fig. 13 at about 75 m and 10 m elevation are also observed
in the SV-wave single-well imaging results (Fig. 14).
The joint analysis performed with well lithology, sonic log
data, conventional surface- and borehole-seismic data confirms that the interferometric approach is suitable to obtain

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

Seismic interferometry experiment in a shallow cased borehole 473

Figure 12 Virtual signals. Single-well imaging gathers obtained by vibroseis cross-correlation (top left panel) and vibroseis deconvolution (top
right panel) and relative amplitude spectra (bottom panels).

reliable signals in the presence of complex overburden for


shallow targets. The interferometric result shows a better
resolution than conventional surface- and borehole-seismic
data. The validation performed with different data suggests
that the acquisition geometry adopted in this experiment is
suitable to obtain a sufficient data coverage able to avoid
artefacts in the interferometric results.
In Fig. 15, P and S interval velocities measured by singlewell imaging gather (10 m offset) are plotted together with
the VP /VS ratio and compared with the sonic-log velocities. Taking into account the different vertical resolution of
these methods (also related to the available receiver intertrace spacing in the well array), the results show good
agreement.


C

DISCUSSION
In this work we use high-frequency seismic signals acquired
in a shallow-medium borehole and closely-spaced surface vibrator sources to calculate single-well virtual gathers with
and without vibroseis deconvolution. Even if it may be in
principle possible to reduce the number of real sources and
still recover essential parts of the interferometric signal (van
Manen et al. 2006), in this test we use dense spatial sampling
for the surface sources subject to variations in their emission
properties due to local coupling conditions. In addition to
correlation with the reference vibrator signal, we investigate
the processing approach by stacking the correlations of the
deconvolved vibrator traces in order to improve the

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

474 F. Poletto et al.

Figure 13 P-wave virtual single-well imaging upgoing wavefield (e) is compared with sonic log data (d), lithology obtained by the well master
log (c), borehole (b) and surface (a) seismic data.

bandwidth in the interferometric signal (Figs 8 and 9). The


frequency analysis shows that the bandwidth of the signal
obtained above 180 Hz is still relevant, with improved S/N
in the virtual single well data obtained at shorter distances
from the boundary surface where the real sources are located
(Fig. 12). This is not surprising because the vibroseis deconvolution improves the bandwidth in the real source signature,
which even if affected by local noise and phase instabilities
in the deconvolved signals obtained at each individual real
source point is rephased by the subsequent interferometry
correlation and then averaged and stabilized by the interferometry integration in the source space. However, note that
this approach does not correct for the propagation effects in
the signature of the virtual signal and the results at different borehole depths differ for frequency attenuation effects


C

(Fig. 12), such as those due to absorption in the wave propagation. These results, obtained in a test well, confirm the
potential of high-frequency borehole interferometry for the
purposes of SWI with seismic sources in the short-medium
range. It is envisaged that the method can be used in conjunction with other interferometry-by-deconvolution approaches
(see, for example, Vasconcelos et al. 2008a,b; Wapenaar, van
der Neut and Ruigrok 2008) to compensate also for the propagation effects in the signal signature.

CONCLUSIONS
A seismic experiment was performed at the OGS
test site to simulate single-well imaging wavefields by
interferometric processing of multi-component data recorded

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

Seismic interferometry experiment in a shallow cased borehole 475

Figure 14 S-wave virtual single-well imaging upgoing wavefield (c) is compared with sonic log data (b) and lithology obtained by the well
master log (a).

Figure 15 P (a) and S (b) interval virtual single-well imaging velocities are compared with sonic log data. The ratio VP /VS is shown in (c).

by a fixed borehole array. The aim of this survey was to obtain a data set with a frequency range of several hundred
hertz usable for simulating single-well imaging acquisition
and processing, useful for a better planning of future borehole tests. The work we present here is an initial analysis
of the data: further developments are foreseen by processing
multi-component data, including SH and comparison with
borehole acoustic data acquired in the instrumented test site.
The presence of a deep water table and of an overburden with


C

loose gravel suggested to use the interferometric approach


to improve the knowledge of the geology in the well. As in
the case of hydrocarbon targets, this near-surface experiment
also allows seismic interferometry to obtain improved local
(downhole) data in the presence of a complex overburden.
Appropriate acquisition parameters were essential to obtain
high-frequency signals, when using a 10400 Hz sweep vibrator source. In particular, we selected very small surface
source intervals to improve the signal quality with dense

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

476 F. Poletto et al.

spatial sampling. The target was to obtain a good approximation of a complete coverage for interferometry (by surface
illumination), which would provide a full reconstruction of
the wavefields and amplitudes in the proximity (below) of the
sources. The results were used to simulate single-well gathers
in the low-frequency approximation. The single-well imaging data were processed and validated by lithological, sonic
log and conventional surface- and borehole-seismic data. The
main geological interfaces in the sedimentary sequence were
detected by the interferometric approach, with a better resolution than the conventional seismic methods.
Processing tests performed by using vibroseis deconvolution
before interferometry demonstrated that vibroseis deconvolution is beneficial and the procedure is robust when we use
vibroseis deconvolution signals to serve as input for virtual
signal processing.

ACKNOWLEDGEMENTS
We thank David F. Halliday, the associate editor and anonymous reviewers for their fruitful comments and suggestions
that helped to improve the manuscript. Thanks to the OGS
crew for field acquisition support. Part of the basic processing
was performed by Seismic Unix.
REFERENCES
Baeten G.J.M. and Ziolkowski A.M. 1990. The Vibroseis Source.
Elsevier.
Bakulin A. and Calvert R. 2004. Virtual source: New method for
imaging and 4D below complex overburden. 74th SEG meeting,
Denver, Colorado, USA, Expanded Abstracts, 24772480.
Bakulin A., Mateeva A., Calvert R., Jorgensen P. and Lopez J. 2007.
Virtual shear source makes shear waves with air guns. Geophysics
72(2), A7A11.
Brittle K.F., Lines L.R. and Dey A.K. 2001. Vibroseis deconvolution:
A comparison of cross-correlation and frequency domain sweep
deconvolution. Geophysical Prospecting 49, 675686.
Calvert R.W. 2004. Seismic imaging a subsurface formation. US
Patent 6,747,915.
Chabot L., Henley D.C., Brown R.J. and Bancroft J.C. 2001. Singlewell imaging using the full waveform of an acoustic sonic. 71st SEG
meeting, San Antonia, Texas, USA, Expanded Abstracts, 420423.
Claerbout J.F. 1968. Synthesis of a layered medium from its acoustic
transmission response. Geophysics 33, 264269.
Gaiser J.E., Vasconcelos I. and Ramkhelawan R. 2009. Elasticwavefield interferometry Pseudo-source VSPs from 3D P-wave


C

vibrator data. 71st EAGE meeting, Amsterdam, the Netherlands,


Expanded Abstracts.
Haldorsen J.B.U. and Borland W. 2008. Use of seismic vibrators
for vertical seismic profiling: Considerations for the choice of
sweep. EAGE Vibroseis Workshop, Prague, Czech Republic, October 2008, E04.
Haldorsen J.B.U., Miller D.E. and Walsh J.J. 1994. Multichannel
Wiener deconvolution of vertical seismic profiles. Geophysics 59,
15001511.
Halliday D., Curtis A. and Kragh E. 2008. Seismic surface waves
in a suburban environment Active and passive interferometric
methods. The Leading Edge 27, 210218.
Hornby B.E. 1989. Imaging of near-borehole structure using fullwaveform sonic data. Geophysics 54, 747757.
van Manen D.J., Curtis A. and Robertsson J.O. 2006. Interferometric
modelling of wave propagation in inhomogeneous elastic media
using time-reversal and reciprocity. Geophysics 71, S147S160.
Mateeva A., Ferrandis J., Bakulin A., Jorgensen P., Gentry C. and
Lopez J. 2007. Steering virtual sources for salt and subsalt imaging.
77th SEG meeting, San Antonio, Texas, USA, Expanded Abstracts,
30443047.
Mehta K., Snieder R., Calvert R. and Sheiman J. 2008. Acquisition geometry requirements for generating virtual-source data. The Leading Edge 27, 620629.
Mewhort L., Bezdan S. and Jones M. 2002. Does it matter what
kind of vibroseis deconvolution is used? CSEG Geophysics 2002
Conference, Expanded Abstracts, 14.
Poletto F., Corubolo P. and Comelli P. 2010. Drill-bit seismic interferometry with and without pilot signals. Geophysical Prospecting
58, 257265.
Poletto F., Petronio L., Farina B. and Schleifer A. 2008. Single Well
Imaging in cased borehole by interferometry. 70th EAGE meeting,
Rome, Italy, Expanded Abstracts, P278.
Sallas J.J. 1984. Seismic vibrator control and the downgoing P-wave.
Geophysics 49, 732740.
Snieder R., Wapenaar K. and Wegler U. 2007. Unified Greens function retrieval by cross-correlation; connection with energy principles. Physical Review E 75, 036103.
Vasconcelos I. and Snieder R. 2008a. Interferometry by Deconvolution: Part 1 Theory for acoustic wave and numerical examples.
Geophysics 73, S115129.
Vasconcelos I. and Snieder R. 2008b. Interferometry by Deconvolution: Part 2 Theory for elastic waves and application to drill-bit
seismic imaging. Geophysics 73, S129141.
Wapenaar K. and Fokkema J. 2006. Greens function representation
for seismic interferometry. Geophysics 71, SI33SI46.
Wapenaar K., Van Der Neut J. and Ruigrok E. 2008. Passive seismic
interferometry by multidimensional deconvolution. Geophysics 73,
A51A56.
Zanferrari A., Poli M.E. and Rogledi S. 2002. The external thrustbelt of the eastern Southern Alps in Friuli (NE Italy). Memorie della
Societa` Geologica Italiana 54, 159162.

2010 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 464476

Das könnte Ihnen auch gefallen