U.S. patent application number 12/551897 was filed with the patent office on 2010-01-14 for method of seismic surveying.
Invention is credited to Benjamin Peter Jeffryes.
Application Number | 20100008187 12/551897 |
Document ID | / |
Family ID | 9934389 |
Filed Date | 2010-01-14 |
United States Patent
Application |
20100008187 |
Kind Code |
A1 |
Jeffryes; Benjamin Peter |
January 14, 2010 |
METHOD OF SEISMIC SURVEYING
Abstract
A method enables the time required to complete a seismic survey
and the noises recorded in the seismic data to be reduced. One
aspect of the method includes actuating a first vibrator group to
start a first sweep at time T0; and a second vibrator group to
start a second sweep at time T1, wherein T0<T1<T0+S1+L, where
S1 is a sweep time of the first vibrator group and L is a listening
time; wherein the time between the first sweep and the second sweep
is T1-T0.gtoreq.((n-1)*f1*S1)/(n*(f1-f0)) where n is a natural
number, f0 is the lower frequency limit of the vibrator sweep and
f1 is the upper frequency limit of the vibrator sweep.
Inventors: |
Jeffryes; Benjamin Peter;
(Cambridgeshire, GB) |
Correspondence
Address: |
Jeffrey E. Griffin, IP Dept.;WesternGeco L. L. C.
10001 Richmond Avenue
Houston
TX
77042
US
|
Family ID: |
9934389 |
Appl. No.: |
12/551897 |
Filed: |
September 1, 2009 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11944690 |
Nov 26, 2007 |
7602670 |
|
|
12551897 |
|
|
|
|
11548521 |
Oct 11, 2006 |
7376046 |
|
|
11944690 |
|
|
|
|
11299410 |
Dec 12, 2005 |
|
|
|
11548521 |
|
|
|
|
10403135 |
Mar 31, 2003 |
7050356 |
|
|
11299410 |
|
|
|
|
Current U.S.
Class: |
367/38 |
Current CPC
Class: |
G01V 1/005 20130101 |
Class at
Publication: |
367/38 |
International
Class: |
G01V 1/28 20060101
G01V001/28 |
Foreign Application Data
Date |
Code |
Application Number |
Apr 6, 2002 |
GB |
0207995.2 |
Claims
1. A method of processing slip sweep seismic data comprising:
grouping the slip sweep seismic data into responses of at least two
groups of vibrators, wherein at least one group has at least two
vibrator sources; generating impulse responses for each source in
each vibrator group; estimating the harmonic noises with
cross-correlating slip sweep seismic data with a harmonic frequency
sweep; and removing the harmonic noises.
2. The method of claim 1 wherein the cross-correlated data is
Fourier transformed.
3. The method of claim 1 wherein the cross-correlated data is time
windowed.
4. The method of claim 1 wherein the data is Fourier transformed
and time windowed.
5. The method of claim 1 wherein the harmonic frequency sweep is
the same as the fundamental frequency sweep, except the sweep rate
is an integer multiple of the sweep rate of the fundamental
frequency sweep.
6. The method of claim 1, wherein the slip sweep seismic data are
responses to K shots at a shot location of N vibrators in M groups,
where K, M, N are integers, K is at least N+1, further comprising
varying the phase of the shots.
7. A seismic data processing system, comprising: a processor; and a
computer readable media containing computer readable instructions,
wherein the instructions control the processor to perform the steps
of: grouping the slip sweep seismic data into responses of at least
two groups of vibrators, wherein at least one group has at least
two vibrator sources; generating impulse responses for each source
in each vibrator group; estimating the harmonic noises with
cross-correlating slip sweep seismic data with a harmonic frequency
sweep; and removing the harmonic noises.
Description
RELATED APPLICATIONS
[0001] U.S. patent application, attorney docket No. 57.0371-US-CON2
is a Continuation Application of the co-pending U.S. patent
application Ser. No. 11/548,521, filed on 11 Oct., 2006, attorney
docket No. 57.0371-US-CON1, now a U.S. Pat. No. 7,376,046, issued
on May 20, 2008, by the same inventor with the same title and
assigned to the same assignee, which is currently pending.
[0002] U.S. patent application Ser. No. 11/548,521 is a
Continuation Application of U.S. patent application Ser. No.
11/299,410, attorney docket No. 57.0371-US-DIV, filed on 12 Dec.,
2005, by the same inventor with the same title and assigned to the
same assignee, which is currently pending.
[0003] U.S. patent application Ser. No. 11/299,410, attorney docket
No. 56.0371-US-DIV is a Divisional Application of U.S. patent
application Ser. No. 10/403,135, attorney docket number 57.0371-US,
filed on 31 Mar. 2003, by the same inventor with the same title and
assigned to the same assignee, which is now U.S. Pat. No.
7,050,356, issued on 23 May, 2006.
[0004] U.S. patent application Ser. No. 10/403,135, attorney docket
No. 57.0371-US claims priority from a GB Patent Application No.
0207995.2, filed on 6 Apr., 2002, by the same inventor with the
same title and assigned to the same assignee, which is now GB
Patent No. 2 387 226, issued on 10 Aug. 2005.
FIELD OF THE INVENTION
[0005] The present invention relates to a method of seismic
surveying. In particular, it relates to a method of seismic
surveying in which two or more groups of vibrator sources emit
seismic energy in such a way that their sweep times overlap with
one another.
BACKGROUND OF THE INVENTION
[0006] The principle of seismic surveying is that a source of
seismic energy is caused to produce seismic energy that propagates
downwardly through the earth. The downwardly-propagating seismic
energy is reflected by one or more geological structures within the
earth that act as partial reflectors of seismic energy. The
reflected seismic energy is detected by one or more sensors
(generally referred to as "receivers"). It is possible to obtain
information about the geological structure of the earth from
seismic energy that undergoes reflection within the earth and is
subsequently acquired at the receivers.
[0007] In practice, a seismic surveying arrangement comprises an
array of sources of seismic energy. This is because it is necessary
to generate sufficient energy to illuminate structures deep within
the earth, and a single seismic source generally cannot do
this.
[0008] Sources of seismic energy are known which emit seismic
energy at more than one frequency. Examples of such seismic sources
are vibrator sources, which emit seismic energy in a frequency
range of, for example, from 5 or 10 Hz to 100 Hz. When such a
vibrator source is actuated, seismic energy is emitted over a
finite time period, and the frequency of the emitted energy changes
during the period over which seismic energy is emitted. For
example, the frequency of the emitted energy may increase
monotonically during the period over which seismic energy is
emitted. The process of operating a vibrator source of seismic
energy to cause emission of seismic energy over the frequency range
of the vibrator will be referred to herein as "sweeping" the
vibrator, and the step of initiating a vibrator sweep will be
referred to as "actuating" the vibrator. Each emission of seismic
energy from a vibrator is known as a "shot". The time period over
which seismic energy is emitted by the vibrator source will be
referred to as the "sweep time", and the "sweep rate" is the rate
at which the frequency changes over the sweep time (a linear sweep
rate is generally used in practice).
[0009] A seismic vibrator source for use on land consists generally
of a baseplate in contact with the ground. Seismic energy is
transmitted into the ground by applying a vibratory force to the
plate, and this is done by applying a control waveform known as a
"pilot sweep" to the vibrator control mechanism. The pilot sweep is
generally a constant amplitude swept frequency signal, although it
tapers off at each end to allow the amplitude of the vibration to
be ramped up and down at the start and finish of the sweep
respectively. In practice the waveform applied to the ground by the
plate is not exactly the same as the pilot waveform; in particular,
as well as applying a force at the desired frequency at any
particular time (known as the "fundamental frequency"), the
vibrator also applies a force at integer multiples of the
fundamental frequency (known as "harmonics").
[0010] Marine vibrator sources of seismic energy are also known.
They are again swept so as to emit seismic energy over a range of
frequencies.
[0011] When a seismic vibrator source is actuated to emit seismic
energy, the seismic energy incident on a receiver is recorded for a
pre-determined period from the start of the sweep time of the
source. The time from the end of the sweep time to the end of the
recording period is generally known as the "listening time", and
data is acquired at a receiver from the start of the sweep time to
the end of the listening time. The data acquired at a receiver in
consequence of actuation of a source is then processed, for example
by cross-correlating the acquired data with the pilot sweep of the
source to produce a record that is the length of the listening
time.
[0012] FIG. 1 is a schematic illustration of the process of a
conventional seismic survey that uses an array of land vibrator
sources. At time T0, one seismic source in the source array is
actuated to start its sweep. In this example, the vibrator sweep
time has a duration S, and the frequency of seismic energy emitted
by the vibrator increases monotonically from a frequency f.sub.0 at
time T0 to a frequency f.sub.1 (f.sub.1>f.sub.0) at the
conclusion of the sweep (at time T0+S). The sweep time is followed
by the listening time, so that the overall time of the process of
actuating and sweeping the source and listening at a receiver for
seismic energy is S+L, where L is the duration of the listening
time.
[0013] In a conventional seismic survey the sources are actuated
such that a receiver will receive seismic energy from only one
source in any given listening period. The minimum delay between the
start of two vibrator sweeps in such a survey is therefore the sum
of the sweep time S and the listening time L. The listening time L
is made sufficiently large that all seismic energy required at a
receiver in a particular listening period was emitted during the
sweep time immediately preceding that listening period.
[0014] The conventional seismic surveying process has the
disadvantage that it can be slow, owing to the need for the minimum
time delay between the starts of two vibrator sweeps to be the sum
of the sweep time and the listening time. One known attempt to
reduce the time required to carry out a seismic survey is the
"slip-sweep" acquisition technique. In the slip-sweep technique the
minimum time delay between the starts of two subsequent vibrator
sweeps is only the listening time, not the sum of the sweep time
and the listening time. The record length after cross-correlation
is the length of the listening time.
[0015] The slip-sweep technique is illustrated in FIG. 2. As in the
method of FIG. 1, one seismic source in the source array is
actuated to start its sweep at time T0, the vibrator sweep time has
a duration S, and the sweep period is followed by a listening time
L. The time T1 at which a second source is actuated to start its
sweep is not however required to satisfy T1>T0+S+L, but is only
required to satisfy T1>T0+L. Since the minimum time delay
between actuation of two sources in the slip-sweep technique is
only the listening time, not the sum of the sweep time and the
listening time, the slip-sweep technique allows the time to
complete a seismic survey to be reduced. It has the disadvantage,
however, that harmonics of the fundamental frequency generated by
one vibrator are present on the seismogram recorded by one or more
preceding vibrators.
[0016] A further known surveying technique is the technique of
"simultaneous shooting". In the simultaneous shooting method two or
more seismic sources disposed at respective shot locations are
actuated to start their sweeps at the same time. The seismic energy
acquired at a receiver will therefore contain events arising from
seismic energy emitted by all sources. In order to allow the events
corresponding to each source to be separated out from one another,
each vibrator must sweep, at its shot location, at least as many
times as there are vibrators in the group, and the recorded data
are then manipulated algebraically to separate the events
corresponding to each source. Typically each vibrator will sweep
for the same length of time, at the same sweep rate, and over the
same frequency range, but the phase relationship between vibrators
changes from one record to another. In the case of a group of two
vibrators, for example, one suitable scheme would be for the two
vibrators to sweep in phase during the first record and to sweep
180.degree. out of phase during the second record. The mean of the
two signals acquired by a receiver gives the signal arising at that
receiver from actuation of one vibrator, and half the difference of
the two signals acquired by a receiver gives the signal arising at
that receiver from actuation of the other vibrator.
SUMMARY OF THE INVENTION
[0017] A first aspect of the present invention provides a method of
seismic surveying comprising the steps of: actuating the or each
vibrator in a first vibrator group at time T0; and actuating the or
each vibrator in a second vibrator group at time T1, where
[0018] T0<T1<T0+S1+L where S1 is the sweep time of the first
vibrator group and L is the time during which data is acquired at a
receiver; wherein at least one of the first vibrator group and the
second vibrator group comprises at least two vibrators.
[0019] The present invention provides a method of seismic surveying
that combines the known simultaneous acquisition technique and the
known slip-sweep acquisition technique. It makes use of the
simultaneous acquisition technique in that at least one of the
groups of vibrators contains two or more vibrators, but the
different groups are swept using a slip-sweep technique. The
present invention provides a reduction in the time required to
carry out a seismic survey compared to the time required by a
conventional slip-sweep technique.
[0020] The invention may be applied to a single vibrator group, in
which case the second vibrator group is the first vibrator group,
or the first and second vibrator groups may be distinct.
[0021] In a preferred embodiment the method comprises the step of
actuating the or each vibrator group at least as many times as
there are vibrators in the respective group and such that the
contribution from each vibrator in a group may be determined by
algebraic operation on receiver records of sweeps made by the
respective group.
[0022] In a preferred embodiment T1-T0>(n-1)S1 f.sub.1/n
(f.sub.1-f.sub.0), where n is a natural number, f.sub.0 is the
lower frequency limit of the vibrator sweep and f.sub.1 is the
upper frequency limit of the vibrator sweep. This enables the noise
in the acquired data arising from the m.sup.th harmonic of the
fundamental frequency to be estimated for all m.ltoreq.n.
[0023] In a preferred embodiment the method comprises the steps of:
actuating the or each vibrator in the first vibrator group at time
T2, where T1<T2<T1+S2+L where S2 is the sweep time of the
second vibrator group; and actuating the or each vibrator in the
second vibrator group at time T3 where T2<T3<T2+S+L; the
first vibrator group is different from the second vibrator group;
and T3-T2.noteq.T1-T0.
[0024] This embodiment allows a further noise reduction technique
to be applied. The varying time delay between a shot of the first
group and the shot of a second group means that harmonic noise will
occur at different times in the records of the two shots. The noise
may therefore eliminated by appropriately summing the two shot
records, on the assumption that each shot record contains the same
signal.
[0025] A second aspect of the present invention provides a method
of seismic surveying comprising the steps of: actuating a first
vibrator at time T0; actuating a second vibrator different from the
first vibrator at time T1, where T0<T1<T0+S1+L where S1 is
the sweep time of the first vibrator and L is the time during which
data is acquired at a receiver; actuating the first vibrator at
time T2, where T1<T2<T1+S2+L where S2 is the sweep time of
the second vibrator; and actuating the second vibrator at time T3
where T2<T3<T2+S1+L and where T3-T2.noteq.T1-T0.
[0026] In a preferred embodiment T3-T2>T1-T0.
[0027] A third aspect of the present invention provides a seismic
surveying arrangement comprising: a plurality of vibrator sources;
and a control means adapted to actuate the or each vibrator in a
first vibrator group at time T0 and to actuate the or each vibrator
in a second vibrator group at time T1, where T0<T1<T0+S1+L
where S1 is the sweep time of the first vibrator group and L is the
time during which data is acquired at a receiver.
[0028] A fourth aspect of the present invention provides a seismic
surveying arrangement comprising: a plurality of vibrator sources;
and a control means adapted to: (a) actuate a first vibrator at
time T0; (b) actuate a second vibrator different from the first
vibrator at time T1, where T0<T1<T0+S1+L where S1 is the
sweep time of the first vibrator and L is the time during which
data is acquired at a receiver; (c) actuate the first vibrator at
time T2, where T1<T2<T1+S2+L where S2 is the sweep time of
the second vibrator; and (d) actuate the second vibrator at time T3
where T2<T3<T2+S1+L and where T3-T2.noteq.T1-T0.
[0029] In a preferred embodiment the control means comprises a
programmable data processor.
[0030] A fifth aspect of the invention provides a medium containing
a program for the data processor of a seismic surveying arrangement
as defined above.
BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Preferred embodiments of the present invention will now be
described by way of illustrative examples with reference to the
accompanying figures in which:
[0032] FIG. 1 is a schematic timing diagram of a conventional
seismic survey;
[0033] FIG. 2 is a schematic timing diagram of a conventional
slip-sweep seismic survey;
[0034] FIG. 3 is a schematic timing diagram of a seismic survey
according to a first embodiment of the present invention;
[0035] FIG. 4 is a block schematic flow diagram of one embodiment
of the present invention;
[0036] FIG. 5 is a schematic illustration of raw seismic data;
[0037] FIG. 6 illustrates the result of processing the raw seismic
data of FIG. 5 according to a first method of the present
invention;
[0038] FIG. 7 illustrates the results of processing the seismic
data of FIG. 5 according to a second method of the present
invention;
[0039] FIG. 8 shows the result of processing the seismic data of
FIG. 5 according to the first and second methods of the
invention;
[0040] FIG. 9(a) is a schematic diagram of a seismic surveying
arrangement according to the present invention; and
[0041] FIG. 9(b) is a schematic diagram of a control means of the
seismic surveying arrangement of FIG. 9(a).
DETAILED DESCRIPTION OF THE INVENTION
[0042] The operation of a seismic surveying arrangement according
to a first embodiment of the present invention is illustrated
schematically in FIG. 3, which is a timing diagram of the method.
This method enables the time taken to complete a seismic survey to
be reduced compared to the conventional slip-sweep method described
above.
[0043] The method assumes that the seismic survey has an array of
seismic vibrator sources. In order to carry out the method, the
vibrators are grouped into two or more groups, with one group of
vibrators being operable independently of the or each other group
of vibrators. The grouping may be a physical grouping, for example
with the vibrators being arranged in a two-dimensional array with
each row or column constituting a group. Alternatively the grouping
may be a notional grouping, in which vibrators in an array are
notionally divided into two or more groups. It is not necessary for
each group to have the same number of vibrators, but at least one
group is required to contain two or more vibrators.
[0044] At time T0 a first group of vibrators is actuated. That is,
each vibrator in the first group starts its sweep at time T0. All
vibrators in the first group will have the same sweep time. (For
each group, all vibrators in the group will have the sweep time. In
principle different groups could have different sweep times,
although it is usual for all groups to have the same sweep time as
one another.) If the sweep time of the vibrators in the first group
is S1, then the sweep period of the first vibrator group lasts from
time T0 to time T0+S1. The sweep period for the first group is then
followed by a listening time, which has a duration L and so
concludes at time T0+S1+L.
[0045] A second group of vibrators is actuated to start their
sweeps at time T1. The time T1 is before the end of the listening
period of the first vibrator group--that is,
[0046] T1<T0+S1+L. The group of vibrators actuated at time T1 is
referred to as a second vibrator group for convenience, but the
second vibrator group either could be different from, or could be
the same as, the first group of vibrators. The sweep period of the
second vibrator group has a duration S2 (which may be equal to S1)
and so extends from T1 to T1+S2, and the listening period then
extends to T1+S2+L (the two groups will have the same listening
time). The time delay between the start of the sweep of the first
vibrator group and the start of the sweep of the second vibrator
group must exceed the listening time (and if the same group is
swept twice in succession then the delay must be greater than the
sweep time of that group).
[0047] FIG. 3 shows only the first sweep of the first group and the
first sweep of the second group. However, the method of the
invention requires that each group of vibrators is swept at least
as many times are there are vibrators in that group, in a sequence
from which the contribution from each vibrator to the seismic data
acquired at a receiver can be determined, for example
algebraically.
[0048] In general, there will be M groups of vibrators, with N
vibrators in each group. If each group makes K shots, where
K.gtoreq.N, there will be a sequence of MK shots. For example, if
there are two groups of three vibrators (M=2, N=3) each group must
make at least 3 shots (K.gtoreq.3). If each group makes exactly
three shots, one possible sequence would be: group 1; group 2;
group 1; group 2; group 1; group 2. Some differentiation between
the sweeps, for example a phase change, is required to allow the
records to be algebraically manipulated to isolate the contribution
of each vibrator. (In principle, if a group were to make more shots
than there are vibrators in the group (that is, if K>N) some of
the shots could be identical provided that there were at least N
independent shots.)
[0049] As another example, consider a single group of 4 (M=1, N=4).
In this case the lowest value of K is 4, so the group must be swept
at least 4 times and the shortest possible sequence is: group 1;
group 1; group 1; group 1.
[0050] As another example, consider six vibrators arranged in three
groups each of two vibrators (M=3, N=2). In this case the lowest
value of K is 2. If each group makes exactly two shots, one
possible sequence would be: group 1; group 2; group 3; group 1;
group 2; group 3.
[0051] In this method of the invention, at least one of the shots
takes place within time S+L of the start of the preceding shot.
Preferably, as many as possible of the shots start less than S+L
after the previous shot. It will be noted that the very last shot
will not contain slip-sweep noise--for example, in the simple
example of shooting with just two vibrators, the second vibrator
would shoot within the listening time of the first vibrator (that
is, within time S+L of the start of the first shot). Both vibrators
would then be moved to their next shot points so that the shot of
the second vibrator would not have any slip-sweep noise on it.
[0052] In the above examples each group of vibrators contains the
same number of vibrators. The invention is not limited to this, and
the groups need not all contain the same number of vibrators
(although that there must be a group containing more than one
vibrator). If the groups do not contain the same number of
vibrators, each group must be swept at least one more time than
there are vibrators in that group.
[0053] FIG. 3 indicates that vibrators in a third vibrator group
(which could be the first or second group) are actuated to start
their sweeps at time T2 that satisfies T2<T1+S2+L, where S2 is
the sweep time of the second group of vibrators. It would, however,
be possible for the third group to be actuated at a time
T2>T1+S2+L since it is necessary for only one shot to start
within S+L of the start of the previous shot.
[0054] Once each vibrator group has been actuated a sufficient
number of times the vibrators may be moved to different positions
to allow a fresh survey to be carried out.
[0055] The response at each receiver owing to the seismic energy
generated by each individual vibrator may be calculated from the
seismic energy acquired at the receiver(s) by using any suitable
technique. One suitable technique is disclosed in UK patent
application No. 2 359 363. In principle the response could be
calculated immediately the data have been acquired, but it is more
usual for the data to be stored, for example on magnetic tape or
disc, for subsequent processing away from the survey location.
[0056] One problem with slip-sweep acquisition is that the data
obtained in the listening period following one sweep will contain
harmonic noise that arises from the subsequent sweep. It is
desirable to eliminate this harmonic noise during processing of the
data. A further embodiment of the present invention provides a
method for estimating the harmonic noise in data acquired in a
listening period from the acquired data.
[0057] The harmonic noise in data acquired in one listening period
is the earth's response to the harmonic output from the subsequent
sweep. The earth's response for the harmonics of the subsequent
shot is the same as the earth's response to the fundamental sweep
of the shot associated with the listening period. Vibrator
deconvolution theory is based on knowing the content of the
fundamental sweep as this provides a measurement of the earth's
response. If it is possible to estimate also the harmonic output of
the vibrator, then it is possible to estimate the harmonic
contribution to the preceding shot by convolving the earth's
response, as determined from the content of the fundamental sweep,
with the harmonic output of the vibrator. Once the harmonic
contribution to the preceding shot has been found in this way, it
is possible to subtract it from the recorded data.
[0058] In principle, it would be possible to use measurements on
the seismic energy emitted at the vibrator to estimate the harmonic
output of the vibrator. However, the effects of the non-linearity
of the earth's behaviour near the vibrator can mean that the
harmonic noise at the receiver differs from the harmonic noise that
would be estimated from measurements on the vibrator. Estimating
the harmonic output from the data itself is therefore a more
reliable method.
[0059] One embodiment of a method for eliminating harmonic noise
from data acquired in the simultaneous slip-sweep acquisition
technique or FIG. 3 will now be described with reference to FIG.
4.
[0060] Initially, at step 1, vibrator sources are grouped into two
or more groups. This grouping step may be a physical grouping step,
for example during the deployment of the vibrators. Alternatively
it may consist of defining notional groups in a vibrator array. At
least one of the groups contains two or more vibrators.
[0061] At step 2, simultaneous slip-sweep data are acquired using
an acquisition technique of the type described generally with
respect to FIG. 3.
[0062] At step 3, the impulse response is calculated for each
group, from the positions of the vibrators in that group. This step
may be carried out in a conventional manner using cross-correlation
and an inversion matrix that is constant with frequency, or it may
alternatively be done in the manner described in GB-A-2 359
363.
[0063] Next, at step 4, the acquired data are cross-correlated with
a harmonic frequency sweep. The sweep that is used in this
cross-correlation step is essentially the same as the fundamental
sweep of the vibrator, except that it has a sweep rate that is an
integer multiple of the sweep rate of the fundamental sweep. Thus,
the sweep used in the cross-correlation step would have twice the
sweep rate of the fundamental vibrator sweep if it is desired to
remove the second harmonic, would have three times the sweep rate
of the fundamental sweep for removing the third harmonic, and so
on. It is not necessary for the sweep used in the cross-correlation
step to extend above the upper frequency limit of the fundamental
sweep--and in any case the upper frequency should not exceed the
Nyquist frequency for the sampling--so a taper is applied to the
harmonic sweep after a frequency near the upper frequency of the
fundamental sweep. For each harmonic, the same harmonic sweep may
be used in the cross-correlation step for all receivers and for all
shots.
[0064] The effect of the cross-correlation step is to concentrate
in time the earth's response to one harmonic. The earth's response
to other harmonics and to the fundamental frequency are spread out
in time. In particular, although the earth's response to the
fundamental sweep frequency will still be the main contributor to
the acquired data, the effect of the cross-correlation step is that
it will now arrive at a later time in the records.
[0065] At step 5, the cross-correlated data is Fourier transformed
over a time window that includes the main first arrival due to the
desired harmonic, but which ends before the earth's response to the
fundamental vibrator frequency is seen or (in the case of
estimating higher harmonics) when the earth's response to a higher
amplitude harmonic is seen. The time window length used in this
step may vary from one receiver to another. For each frequency and
receiver, the result of the Fourier transform is a vector G.sub.n
having length N, where N is the number of shots. This vector is the
early part of the earth's response to the chosen vibrator
harmonic.
[0066] At step 6, the impulse response for each receiver and
vibrator position is treated with a Fourier transform of the same
length as in step 5. The Fourier transform is carried out over the
same time window, or a slightly shorter time window, than the
Fourier transform in step 5. The results of this Fourier transform
step are a vector R.sub.n for each receiver and frequency, with
R.sub.n having a length M where M is the number of vibrator
positions.
[0067] At step 7, the matrix G.sub.nR*.sub.n is calculated for each
frequency and receiver. R*.sub.n is the complex conjugate of the
transpose of R.sub.n. This yields an N.times.M matrix. This matrix
is then averaged over receivers to give the cross-correlation
matrix G.sub.nR*.sub.n. A receiver-based normalisation may be
applied in this averaging process, to allow for the varying signal
amplitude at different receivers.
[0068] At step 8 the M.times.M matrix R.sub.nR*.sub.n is calculated
for each frequency and receiver. This is then averaged over
receivers to give the auto-correlation matrix R.sub.nR*.sub.n. If a
receiver-based normalisation was applied in step 7, the same
normalisation should be applied in step 8.
[0069] In step 9 the estimate of the n.sub.th harmonic H.sub.n
emitted by the vibrator is determined. In principle H.sub.n is
given by H.sub.n=G.sub.nR*.sub.nR.sub.nR*.sub.n.sup.(l). However,
in practice the auto-correlation matrix will be badly
conditioned--that is the smallest eigenvalue will be much smaller
than the largest eigenvalue, and taking an exact inverse runs the
risk of being dominated by noise. A standard singular value
decomposition of R.sub.nR*.sub.n decomposes it into the product of
three matrices, U, V, and .LAMBDA., where
[0070] R_hd nR*.sub.n=U.LAMBDA.V*, U and V are such that
UU*=VV*=*=I where I is the identity matrix) and .LAMBDA. is
diagonal and real. U and V are the matrices of right and left
eigenvectors respectively, .LAMBDA. is composed of the eigenvalues.
The true inverse matrix is given by
[0071] V.LAMBDA..sup.-1U* The estimate of H.sub.n is given by
H.sub.n=G.sub.nR*.sub.n(VLU*) where L is a diagonal matrix,
identical to .LAMBDA..sup.(-1), except the smallest elements have
been replaced by zeros. One way of deciding how many eigenvalues to
retain and how many to set to zero comes from comparing the
root-mean-square (rms) size of the elements of R.sub.nR*.sub.n. The
rms size of the elements of VLU* times the rms size of
R.sub.nR*.sub.n should be of order 1. Often only one eigenvalue is
necessary. H.sub.n is an N by M matrix for each frequency and set
of shots.
[0072] At step 10 the matrix H.sub.n is Fourier transformed back to
the time domain to obtain h.sub.n. The Fourier transform is carried
out for a time window around T=0 using a smooth taper, and having a
half width of around 0.25 seconds or less.
[0073] At step 11, h.sub.n is convolved with the harmonic sweep
used in step 4. The result of this convolution is the estimate of
the vibrator output for the n.sup.th harmonic. This is an estimate
of the noise that appears on the data acquired during the preceding
shot. The data acquired for the preceding shot can be corrected for
the harmonic noise, for example by subtracting the harmonic noise
estimated from the raw data acquired at the receiver.
Alternatively, if the deconvolution stage is performed on
correlated data, the harmonic noise can be correlated with the
appropriate pilot sweep and then subtracted from the appropriate
correlated records.
[0074] The above steps may be repeated for each harmonic which it
is desired to remove. Typically, the second and third harmonics
have the greatest amplitude, so that removal of just these
harmonics may be sufficient.
[0075] At step 12 the shot records, which have now had the harmonic
noise removed, are separated into their individual shot point
components. This step may be done on either uncorrelated or
correlated shot records. Alternatively, the separation stage can be
carried out on the summed noise estimates alone, and these
separated noise estimates are then subtracted from previous
calculated individual shot point components.
[0076] The above method has been described with reference to the
simultaneous sub-sweep acquisition method of FIG. 3. It may
alternatively be applied to a conventional slip-sweep acquisition
method of the type shown in FIG. 2. The method can be simplified
when applied to such a conventional slip-sweep technique, since the
cross-correlation and the auto-correlation are not matrixes, but
are a number at each frequency. Thus, in step 9 the true inverse of
the auto correlation can be used.
[0077] Use of this method requires a certain minimum time between
sweeps. In order to remove the n.sup.th harmonic, the minimum time
between consecutive sweeps is given by:
T m i n = ( n - 1 ) f 1 S n ( f 1 - f 0 ) ( 1 ) ##EQU00001##
[0078] This assumes that the fundamental sweep varies linearly
between frequency f.sub.0 and frequency f.sub.1 and has a sweep
time of S.
[0079] Thus, to remove the second harmonic the slip between
consecutive sweeps (for example, T1-T0 in FIG. 2 or FIG. 3) must be
slightly more than half the overall sweep duration, to remove the
third harmonic it must be slightly more than two thirds of the
sweep time, etc.
[0080] FIGS. 5 and 6 illustrate results of the method of FIG. 4.
FIG. 5 shows part of a data record acquired at a receiver during
simultaneous actuation of three seismic vibrators, with each
vibrator being actuated four times. Although the data was acquired
with a conventional simultaneous acquisition technique, rather than
a simultaneous slip-sweep acquisition technique, the records were
summed to simulate data acquired by a simultaneous slip-sweep
technique of the present invention. It will be seen that there is
high frequency harmonic noise across much of the data. The twelve
traces shown in FIG. 5 were acquired at twelve separate receiver
locations.
[0081] FIG. 6 illustrates the result of applying a harmonic
estimation and removal method of the type described with reference
to FIG. 4 to the second and third harmonics in the data of FIG. 5.
It can be seen that the high frequency harmonic noise has been
significantly reduced.
[0082] An alternative technique for removing harmonic energy will
now be described. In contrast to the above-described method of
harmonic noise reduction, which relies on being able to estimate
the harmonic noise, the technique described below requires very
little knowledge of the harmonic energy. The method requires only
knowledge of at which time different orders of harmonic energy
arrives at a receiver. This method may be applied to both the
simultaneous slip-sweep acquisition of FIG. 3 or to conventional
slip-sweep acquisition of seismic data. One restriction on this
technique is that there must be at least one more sweep made than
the number of shot points that are being separated. Thus, for
standard slip-sweep data acquisition each vibrator must be actuated
at least twice in each shot location. In a simultaneous slip-sweep
acquisition technique in which each group includes two vibrators,
at least three sweeps must be made in each location.
[0083] Consider initially the case of a standard slip-sweep
acquisition technique in which each vibrator group makes two shots
in each location. After cross-correlation or deconvolution the
signal components of the data record for one shot will be the same
as the signal component of the data record for the other shots. If
the harmonic noise component in the two records can be arranged to
differ, then it is possible to reduce or eliminate the harmonic
noise by combining the two records appropriately. One way of doing
this that has been proposed previously is to vary the phase of the
shots. If the phase of the n.sup.th harmonic is n times greater
than the phase of the fundamental sweep (which is usually
approximately true), then by choosing an appropriate phase
difference and summing the records after cross-correlation, one or
more harmonics will cancel out. If the two shots are 90.degree. out
of phase then the third harmonic will cancel. If the two shots are
180.degree. out of phase then the second and fourth harmonics will
cancel. If there are three shots, each being 120.degree. out of
phase with the other two shots, then the second and third harmonics
will cancel.
[0084] The method described below uses a combination of an
acquisition technique and a processing technique, and does not
depend on any algebraic phase relation.
[0085] The method described below is based on the principle that,
if harmonic noise on one shot appears at a different time from
harmonic noise on another shot, then appropriate stacking can
eliminate or substantially reduce the noise. The stacking method
that is used is time-frequency diversity stacking. This stacking
method works well when the signal component of the two records is
the same, but the noise appears in different locations in the
time-frequency domain.
[0086] To make harmonic noise appear at different times in
different shots, one method is to vary the delay between shots. The
chosen time difference between a shot in one location and the
subsequent shot (at another location) should differ from the time
delay between another shot made in the first location and
subsequent shot thereto such that, after time-frequency
decomposition, the largest peaks in the noise do not substantially
overlay one another. The exact time delay that is required will
depend on the sweep rate, but it has been found that a difference
of between 1 and 2 seconds is generally sufficient if the sweep
time is less than 10 seconds.
[0087] In a variation of this technique the effect of the time
delay between shots is mimicked using multiple vibrator groups
where the noise is arranged to come from shots from different,
widely separated vibrator groups. The spatial separation between
the vibrator groups will induce a time delay so that, for most
receivers, harmonic noise will arrive at different times and
time-frequency diversity stacking will be effective at removing the
harmonic noise. For some receivers, the peaks in the noise from
different shots will overly one another, but this noise can be
removed during stacking using the technique disclosed in UK patent
application No 2 359 363.
[0088] In the case of N vibrators, it is necessary to make at least
(N+1) shots of encoded sweeps, from which the n individual shot
records can be extracted. For example, consider two groups of two
vibrators each with a sweep time S of 8 seconds and a listening
time L of 5 seconds. The fastest that these vibrators can perform
three shots for each group is as follows:
TABLE-US-00001 TABLE 1 Group 1 shot 1 starts 0 seconds Group 2 shot
1 starts 5 seconds 5 second time slip Group 1 shot 1 ends 8 seconds
Group 1 shot 2 starts 10 seconds Group 2 shot 1 ends 13 seconds 5
second slip time Group 2 shot 2 starts 15 seconds 5 second slip
time Group 1 shot 2 ends 18 seconds Group 1 shot 3 starts 20
seconds Group 2 shot 2 ends 23 seconds 5 second slip time Group 2
shot 3 starts 25 seconds 5 second slip time Group 1 shot 3 ends 28
seconds Group 2 shot 3 ends 33 seconds 5 second slip time
[0089] Adding delays between the shots made by the first group and
the shot that follows each shot of the first group enables the
harmonic noise to be placed at different times in the shot records.
For example, the above shot scheme can be modified by increasing
the delay between a shot of the first group and the corresponding
shot of the second group, as follows:
TABLE-US-00002 TABLE 2 Group 1 shot 1 starts 0 seconds Group 2 shot
1 starts 5 seconds 5 second time slip Group 1 shot 1 ends 8 seconds
Group 1 shot 2 starts 10 seconds Group 2 shot 1 ends 13 seconds 5
second slip time Group 2 shot 2 starts 16 seconds 6 second slip
time Group 1 shot 2 ends 18 seconds Group 1 shot 3 starts 22
seconds Group 2 shot 2 ends 24 seconds 6 second slip time Group 2
shot 3 starts 29 seconds 7 second slip time Group 1 shot 3 ends 30
seconds Group 2 shot 3 ends 37 seconds 7 second slip time
[0090] In this modified shot scheme of Table 2, the varying delay
between a shot of the first group and the corresponding shot of the
second group means that harmonic noise will appear at different
times in the shot records. It is therefore possible to eliminate
the harmonic noise using the "diversity simultaneous inversion
technique" disclosed in GB 2 359 363.
[0091] FIG. 7 illustrates the results of applying the diversity
simultaneous inversion technique to the data shown in FIG. 5. It
will again be seen that a considerable reduction in harmonic noise
has been provided.
[0092] In the shot scheme of table 2, the time delay between a shot
of the first vibrator group and the corresponding shot of the
second vibrator group increases with the shot number. The invention
is not limited to this, and the time delay between a shot of the
first vibrator group and the corresponding shot of the second
vibrator group could alternatively decrease with the shot
number.
[0093] The technique of diversity simultaneous inversion may be
combined with harmonic estimation and removal for those shots where
the time delay satisfies equation (1) above. For example, if the
vibrator sweep is from 10 Hz to 100 Hz, then the second harmonic
can be estimated and removed if the slip time is greater than 4.5
seconds, the third harmonic can be estimated and removed if the
slip time is more than 6 seconds, and the fourth harmonic can be
removed if the slip time is more than 6.7 seconds. Thus, for the
acquisition scheme of Table 2, the second harmonic may be estimated
and removed from the harmonic noise in all shots, the third
harmonic may be estimated and removed from the noise in the second
and third shot of each group, and the fourth harmonic may be
estimated and removed from the noise in the third shot of each
group.
[0094] FIG. 8 illustrates the results of applying both the harmonic
estimation and removal technique and the diversity simultaneous
inversion technique. FIG. 8 shows the results of carrying out the
diversity simultaneous inversion technique to the data shown in
FIG. 6--which has already undergone a harmonic estimation and
removal technique for the second and third harmonics. It will be
seen that subsequent application of the diversity simultaneous
inversion technique has resulted in further reduction in noise.
[0095] The manner in which the above techniques are used depends on
factors such as the number of groups, the time required to move a
vibrator from one desired shooting position to another, and the
ratio of the total vibrator sweep time to the listen time.
[0096] If the listen time is equal to the total sweep time divided
by the number of vibrators in a group, then an efficient
implementation is to have two groups. At any one time, one group
can be being re-positioned while the other group is acquiring data.
The group that is acquiring data shoots the same number of shots as
there are vibrators in the group. Each vibrator sweeps nearly
continuously, with only a short pause at the end of each sweep to
reset the equipment. When the first group has finished shooting,
the second group is in position, and begins shooting while the
first group moves up to the next location. The slip time is 100% of
the sweep time, so harmonic estimation and removal may be applied
to the acquired data. However, as the number of shots is equal to
the number of vibrators in each group, diversity-simultaneous
inversion is not possible.
[0097] If, in another example, the sweep time for each vibrator is
around twice the listening time, then a method using three groups
can usefully be employed. In the case of nine vibrators, it will be
possible to have three groups of three vibrators, with vibrators in
each group being actuated four time with varying delays between
shots. The shooting pattern would be arranged so that the first two
shots of each group alternate with the last two shots of the
preceding group and so that the third and fourth shots of each
group alternate with the first two shots of the subsequent group.
For those shots where equation (1) is satisfied, harmonic noise can
be estimated and removed. Diversity simultaneous inversion may then
be applied to the acquired data for each group.
[0098] FIG. 9(a) is a schematic illustration of a seismic surveying
arrangement according to an embodiment of the invention. The
seismic surveying arrangement comprises a plurality of vibrators
1,2,3,4 and a control means 5. Two groups of vibrators A, B are
defined in FIG. 9(a), with each group containing two vibrators in
each group, but the seismic surveying arrangement of the invention
is not limited to this particular number of groups or to this
number of vibrators in a group. The control means 5 is able to
actuate each group independently of the other group. For example,
the control means may be electrically connected to each group of
vibrators so that it can send an electrical signal to a selected
group to actuate each vibrator in the selected group. The control
means is adapted to actuate the vibrators according to, for
example, a "simultaneous slip-sweep" method of the type described
with reference to FIG. 3 or a "varying time delay" method of the
type described with reference to Table 2.
[0099] Four vibrators are shown in FIG. 9(a), arranged into two
groups each containing two vibrators but a seismic surveying
arrangement the invention is not limited to these numbers of
vibrators and groups. A seismic surveying arrangement of the
invention for use with the "simultaneous slip-sweep" method may
contain two or more independently actuable groups of vibrators,
with at least one group containing more than one vibrator. A
seismic surveying arrangement of the invention for use with the
"varying time delay" method may contain two or more independently
actuable groups of vibrators, or alternatively may contain two or
more independently actuable vibrators.
[0100] The seismic surveying arrangement of FIG. 9(a) further
comprises an array of one or more seismic receivers (two receivers
6,7 are shown in FIG. 9(a), but the seismic surveying arrangement
is not limited to two receivers).
[0101] FIG. 9(b) is a schematic block diagram of the control means
5. The control means comprises a programmable data processor 8 with
a program memory 9, for instance in the form of a read only memory
ROM, storing a program for controlling the control means 5 to
actuate the vibrators 1,2,3,4 according to, for example a method as
illustrated in FIG. 3 or Table 2 or as defined by equation (1)
above. The system further comprises non-volatile read/write memory
10 for storing, for example, any data which must be retained in the
absence of power supply. A "working" or "scratchpad" memory for the
data processor is provided by a random access memory (RAM) 11. An
input interface 12 is provided, for instance for receiving commands
and data. An output interface 13 is provided, for instance for
outputting actuation signals to a selected receiver or to a
selected group of receivers. A program defining the actuation
sequence of the receivers or receiver groups may be supplied via
the input interface 12 or may alternatively be provided by a
machine-readable store 14.
[0102] The program for operating the control means and for
performing the method described hereinbefore is stored in the
program memory 9, which may be embodied as a semi-conductor memory,
for instance of the well-known ROM type. However, the program may
be stored in any other suitable storage medium, such as magnetic
data carrier 9a (such as a "floppy disc") or CD-ROM 9b.
[0103] The present invention is applicable to both land-based
vibrator sources of seismic energy and the marine vibrator sources
of seismic energy.
[0104] While the invention has been described in conjunction with
the exemplary embodiments described above, many equivalent
modifications and variations will be apparent to those skilled in
the art when given this disclosure. Accordingly, the exemplary
embodiments of the invention set forth above are considered to be
illustrative and not limiting. Various changes to the described
embodiments may be made without departing from the spirit and scope
of the invention.
* * * * *