Sie sind auf Seite 1von 5

425

Seismic 4
Statics determined

from near-surFace

third assumption needs a more critical examination

for 3-D

data since midpoint spread may be dppreciable,

ics techniques alone in that the models are largely determin-

when traces are gathered in common midpoint (CMP) bins of

istic. The interpretive

significant lateral extent.

evaluation

leads to geologic credibility

Downloaded 12/01/15 to 132.239.1.230. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

models based on

refraction analyses offer several advantages to residual stat-

Attention

of the near-surface

model

of the statics to be applied.

to the details of the models generated at line tic

positions and subsurface control points allow\

a consistent

examples

of the application

The first step of this method is obtaining a time shift for


each trace by crosscorrelation

of the described

technique are presented. These examples show the validity


circumstances.

s4.4

3-D Residual Statics

A. R. Mosrr, Atw Oil d Gns: awl D. B. Jovcrtrovicir.Grr(t


R&D
of surface-consistent

residual

statics

is described in this paper. The method

employed is similar to 2-D statics procedures utilizing reflection time delays in CDP gathered data. We first review the
basic assumptions
a CDP

and equations
reference

of the

frame.

Next.

tween 2-D and 3-D statics are outlined.


the 3-D statics program

statics

method

differences

be-

implementation

of

ih then described.

The

II)

t (C = I, ttz), D, is the

offset for trace t, R, i\ the static for receiver

normalized

The determination

tract.

7, = R,,,, + ,),(I, - cai,, + D,M~ I,).


where l( is the time delay for trace

for 3-D surveys

a model

solution for shot-and-

four components:

of the method as well as its robustness in varied geologic

employing

with

second step is the curface-consistent

receiver statics. Each time delay i\ modeled a\ the sum of

model of the near-surface of an area to be developed.


Several

particularly

position i (i = I, nRl, S, is the static for shot position,; u = I,


ns), CL is a structure parameter associated with midpoint

tt,w),and ilfl is a residual normal moveout

index X (X = I.
parameter

(KNMO).

accounting

for velocity

errors,

also

depending upon midpoint index X. There are m equations in


II = llfi L 115 + 2n&# unknowns,
receiver

and midpoint

overdetermined.

so with the redundancy

positions,

However.

of

the set of equations

as demonstrated

is

by Wiggins et

al. (1976) for linear surveys, the problem is ill-conditioned.


The set of equations
minimization

can be inverted

by a least-squares

of:

A test of the

procedure is next shown utilizing simulated data corresponding to a crossed array


statics.

structure,

3-D survey.

and residual

For thi:, case, known

normal

terms were put in the data. Parameters


program demonstrate

moveout

(NMO)

determined

bb the

its ability to extract idealized statics.

Finally, a case study is presented showing the successful use

of this procedure on real data. The object of this 3-D jurvcq


was to delineate

thin,

meandering

thickness over the survey


data

region.

was based on the tuning

sands which
Interpretation

effect

which

change
of these

causes the

The

last four

terms pick

a unique

solution

for this ill-

amplifudes of reflections from these sands to change propor-

conditioned problem with minimum norm. The normal equa-

tionally

tions resulting from minimization

to their thickness.

Hence,

essential in this interpretation

static corrections

since degradation

were

of pulse

the Gauss-Seidel

offare

usually inverted by

method, yielding:

amplitude and phase due to statics would have obscured the


subtle wavelet changes being sought.

Residualstaticsutilizing reflectiontime delaysin CDP


gathered traces
The purpose of static corrections IS to remove trace time
delays resulting from the transit of seismic energy through
the variable,

near-surface

weathered

tions are, of course. as important

layer.

R,tlI)

These correc-

for land 3-D data as for

(3)

linear seismic surveys. The method we discuss for 3-D static


corrections
utilizing

is based on the same model as for 2-D statics

reflection

time delays from CDP gathered

traces.

This has been thoroughly discussed by Hileman et al. (196X),


Taner et al. (1974). Wiggins et al. (1976). and others. We
present a brief review here so that special considerations for
3-D data may be pointed out.
The

major assumptions

surface effects introduce

for this method

are: (I)

near-

only static time delays: (2) each

surface location has an associated fixed time delay regardless of the wave path (surface consistency): and (3) ail traces
in a given CDP gather exhibit the same subsurface structure.
The first two assumptions have to do with the velocity in the

for the parameter

near-surface

This

numbers of traces associated with the corresponding param-

has no dependence on whether the data are 2-D or 3-D. The

eter; sums are over all traces associated with the parameter

being much less than in layers beneath.

iterates. Here IV,.

N:.

and I%,, are the

426

Seismic 4

being evaluated (denoted as l E C&where. for example.


the set of traces for parameter

cb,is

R,) and 1 is the iteration

number.

trace is used in the partial stack tor the next tnrce. and YOon.
This procedure is iterated a specrficd number of times. The

ofa master

other option is the formation

trace from the stack

of adjacent common midpoint bins. Partial stacking is u$cd

Downloaded 12/01/15 to 132.239.1.230. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Special considerationsfor 3-D statics

to align traces in adjacent bins and the stacks from the\e

The above method requires no distinction


and 3-D data; however,
necessary
rather
grid.

2-D

bins. The level of adjacent bins radiating outward from the

some special considerations

are

central bin is an adjustable parameter.

for its application

than forming

midpoint

between

CDP

to three dimensions.

gathers along

bins are constructed

increase fold.

parameters.

to regularize

and reduce

the midpoint

the number

Binning may put components

Other options in the time shift pt-ogram are the crosacorre-

First.

line. common-

of midpoint

into trace shifts

lation time window


and midpoint

of time delay data may result from the azi-

muthal dependence

of moveout.

The form of these time

shifts is

The

file consisting of shot, receiver,

indices: offset

normalized

cros\correlation;

and time shift for each trace.


A second program reads the file of time shifts and solves

arising from subsurface structure within bins. An additional


contamination

and the range of shifts examined.

output of this program is

for surface-consistent

statics. An etficient way of performing

the Gauss-Seidel itrrative

procedure i5 to create three copies

of the trace file sorted

by shot,

receiver,

and midpoint

indices. These indices are made sequential \o they may be


(4)

used as pointers in parameter arrays. As can be seen from


equation (3), having sorted files makes the calculation
parameter iterates a simple vector operation.

taking the shot parameters. all trace records associated with

, _ Di sin H

w =

co? (a, -

4)

the first shot are read in sequence from the file sorted by shot

4zo

index. These records have time delays and pointers for the
corresponding

I
t ~(sin

Hr,

Yc sin ~5.

the reflector depth at the bin center.

algorithm then reads the next \et of sequential records for


Z. is

(0, 4) are the dip and

strike (measured with respect to v-direction)

of the reflector.

Di is the trace otTset. LY(is the offset azimuth, and (x,, .v,) are
of the distance from the bin center to the

midpoint for trace P.


Possibilities for treating these time shifts in 3-D statics are:
(1) time shifts can be removed

using interpretative

data

obtained prior to statics: (2) reflector parameters can be fit in


the time delay model: and (3) (the method we chose) bin size
window can be chosen to make these

shifts negligible (of course, the data must cooperate!).

This

can be done by limiting offsets. picking an appropriate


Another

adjacent

CMP

is the formation

of a model

bins are used, they must be

tangent in a 2-D sense requiring two indices for labeling the


midpoint

index.

Another

reason for 2-D

labeling of mid-

points is smoothing of the structure and RNMO

terms.

is the volume of data which must be processed. 3-D surveys


2-D survey
receivers

with

110 ft shot-and-receiver

a IO mile.

spacing and 4X

per shot has about 23 000 traces. Thus.

consideration

algorithm advances to the next parameter


receivers.

a key

in 3-D statics is efficiency in both storage and

time

This program has many options. Included are elimination


of trace records based on time shift and crosscorrelation.
elimination

of parameters

order of parameter

are two options for forming

model traces.

One is

These programs were tested on a simulated crossed array


survey depicted in Figure
crossed arrays
200 ft.

Midpoint

is

stored and the trace is shifted by this amount. The shifted

and 31 shots each. Array

binning

spacing was

was chosen to be 3-by-3

true

midpoints, There are a total of I I 025 traces. The number of


are 205 receiver positions. 205 shot posi-

tions. 441 structure terms, and 441 RNMO


of

terms for a total

I 391 parameters. Random. but surface-consistent.

and-receiver

statics uniformly

distributed

between

shot-

532 ms

were applied to traces consisting of a Ricker wavelet (20 HZ


central frequency)

at a fixed reflection time plus 20 percent

Gaussian noise. In addition. random structure terms from a


between

~16 ms and constant

I6 ms

terms were placed in the data. This type of data set is


useful in testing statics since the statics are known

and the method assumptions are exactly

satisfied.

Figures lb through le show traces for the bin outlined in


Figure

crosscorrelation

la. This survey consisted of 25

of 21 receivers

centers were 1000 ft apart and shot-and-receiver

structure,

to maximum

statics which may be

Test of 3-D statics programs

partial stacking in which each trace in a gather is crosscorretime shift corresponding

of weighting

The final result of this

program is a set of shot-and-receiver

Iated with the stack of the other traces in the gather.

The

and variation

number.

applied to the trace\.

extremely

The first stage in 3-D statics is obtaining trace delays.

based on fold, selection of the

calculation,

coefficients with iteration

RNMO

There

and read5 from the trace file sorted by receiver

uniform distribution

Implementation of 3-D statics

the

set. for example.

index.

static parameters

The most important distinction of 3-D as contrasted to 2-D


often have 100 000 traces or more. Comparatively,

S?, and 40 on. After all shot parameters are calculated.

bin

on a deep, flat event.

3-D consideration

trace. When

parameters.

are formedand

followed by the sum and division needed to calculate Sr. The

rn is the reflection time at the bin center (To = 2Z&),

size, and crosscorrelating

structure, and RNMO

R,,, ,, CA,,). and D,Mai,,

vector subtraction is carried out in an array processor call

ri = .Y~cos 4 t

and crosscorrelation

receiver.

of T,,

Vectors

t 2& sin (Iv,).

Z,,-

the components

of

For example,

la. Figure

I b shows traces before artificial \tatic\.

and RNMO

were applied.

same traces after these were


shifts were applied

in Figure

put

Id.

I-igure

Ic shows the

in. Surface: inconsi\tcrrt


The result\

of suri-r.~

Seismic 4

Downloaded 12/01/15 to 132.239.1.230. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

I,#
,

427

Y..
.
.
.
1
:
..
..
..
.
.
..
..
...
*
...
. ._ .
.,
..
.
.
..
..
::::::::,:;:::::::::::::::,:,:;:::::::::
:::::::::i::::::::::
,
.
..
.

~ i_~~~~
:::::::::::::::::::.:::::
:i::::::::J:::::::::i:::...:
i,iiiiiiiiiiiliiiiiliiiiiiij.iiiiillilii]~~~~~~~~~~:~:~~~~~
, . . . i,iiiiijiiiiiiiijjiiiiiiiiiiiiiiiiililij,~~:~:~~~~~~
i:ix;: . * .
I:;
;jiiiiiiiirjj:iiiijj:iiiji::::
.,..,.,.,_...,....................
..: ,,: :i::
. : ::.
... ..*........,.. :.,
.
:.:: ::.
, , . . . ..,,.,,,.,
,.,.*.,. ..... *.....a.,.*
.... . . * .,.::::! .:.:. . .
:.
::::::I:
::::::
::::::::::::.:
!
:
.
::.,::,::
:, ::::::.!:::::::.I:::,I.
: .::
:..::::
: : ::: : : ..
.
._
.
.
*
1
:
I.
..
..
,(
.
?
.?
Ye+.
-1.1
XC00R0

v
*

QATHER

STACKS

Nc 8trtkr In data

bndcrnrtatkohdrta

FIG. 1. (a) Simulated cross-array survey. (b) Gather traces before artificial statics placed in data. (c) Gather traces after artificial
statics placed in data. (d) Gather traces after being shifted by surface inconsistentstatics. (e) Gather traces after being shifted by
surface consistent statics. (f) Stacked traces before artificial statics placed in data. (g) Stacked traces after artificial statics placed
in data (gained by a factor of 2). (h) Stacked traces after being shifted by surface inconsistentstatics. (i) Stacked traces after being shifted by surface consistent statics. (j) Histogram of (known-computed) receiver statics. (k) Histogram of (known-computed)
shot statics.

Downloaded 12/01/15 to 132.239.1.230. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

428

C
After

1,

_,--____

rtatka
~~__ ..l__

iI

.I

Before rtatkk

FIG.2. (a) Southmayd 3-D survey coverage. (b) Line from survey before statics applied. (c) Line from survey after statics applied.
(d) Detailed view of line, midpointsfrom 8 to 15, times from 1.I to 1.2 s, beforestatics. (e) Detailed view of line, midpointsfrom 8
to 15, times from 1.I to 1.2 s, afterstatics. (f) Detailed view of line, midpointsfrom 16 to 23, times from 1 5 to 1.7 s, before statics.
(g) Detailed view of line, midpoints from 16 to 23, times from 1.5 to 1.7 s, after statics.

consistent statics are displayed

in Figure

le.

Figures

If

amplitude and wavelet character.


is destroyed.

traces along the line illustrated in Figure la. Deterioration

consistent statics, shows the proper structure.

of

Figure

Ii.

after

but continuity

through Ii show the same sequence of results for the stacked


stacks with statics in the data is quite evident in Figure lg.

bulk shift resulting from long-wavelength

Figure

cannot

Ih show\

that ~rface

inconsistent

statics restore\

be t-r\olved

i\

present.

of structure

the application

Figurc5

of surface
ho\vcver.

;I

component5 which
1.j and

Ik

9how

Seismic 4
histcrgrnm\ of the dillercnce

between knoc\n and computed

static\. .4s can be seen. the statics (originally

~31 m\) have

hecn reduced considerably.

429

tion of ~urtili.~:-consistent prc\txk


required the introductiun

3-D ditta analysis: the correlation


computed from prcmigrated

Downloaded 12/01/15 to 132.239.1.230. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Real 3-D survey-Southmayd

t&time

The 3-D statics programs were successfully run on a real


3-D survey from Grayson County.
a producing

Texas. This is a \ul-vey ol

field designed to delineate

thin.

meandering

sands which change thickness over the survey region. The


survey w;i\ relatively
square milts.

method\.

small for 3-D; 70 000 traces covering 6

Shot and receiver

lines as well as midpoint

surface provides

means of investigating

fhi\

technique

of a new computational
time

tool fol

A qu;intiry

surfxe.

stacked 3-D data, the correlathe seimic

dat;l

with 21

;~naly~t

the time intcrrelation~hi~~

of jtach

3-D data. It contains the imprint of both geological structure


and static error\.

while being relatively

Inxcn\itive

to geolog-

ical faults.
By means of a model \ludq
geometries are simulated in

;I

111which three acquisition


3-D program (Imear.

sw;lth,

coverage are shown in Figure la. The average fold is 5. Shot-

and areal spreads). we look in gl.eater- detail into the nature

and-receiver

of the correlation

spacing was 220 ft. giving midpoint spacing of

time surEIce. The prexnt

study demon-

110ft. There is a strong, fairly flat reflector (seen in Figures

strates the following aspects of the correlation time surface:

Ib and 7~) at 1600 m (-8000 ft) due to an unconformity

(1) The correlation

base of the Pennsylvanian.


cro~scorl.elation.

This provided

Because of its tIatncs5.

point binning (330 ti) was chosen. lhi\


which along with the 587 receiver

My-3

resulted in

particularly for swath and liner receiver spread\. (1) It is not


affected by vertical faults parallel to the acquisition direction

determination.

allowed.

at

After

1600 ms. Trace

shifts of i?3

surface-consistent

(i.e.. it MJill not confuse them with statics errors). (3) Static\
errors and geological structure (with the exception of faults)
both place their imprint on the correlation

time surface.

( . 45) and good

fold

signal-to-noise ratio. Ihe cros~corl.el;ltiun window wa\


ms centered

the

data.

true mid-

statics program. the pal-tial stack option

U;I~ chosen because of the high ctftxtive

time surface cdn he used to analyre

of the reGdua1 statics solution for 3-D

effectiveness

I 468 bins

positions and 45.5 shot

positions gave 3 978 par-ametcrs for static\


For the firit-stage

at the

;I good event fog

statics were

1000

ms were
applied.

Introduction
In a recent

publication

(Schultz

and

Lau.

19X4). we

second pas5 of statics utilizing the master trace option was

introduced two new concepts for the more rflcctive

run. Shifts found from this were small f-,X

ms) and were

ing and analysis of 3-D land seismic data: the correlation

fashion :IS trim

time surface, and a trend analysis applied to that surface. In

applied to the data in a surface-inconsistent


statics.

that

Figures 2b and 3c show one line from the 3-D hurvry

prior

study,

technique

we

permitted

showed

that

proccs+

the use of this new

a set of J-D land data (shot using a

before and after Italics but with otherwise identical procesc-

swath

ing. The mo\t notable improvement

application of a set of static time shifts computed from and

section around

is on the I-ight side of the

I.1 s. This is shown in a detailed

view in

Figures 2d and ?e. The point of this survey was to track the
changing thickness of the Davis sands 1-7900
the uncomtbr-mity)

ft. just above

across the producing field. These sandj

acquistion geometry)

to be improved

applied to the post-stack (but premigrated)

through the

3-D data.

In this study. we apply thin new appoach to synthetic data


to demonstrate

some of the features of [he correlation

surface. We will not consider the trend analysis a5 part

lime

ofthe

are at most 100 ft thick which is well belobv the dominant

present study. although its inclusion is implicit when any

seismic wavelength.

attempt 1s made to compute static corrections alicr stack.

interpretation

The

effect

was used in the

to relate reflection amplitude direc[ly to sand

thickness. Contamination
destroyed

tuning

of the data by statics would have

this relation as illustrated

by another large-scale

Two particular featlure\ of the correlation time \urface are


investigated

in this study:

First. that the correlation

time

surface contains the imprint of both geologic struc~ilre and

display in Figures 2f and 2g (notice the reflection just above

residual statics errors. If an appropriate

I.6 s). Thus. statics correction was an essential part of 3-D

be made (such as that used in the prior study). the statics


errors

processing for this survey.

may

bc separable

from

trend analysis can

structure

and post-jtack

corrections can be made to the data. Second. that vertical

References

faults parallel to the acquisition direction will not be imprint-

Hilrman, J. A., Embree, P.. and Pflueger, J. C., 196X. Automated


static corrections: geophys Prow. 16. 326-35X.
Taner, M. T.. Koehler. F:, and Alhib. K: .A.. 1974. Estimation and
correclion of near-surface time anomalies: Geophysics. 39, 441463.
Wiggins. K. A.. Lamer. K. L., and Wisecup. K. D.. 1976, Kehldual
q;taticsanalysis as a linear inverse problem: Geophysics. 41, 92293x.

ed on the correlation

time surface, and therefore will not bc

confused with statics errors.

The synthetic model


A synthetic data set was cleated on which the study was
performed.
Figure

The

geological

events were generated

Correlation time Surface: A New Tool


for 3-C Statics Analysis

54.5

structural

I; it shows approximately

model

is shown

parallel to the horizon of Figure

separated by I00 ms, and placed at approximately


the section. Random
created with

in

400 ms of relief. Three


I.

3.0 > on

noise was added. .4 3-D data set W:IS

180 x 360 CDPs

(bin dimension 50 x 3

giving total survey dimensions of Y x Y km).

A method
1984) where.

was recently
through

additional contribution\
statics solution. IIs%

the long-wavelength

introduced

analysis

of 3-D

(Schultz

and Lau,

data after stack,

could be made to the total residual


statics correction\

involve

some of

spectral components beyond the resoltr-

Synthetic

data

were

collected

using each of thl-ee

acquisition geometries: first. a linear spread ofrcceivcr\

138channels) Gfh

;I 25

(I A

m in-lint 1gr-oupseparatic>n: \econtl. iti

swath spread of receivers (1 x 37 channel\)

with ,I 75 111in

line and 50 m cr_w,sline group scpxation:

third.

;~n L~lI~I

Das könnte Ihnen auch gefallen