Removing irregularities from seismic data caused by tube waves
Method of removing irregularities caused by tube waves from seismic data obtained in a borehole seismic survey, using a set of seismic or sources and a set of seismic receivers each arranged in a borehole, in which seismic data is time shifted based on an estimated tube wave velocity, filtered and then the time shift is reversed and applied to the original data set resulting in a corrected set, which is again time shifted opposite to the original time shift to remove and identify tube wave irregularities.
1. A method of removing irregularities caused by tube waves from seismic data obtained in a borehole seismic survey, using a first set of transducers (seismic receivers or sources) and a second set of transducers (seismic sources or receivers) and wherein at least the transducers of the first set are arranged in a borehole, which method comprises the steps of
(a) obtaining a set of original seismic data p m (t,k), wherein the integer k refers to the k'th transducer of the first set of transducers and the integer m refers to the m'th transducer of the second set of transducers;
(b) subjecting the original seismic data set to a time-shift based on an estimated tube wave velocity to align the tube waves in time so that a set of time-shifted data {tilde over (p)} m (t,k) is obtained;
(c) applying a filter to the time-shifted seismic data set {tilde over (p)} m (t,k) to obtain a set of time-shifted, predicted tube wave data {tilde over (q)} m (t,k);
(d) undoing the time shift for the time-shifted, predicted tube wave data set {tilde over (q)} m (t,k) to obtain a set of predicted tube wave data q m (t,k);
(e) subtracting the predicted tube wave data set q m (t,k) from the original seismic data set p m (t,k) to obtain a set of corrected seismic data p c m (t,k), and the original seismic data set by the corrected seismic data set;
(f) subjecting the set of corrected, original seismic data to a time-shift based on a tube wave velocity opposite to the velocity used in step (b) to align the tube waves in time so that a set of time-shifted data {tilde over (p)} m (t,k) is obtained; and
(g) repeating steps (c), (d) and (e) to obtain a set of seismic data of which irregularities caused by tube waves have been removed, characterized in that the filter applied in step (c) is a structurally oriented edge-preserving filter.
2. The method according to claim 1 , wherein step (c) comprises applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} m (t,k) to obtain a set of time-shifted, predicted tube wave data {tilde over (q)} m (t,k), correlating the set of time-shifted, predicted tube wave data with the time-shifted seismic data set, and removing the less well correlated elements from the set of time-shifted, predicted tube wave data.
3. The method according to claim 2 , wherein correlating the set of time-shifted, predicted tube wave data with the time-shifted seismic data set, and removing the less well correlated elements from the set of time-shifted, predicted tube wave data comprises determining a correlation measure and if the correlation measure is less than a predetermined minimum value, the time-shifted, predicted tube wave data are made equal to zero.
4. The method according to claim 3 , wherein the correlation measure is
C
(
t
n
,
j
)
=
2
∑
l
=
-
n
t
n
t
p
~
(
t
n
+
l
Δ
t
,
j
)
·
q
~
(
t
n
+
l
Δ
t
,
j
)
∑
l
=
-
n
t
n
t
{
p
~
2
(
t
n
+
l
Δ
t
,
j
)
+
q
~
2
(
t
n
+
l
Δ
t
,
j
)
}
.
5. The method according to claim 2 , wherein also the transducers of the second set of transducers are located in a borehole, which differs from the borehole in which the transducers of the first set are located, which method further comprises the steps of
(h) subjecting the set of seismic data obtained in step (i) to a time-shift to based on an estimated tube wave velocity to align the tube waves in time so that a set of time-shifted data {tilde over (p)} n (t,l) is obtained, wherein the integer l refers to the l'th transducer of the second set and the integer n refers to the n'th transducer of the first set;
(i) applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} n (t,l) to obtain a set of time-shifted, predicted tube wave data {tilde over (q)} n (t,l);
(j) undoing the time shift for the time-shifted, predicted tube wave data set {tilde over (q)} n (t,l) to obtain a set of predicted tube wave data q n (t,l);
(k) subtracting the predicted tube wave data set q n (t,l) from the original seismic data set p n (t,l) to obtain a set of corrected seismic data p c n (t,l), and replacing the original seismic data by the corrected seismic data;
l) subjecting the set of corrected, original seismic data to a time-shift based on a tube wave velocity opposite to the velocity used in step (h) to align the tube waves in time so that a set of time-shifted data {tilde over (p)} n (t,l) is obtained, and repeating steps (i), (j) and (k);
(m) repeating steps (i) through (l) for each transducer l of the first set of transducers and for each transducer n of the second set of transducers to obtain a set of seismic data of which irregularities caused by tube waves have been removed.
6. The method according to claim 2 , wherein applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} i (t,j) to obtain a set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) comprises the steps of:
1) determining for each element of the data set {tilde over (p)} i (t,j) the local orientation of the time-shifted seismic data, which local orientation is the orientation of a plane tangent to the time-shifted seismic data for that element;
2) determining for each element whether there is an edge in its neighbourhood; and
3) carrying out an averaging operation for each element of the data {tilde over (p)} i (t,j), wherein the direction of the averaging operation is the local orientation of the data and wherein the averaging operation does not go over the edge, to obtain the set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j).
7. The method according to claim 2 , wherein subjecting a data set p i (t,j) to a time-shift to align the tube waves in time is done in accordance with the following equation {tilde over (p)} i (t,j)=p i (t±z j /v tube ,j), wherein z j is the depth along the borehole of transducer j, and wherein v tube is the magnitude of the tube wave velocity, and wherein undoing the time shift for a time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) is done in accordance with the following equation q i (t,j)={tilde over (q)} i (t∓z j /v tube ,j).
8. The method according to claim 1 , wherein also the transducers of the second set of transducers are located in a borehole, which differs from the borehole in which the transducers of the first set are located, which method further comprises the steps of
(h) subjecting the set of seismic data obtained in step (i) to a time-shift to based on an estimated tube wave velocity to align the tube waves in time so that a set of time-shifted data {tilde over (p)} n (t,l) is obtained, wherein the integer l refers to the l'th transducer of the second set and the integer n refers to the n'th transducer of the first set;
(i) applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} n (t,l) to obtain a set of time-shifted, predicted tube wave data {tilde over (q)} n (t,l);
(j) undoing the time shift for the time-shifted, predicted tube wave data set {tilde over (q)} n (t,l) to obtain a set of predicted tube wave data q n (t,l);
(k) subtracting the predicted tube wave data set q n (t,l) from the original seismic data set p n (t,l) to obtain a set of corrected seismic data p c n (t,l), and replacing the original seismic data by the corrected seismic data;
(l) subjecting the set of corrected, original seismic data to a time-shift based on a tube wave velocity opposite to the velocity used in step (h) to align the tube waves in time so that a set of time-shifted data {tilde over (p)} n (t,l) is obtained, and repeating steps (i), (j) and (k);
(m) repeating steps (i) through (l) for each transducer l of the first set of transducers and for each transducer n of the second set of transducers to obtain a set of seismic data of which irregularities caused by tube waves have been removed.
9. The method according to claim 8 , wherein step (i) comprises applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} n (t,l) to obtain a set of time-shifted, predicted tube wave data {tilde over (q)} n (t,l), correlating the set of time-shifted, predicted tube wave data with the time-shifted seismic set, and removing the less well correlated elements from the set of time-shifted, predicted tube wave data.
10. The method according to claim 9 , wherein applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} i (t,j) to obtain a set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) comprises the steps of:
1) determining for each element of the data set {tilde over (p)} i (t,j) the local orientation of the time-shifted seismic data, which local orientation is the orientation of a plane tangent to the time-shifted seismic data for that element;
2) determining for each element whether there is an edge in its neighbourhood; and
3) carrying out an averaging operation for each element of the data set {tilde over (p)} i (t,j), wherein the direction of the averaging operation is the local orientation of the data and wherein the averaging operation does not go over the edge, to obtain the set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j).
11. The method according to claim 9 , wherein correlating the set of time-shifted, predicted tube wave data with the time-shifted seismic data set, and removing the less well correlated elements from the set of time-shifted, predicted tube wave data comprises determining a correlation measure and if the correlation measure is less than a predetermined minimum value, the time-shifted, predicted tube wave data are made equal to zero.
12. The method according to claim 9 , wherein subjecting a data set p i (t,j) to a time-shift to align the tube waves in time is done in accordance with the following equation {tilde over (p)} i (t,j)=p i (t±z j /v tube ,j), wherein z j is the depth along the borehole of transducer j, and wherein v tube is the magnitude of the tube wave velocity, and wherein undoing the time shift for a time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) is done in accordance with the following equation q i (t,j)={tilde over (q)} i (t∓z j /v tube ,j).
13. The method according to claim 8 , wherein applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} i (t,j) to obtain a set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) comprises the steps of:
1) determining for each element of the data set {tilde over (p)} i (t,j) the local orientation of the time-shifted seismic data, which local orientation is the orientation of a plane tangent to the time-shifted seismic data for that element;
2) determining for each element whether there is an edge in its neighbourhood; and
3) carrying out an averaging operation for each element of the data set {tilde over (p)} i (t,j), wherein the direction of the averaging operation is the local orientation of the data and wherein the averaging operation does not go over the edge, to obtain the set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j).
14. The method according to claim 8 , wherein subjecting a data set p i (t,j) to a time-shift to align the tube waves in time is done in accordance with the following equation {tilde over (p)} i (t,j)=p i (t±z j / v tube ,j), wherein z j is the depth along the borehole of transducer j, and wherein v tube is the magnitude of the tube wave velocity, and wherein undoing the time shift for a time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) is done in accordance with the following equation q i (t,j)={tilde over (q)} i (t∓z j /v tube ,j).
15. The method according to claim 1 , wherein applying a structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} i (t,j) to obtain a set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) comprises the steps of:
1) determining for each element of the data set {tilde over (p)} i (t,j) the local orientation of the time-shifted seismic data, which local orientation is the orientation of a plane tangent to the time-shifted seismic data for that element;
2) determining for each element whether there is an edge in its neighbourhood; and
3) carrying out an averaging operation for each element of the data set {tilde over (p)} i (t,j), wherein the direction of the averaging operation is the local orientation of the data and wherein the averaging operation does not go over the edge, to obtain the set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j).
16. The method according to claim 15 , wherein applying the structurally oriented, edge-preserving filter to the time-shifted seismic data set {tilde over (p)} i (t,j) to obtain the set of time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) comprises the steps of:
(1) defining a window of n L +n R +1 transducers centred on transducer j and 2n t +1 time samples centred on t n =t 0 +nΔt, wherein t 0 is the time at which sampling starts;
(2) calculating a semblance measure in accordance with the following equation
S
0
(
t
n
,
j
)
=
∑
l
=
-
n
t
n
t
(
∑
k
=
-
n
L
n
R
p
~
(
t
n
+
l
Δ
t
,
j
+
k
)
)
2
(
n
L
+
n
R
+
1
)
∑
l
=
-
n
t
n
t
∑
k
=
-
n
L
n
R
p
~
2
(
t
n
+
l
Δ
t
,
j
+
k
)
,
further semblance measures S + and S − wherein the data used to calculate the two further semblance measures is {tilde over (p)}(t±αz k ,j+k), wherein α is a slowness parameter, and wherein z j is the distance along the borehole between transducers j and k and determining the slowness parameter α max that pertains to the maximum of S + , S 0 and S − , and calculating time-shifted tube wave data by taking the median of {tilde over (p)}(t+α max z j ,j 0 +k), wherein −n L ≦k≦n R ;
(3) repeating step (1) for a left range (L), wherein n L =n k and n R =0, a central range (C), wherein n L =n k and n R =n k and for a right range (R), wherein n L =0 and n R =n k to obtain S L max and {tilde over (q)} L , S C max and {tilde over (q)} C , and S R max and {tilde over (q)} R , respectively, wherein n k is a predetermined integer; and
(4) selecting the time-shifted, predicted tube wave data as follows, if S C max >ε.max(S L max , S R max ) then the time-shifted, predicted tube wave data are {tilde over (q)} C , else, if S R max >S L max then the time-shifted, predicted tube wave data are {tilde over (q)} R , else the time-shifted, predicted tube wave data are {tilde over (q)} L , wherein ε is a predetermined constant.
17. The method according to claim 16 , wherein if the semblance measure selected in step (4) is less than a minimum value the time-shifted, predicted tube wave data are made equal to zero.
18. The method according to claim 16 , wherein subjecting a data set p i (t,j) to a time-shift to align the tube waves in time is done in accordance with the following equation {tilde over (p)} i (t,j)=p i (t±z j /v tube ,j), wherein z j is the depth along the borehole of transducer j, and wherein v tube is the magnitude of the tube wave velocity, and wherein undoing the time shift for a time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) is done in accordance with the following equation q i (t,j)={tilde over (q)} i (t∓z j /v tube ,j).
19. The method according to claim 10 , wherein subjecting a data set p i (t,j) to a time-shift to align the tube waves in time is done in accordance with the following equation {tilde over (p)} i (t,j)=p i (t±z j /v tube ,j), wherein is the depth along the borehole of transducer j, and wherein v tube is the magnitude of the tube wave velocity, and wherein undoing the time shift for a time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) is done in accordance with the following equation q i (t,j)={tilde over (q)} i (t∓z j /v tube ,j).
20. The method according to claim 1 , wherein subjecting a data set p i (t,j) to a time-shift to align the tube waves in time is done in accordance with the following equation {tilde over (p)} i (t,j)=p i (t±z j /v tube ,j), wherein z j is the depth along the borehole of transducer j, and wherein v tube is the magnitude of the tube wave velocity, and wherein undoing the time shift for a time-shifted, predicted tube wave data set {tilde over (q)} i (t,j) is done in accordance with the following equation q i (t,j)={tilde over (q)} i (t∓z j /v tube ,j).