Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
1999-09-14
2001-02-13
McElheny, Jr., Donald E. (Department: 2862)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
Reexamination Certificate
active
06188964
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention comprises methods for generating surface-consistent residual statics. In particular, the invention relates to methods for calculating static corrections for seismic data by minimizing an objective function using global optimization thereby to maximize a corresponding stack power. The stack power is decoupled from the overlapping between different seismic traces into several one-dimensional problems. The surface-consistent residual statics that minimize the objective function are determined by using a Stochastic Pijavskij Tunneling technique to eliminate regions where the global minimum of the objective function is unlikely to exist so that the global minimum of the objective function can be reached without determining the objective function in these eliminated regions.
2. Background Art
The Problem
Seismic exploration is a predominant geophysical activity conducted to find commercial accumulations of oil, gas, or other minerals, to study the nature of the Earth for the foundations of roads, buildings, dams, tunnels, nuclear power plants, and other structures, and to search for geothermal areas, water resources, archeological ruins, etc.
In seismic exploration, seismic waves or signals are generated by shots from one of several types of energy sources and detected by arrays of sensitive devices or receivers called geophones or hydrophones. The most common measurement made is of the travel times of seismic waves, although attention has been increasingly directed to the amplitude of seismic waves or changes in their frequency content or wave shape.
The measurement of seismic signals may be seriously interfered by external factors such as wind, vehicular and pedestrian traffic, microseisms and at sea, ship noise and environmental noise due to marine life. Moreover, solid friction in the Earth as well as seismic waves scattering attenuate the measured signals. Thus, a plurality of traces from different shots resulting from the interaction of a plurality of cooperating sources and receivers, but having the same subsurface incident point, are gathered and combined together by use of various algorithms well known to those skilled in the geophysical art. Multiple coverage tends to destructively attenuate random noise but enhance coherent reflection data. One such method is referred to as a Common Mid Point (“CMP”) stacking which will now be illustrated with reference to FIG.
1
.
In
FIG. 1
, sound waves generated by acoustic sources S
1
-S
3
located on surface respectively are detected by receivers R
1
-R
3
, after sound wave reflection from interface
14
along ray paths
20
and
30
,
22
and
28
,
24
and
26
. The number of the energy sources can be different from that of the receivers. The ray paths shown all converge at a common mid point
16
. The surface separation between the respective sources and correspondingly numbered receivers is termed the offset. Correction for angularity (also called Normal Moveout or “NMO”) to the respective ray paths by stacking them together would collapse them to zero-offset as represented on the surface by S
0
/R
0
to form two-way travel path
18
impinging on the common mid point
16
. In this case of zero dip, mid point
16
also forms a common depth point as well. Because all of the ray paths are incident on the same subsurface point, here CMP
16
, the traces can be stacked validly to enhance the signal-to-noise ratio.
Near-surface lateral velocity variations and surface elevation changes create travel-time variations that may be approximated by surface-consistent static time shifts. In
FIG. 1
there are shown two subsurface earth layers I and II, separated by line
12
, characterized respectively by acoustic propagation velocities of V
I
=1500 meters per second (m/s) and V
II
=3100 m/s. P
si
and P
rj
identify those portions of the near-vertical ray paths that traverse the variable-thickness upper low velocity layer I, after ray bending due to Snell's law of refraction at the interface
12
, beneath sources S
i
and receivers R
j
. Because of a longer combined path length of P
s3
+P
r3
through the low velocity layer I, the total travel time along ray path
20
,
30
from S
3
to R
3
, after application of NMO, will still be longer than the total travel time along either ray path
22
,
28
from S
2
to R
2
or ray path
24
,
26
from S
1
to R
1
, corrected for NMO, where the combined ray path P
s2
+P
r2
or P
s1
+P
r1
through the low velocity layer I is shorter. Note that
FIG. 1
gives an example where the velocity in the low velocity Layer I has a constant volume. In reality, this low velocity Layer I, which is usually 4 to 50 meter thick, is characterized by seismic velocities that are not only low (usually between 200 and 2,000 m/s), but at times highly variable. Proper statistical methods need to be used to estimate the low velocity layer. It is known in the art, therefore, that even though the low-velocity segments of the total ray path are relatively short, travel-time differences are not negligible due to the very low velocity at or near the surface I. The respective time delays s
i
and r
j
due to a variable-thickness low velocity layer, here the surface layer I, are defined as the surface-consistent statics which must be applied as corrections to the reflection travel times prior to stacking for maximizing inter-trace reflection coherency. The statics are termed surface-consistent because they are due to irregularities of the near-surface low velocity layer.
In the example of
FIG. 1
, only a few of sources and receivers are shown for simplicity. The relative static corrections can be easily obtained for this simple, constant velocity case. In reality, calculating surface-consistent residual statics correction to compensate for time shifts in stacked many thousands of seismic traces, many of indifferent or poor quality, recorded from interfacing between up to thirty cooperating seismic sources and receivers poses a challenge to the art.
The Prior Art
Since the earliest days of seismic exploration, geophysicists have recognized the need to correct for the low velocity in the weathered and unconsolidated sediments near the earth's surface. The data processing procedure has been described by Yilmaz, in a book entitled “Seismic Data Processing,” published by Society of Exploration Geophysics, Tulsa, Okla., 1987, Marsden, in Static Corrections—a Review, The Leading Edge, 12, Part 1, pp. 43-49, Part 2, pp. 115-120, and Part 3, pp. 210-216, 1993, and Sheriff and Geldart, in Exploration Seismology, Second Edition, Cambridge University Press, Cambridge, England, 1995. The first corrections for elevation and low velocity are field statics. A reference level is determined that is below the low velocity layer (LVL) and field statics move the sources and receivers to the reference level. CMP gathers are used to generate a set of preliminary velocity picks that are used to calculate NMO corrections. Residual statics corrections are calculated using the corrected data. The process is repeated until the results converge.
The conventional method for calculating residual statics corrections was developed by Taner, et al., in “Estimation and correction of near-surface time anomalies,” Geophysics, 39, pp. 441-463, 1974 and Wiggins, et al., in “Residual statics analysis as a general linear inverse problem,” Geophysics, 41, pp. 922-938, 1976. The time delays caused by the passage of seismic signals through the LVL should depend on path. After the NMO corrections, it is assumed that all of the paths are vertical and estimate a single time delay that is “surface consistent” (each source and receiver location has a time delay that does not depend on the wave path). The time delay, denoted by t
srk
, for a trace that follows a path from a source (s) to a receiver (r) via a common midpoint (k) is determined by maximizing the cross correlation between the trace and the CMP gather. The total time delay has four components: the source
Barhen Jacob
Oblow Edward M.
Reister David B.
McElheny Jr. Donald E.
Needle & Rosenberg P.C.
UT-Battelle LLC
LandOfFree
Method for using global optimization to the estimation of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for using global optimization to the estimation of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for using global optimization to the estimation of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2598319