Ci~HYSICS.

VOL.

44.

NO.

6 (JUNE

1Y7Y); P. IMI-1063.

16 FIGS.,

I TABLE

Complex seismic trace analysis

M. T. Taner*, F. Koehler*, and R. E. Sheriff*
The conventional seismictrace can be viewed asthe real component acomplex trace which can be uniquely calculated under usual conditions. The complex trace permits the unique separation of envelope amplitude and phase information and the calculation of instantaneousfrequency. These and other quantities can be displayed in a color-encoded manner which helps an interpreter see their intcrrelationahipand spatial changes. The significance of color patternsand their geological interpretation is illustratedby example\ of seismic data from three areas.

of

INTRODUCTION of amplitude and phase information. two of the This paper has two objectives: specifically to (I) quantities (called attributes) which are measured explain the application of complex trace analysis to in complex trace analysis, The amplitude attribute seismic data and its usefulnessin geologic interprcta- is called reflection strength. The phase information and (2) illustrate the role of color in conveying tion is both an attribute in its own right and the basis frequency measurement.Amplitude seismic information to an interpreter. Expressing for instantaneous seismic data in complex form also yields computa- and phaseinformation are also combined in additional tional advantageswhich are discussedin Appendix A. attributes, weighted average frequency and apparent Transformations of data from one form to another polarity. Signal analysis can also be viewed as a communiare common in signal analysis, and varioustechniques are used to extract significant information from time cations problem. The ob.jective is to make an interseries (seismic data). Interpreting data from different preter aware of the information content of data, inpoints of view often results in new insight and the cluding an appreciationfor the reliability of measurediscovery of relationships not otherwise evident. ments and how information elements relate to each The transformation of seismic data from the time other. The display of data is an inherent part of the domain to the frequency domain is the most common analysis. Seismic data are conventionally displayed example of data rearrangement which provides in- in variable area, variable density, vat-iableamplitude sight and is useful in data analysis. The Fourier trans- (wiggle). or a combination of these forms. Display form, which accomplishes this. allows us to look at scale and vertical-to-horizontal scale ratio are variaverage properties of a reasonably large portion of ables whosejudicious choice can aid analysis(Sheriff a trace, but it does not permit examination of local and Farrell. 1976). Display parameters also include variations. Analysis of seismic data as an analytic trace superposition,bias, and color. Color has proven signal, complex wuw unalyis, is a transform tech- to be especially effective in complex trace analysis. nique which retains local significance. Complex trace The literature on the use of color In geophysics is analysis provides new insight, like Fourier trans- limited. Balch (1971) discussed the use of color forms. and is useful in interpretation problems. seismic sections as an interpretation aid, and geoComplex trace_ analysis tiff&s a natural_sepamtinn~ physical advertisements have illustrated limited use

Presented at the 46th Annual International SEC Meeting October 27, 1976 in Houston and at the 47th Annual International SEG Meeting September 21, 1977 in Calgary. The subject matter constituted the lectures given by M. T Taneras AAPG Distinguished Lecturer 1975 and by R. E. Sheriff as SEC Distinguished Lecturer in 1977. Manuscript received by the in Editor January 23, 1978; revised manuscript received August 7, 1978. *Seiscom Delta Inc., 7636 Harwin, P. 0. Box 36928. Houston, TX 77036. OOl6-8033/79/0601-lO41$03.00. @ 1979 Societ! of Exploration Geophysicists. All rights reserved 1041

1042

Taner et al

cos (wt + 0) for all real values ol and 0 and all A w>o. These rules determine J (I) uniquely for any func* tion ,f t) which can be represented by a Fourier ( series or Fourier integral. The use of the complex trace F(r) makes it possible to define instantaneous amplitude, phase, and frequency in ways which are logical extensions of the definitions of these terms for simple harmonic oscillation. Complex traces can ;IIVI be used in similarity calculations. enabling us to lind more precisely the relative arrival times of a common signal appcaring on different traces (Appendix A). The real seismic trace ,f (t) can he expressed in tcnns of a time-dependent amplitude A (t) and a timedependent phase 0(r) as .f (I) = A (t) cos d(t). The quadraturetrace ,f* (1) then is (I)

(b)
FIG. I Frequency domain representationsof (a) real and (b) complex traces.

f* (t) = A (t)

sin

O(t),

(2)

and the complex trace F(r) is F(t) = J t) + jf(t) ( = A (t) P(1). (3)

If f (t) and .J* (f) are known. one can solve for A(t) and 0(t):
of color in emphasizing reflection amplitude anomalies (bright spots) in showing the direction of cross-dip, etc. A (t) = [J(t) and e(t) =
CALCULATION OF THE COMPLEX TRACE
tan- [f* (t) /.f (t)]. (5)

+ f*r)] (

= IF(t) (,

(4)

A(t) is called reflection strength. and 19(r)is called

Complex trace analysis is discussed in electrical engineering and signal analysis literature (Gabor. 1946; Bracewell, 1965; Cramer and Leadbctter, 1967; Oppcnheim and Schafer. 1975). Some applications to seismic signal problems are given in Farnbach (197% and Taner and Sheriff (1977). However, explanation of the application to seismic signal analysis is not available in the geophysical literature.
Basic definitions

instantaneous phase (Bracewell, 1965). The rate of change of the time-dependent phase gives a time-dependent frequency

This can be expressed in convolutional form as -xi

m(t) =

Complex trace analysis treats a seismic trace .f (t) as the real part of an analytical signal or complex trace, F(t) =.f(t) +if*(r). The quadrature (also called conjugate or imaginary) component ,f* (t) is uniquely determinable from f(t) if we require that f 0) *

I --m

d(7) e(t ~ T)d7,

(7)

I) be determined from f t) by a linear convolution (

operation. and 2) reduces to phasor representationiff (f) is a sinusoid, that is, f*(r) = A sin (wt + 0) if f(t) = A

where d(7) is the differentiation filter (Rabiner and Gold. 1975, p. 164). A difficulty with this is that the phase must be continuous, whet-cas the arctangent computation of equation (5) give\ only the principle value. We then have to unwind the phaseby determining the location of2n- phasejmps and correcting them. A more convenient way of computing the instantaneous frequency is to compute the derivative of

Complex Seismic Trace Analysis

1043

the arctangent function itself

w(t) = f {tan- 0) /f (t)ll, lY*

which results in

(8)

f o(t) =

@)

4 0)_f*
dt

@)

df
dt

P(t) +f*t) (

(9)

FIG. 2. Normalized Hilbert time-domain operator truncated to I9 points.

where the derivatives off(t) and f* (t) can be computed in convolutional form as in equation (7). We also define a weighted average frequency G(t) as oc 0. Then f*(t)

= lornC(o) sin [wt + 4(w)]dw;

i%(t) = -= m A(t - T)L.(T)dT I -cc

A(t - 7) w(t - T)L(r)dT t (10)

and

(12)
g C(o)ej[wf++(w)ldw. I0

F(t) =

where L(T) is a low-pass filter. Apparenrpolarity is defined as the sign of f(r) when A(t) has a local maximum. Positive or negative sign is assigned assuming a zero-phase wavelet and a positive or negative reflection coefficient, respectively.
Calculation of the quadrature trace

We give equivalent ways of defining f*(t) and F(r), first in terms of Fourier integrals and then by convolution in the time domain using the Hilbert transform. We assumethat f (t) is real, defined for --XI < t < 33, and can be represented by the Fourier integral formula

The frequency-domain representationsof a real trace and its complex trace equivalent are shown in Figure 1. The amplitude spectrum of the complex trace C(o) vanishesfor o < 0 and has twice the magnitude for w > 0. The phase 4(o) is unchanged (except it is not defined for o < 0). The complex trace can thus be found by (I) Fourier transforming the real trace, (2) zeroing the amplitude for negative frequencies and doubling the amplitude for positive frequencies, and then (3) inverse Fourier transforming. An equivalent formula for f*(t) is given by the Hilbert transform (Rabiner and Gold, 1975) f*(7) =+ P.V. Im f(t),,

-,7-t f(t)
and

(13)

= _fa B(o)ej d --m

o,
(11)

f(t)

= j-omC(w) cos [wt + $(o)]dw,

1 Andy 4(w) =a@(,), w>

where P.V. j-mmmeans the Cauchy principle value a P.V. _-m= ;: [/;;E+l::]. (14) I The Hilbert transform can be .used to generate the quadrature traces ftom the real trace or-vice versa by the convolution operation, which in digital form is

-where C(w) -2jBjwji

f*(t)=+
and

2 f.(t-nAt)= n=-cc

1n # 0,
i

(1%

f*(t)

=i

f f(t - nAt) sinz(~ ) 2 n=-cc

1044

Taner et al

FIG. 3. Real (a) and quadrature(b) traces for a portion of an actual seismic trace. Also shown is the envelope [dotted curve in (a, b)], phase (c), instantaneousfrequency (d), and weighted average frequency [dotted curve in (d)].

Complex Seismic Trace Analysis

FIG. 4. Isometric diagram of portion of an actual seismic trace.

where Ar is the sample interval. The inverse convolution is merely the negative

(t)

The normalized Hilbert time-domain operator [equation (15)], shown in Figure 2, is odd, vanishes for even n, and decreasesmonotonically in magnitude as In/ increases for odd n. It is usually applied in a modified truncated version.

Graphical representations and examples The real f(r) and quadraturef*(t) traces can be plotted in any of the conventional ways used for seismic traces. Variable amplitude plots for a portion of an actual seismic trace are shown in Figures 3a and 3b for the real and quadraturetraces. The complex trace F(r) can be thought of as the trace in complex space of a vector which is continually changing

its length and rotating, thus tracing out an irregular helix as shown in Figure 4. We may then think of A(t) as the time-varying modulusand O(f) as the timevarying argument of this vector. The seismic trace shown in Figures 3 and 4 is from an East Texas survey. The real and quadraturetraces are given by the projection of the tract of the rotating vector on the real and imaginary planes, as shown in Figure 4. The length of the vector is A (I) and its angle with the horizontal is O(rj. Figure 5a shows a simple Ricker wavelet f (I) and the quadraturetracef* (I) derived from it. Also shown is the magnitude IF(t)1 = A(r) and the phase O(f). Figure 5b is an isometric diagram of the same wavelet showing the quadraturecomponent_/ (I) in the imagi* nary plane perpendicularto the real componentf (t). Figure 5c is a polar plot of A(t) = IF(r) I at successive and equal time intervals, and Figure Sd is the corresponding amplitude spectrum A (0). Data for this example are tabulated in Table I of Appendix B. Note in Figures 3a and 3b that both real and quadrature traces are identical except phase shifted by 90 degrees. Except for this phase shift, a geophysicist would observe the same features, that is, the same

1046

Taner et al

- r/r/ --- P,,, -w/f,, 8,,,

/

-y

.., 2,~ .,.,..

(4

(W

FIG. 5. (a) Real partf(t), quadraturepart f*(t), complex amplitude IF(t and phase O(r) of 25Hz Ricker wavelet. (b) Isometric diagram of real, f(r), and quadrature, f*(r), components of 25Hz Ricker wavelet. (c) Polar plot of A(r) = IF r)1 for a 25Hz Ricker wavelet. (d) Spectrum B(o) of 25Hz Ricker wavelet. (

Complex Seismic Trace Analysis

1047

coherency and the same signal-to-noise ratio. on real and quadrature seismic sections. The reflection strength A(t) is the envelope of the seismic trace. We might imagine the reflection strengthrotatedaboutthe time axis so as to appearlike beads on a string, sometimes overlapping, each bead representing the arrival of new energy. The vector rotates within each of these beads and the phase (Figure 3c) occasionallyhas to back up or hurry ahead to represent succeeding energy. The instantaneous frequency curve (Figure 3d) jumps sharply whenever the rotating vector is locking onto new energy but does not change appreciably during each bead of energy. An average of the instantaneousfrequency, such as given by the weighted average frequency, yields roughly the same value we would obtain if we were to measure the period between successivepoints of similar phase for succeedingcycles, as is often done to determine dominant frequency. For the 2%Hz Ricker wavelet of Figure 5, W = 24.5 Hz, which is close to the 25.Hz value of o at the peak of its amplitude spectrum (Figure 5d).

SIGNIFICANCE

OF ATTRIBUTES

Attribute measurements based on complex trace analysis were defined in the preceeding section. We now examine their significance and color representations as originally described by Taner et al (1976). Reflection strength Reflection strength (amplitude of the envelope) is defined by equation (4). Reflection strength is independentof phase. It may have its maximum at phase points other than peaks or troughs of the real trace, especially where an event is the composite of several reflections. Thus, the maximum reflection strength associated with a reflection event may be different from the amplitude of the largest real-trace peak or trough. High-reflection strength is often associated with major lithologic changes between adjacent rock layers, such as acrossunconformities and boundaries associated with sharp changes in sea level or depositionalenvironments. High-reflection strengthalso is often associatedwith gas accumulations. Strength of reflections from uncomformities may vary as the subcropping beds change, and reflection strength measurement may aid in the lithologic identification

of subcropping beds if it can be assumed that deposition is constant above the unconformity so that all the change can be attributed to subcroppingbeds. Lateral variations in bed thicknesses change the interference of reflections;suchchangesusually occur over appreciable distance and so produce gradual lateral changes in reflection strength. Sharp local changes may indicate faulting or hydrocarbon accumulations where trapping conditions are favorable. Hydrocarbon accumulations, especially gas, may show as high-amplitude reflections or bright spots. However, such bright spots may be noncommercial and, conversely, some gas productive zones may not have associatedbright spots. Observing where, within a reflection event, the maximum reflection strengthoccurs provides a measure of reflection character. Occasionally, this can be used to indicate reflection coefficient polarity as shown by Taner and Sheriff (1977, p. 327). Constancy of character may aid in distinguishing between reflection events from a single reflector and those which are a composite of reflections The strength of reflections from the top (11 massive beds tends to remain constant over a large region. Reflections of nearly constant strength provide good references for time-interval measurements. The usual color-encoding of reflection strength is referenced to the maximum reflection strength which occurs on a seismic section or in an area, using a different color for each dB step (Figure 6a). Using the same color reference for the data over an area provides color ties at line intersections. providing data recording conditionswere unifornl or corrections for nonuniform recording conditions were made in processing. The reference can be changed where desired. Instantaneous phase The instantaneousphase, defined hy equation (5), emphasizes the continuity of events. Instantaneous phase is a value associatedwith a point in time and thus is quite different from phase a$ a function of frequency, such as given by the Fourier transform. In phase displays, the phase corresponding to each peak. trough, zero-crossing, etc. of the real trace is assigned the same color so that any phase angle can be followed from trace to trace. Becausephaseis independentof reflection strength, it often makes weak coherent events clearer. Phase displays are effective in showing discontinuities, faults, pinchouts, angularities, and events with different dip attitudes which interfere with each other.

1048

Taner et al green end, the higher frequencies. Frrquencie\ lower than 6 Hr arc usually left uncolored. Weighted average frequency Weighted average frequency. defined by equation (IO). emphasizes the frequency of the stronger reflection events and smooths irregularities caused bj noise. The frequency values approximate dominant frequency values determined by measuring peak-topeak times or times between other similar phase points. Like instantaneous frequency displays, weighted average frequency displays are sometimes excellent for enhancing reflection continuity. Hydrocarbon accumulations often arc cvidenccd by IOU frequencies. Apparent polarity While all attribute measurementsdepend on data quality and the quality of the recording and processlng, apparent polarity measurementsare especially sensitive to data quality. Interference may result in the reflection strength maximum occurring near a zero-crossing of the seismic trace so that the polarity may change sign as noise causes the zero-crossing of the trace or the location of the reflection strength maximum to shift slightly. The analysis of apparent polarity assumes a single reflector, a zero-phase wavelet. and no ambiguity due to phase inversion. However, since most reflection events are composites of several reflections, polarity often lacks a clear correlation with reflection coefficient and hence it is qualified as apparent polarity. Polarity sometimesdistinguishesbetw,eendifferent kinds of bright spots(Figures 7c and 7f). Bright spots associated with gas accumulations in elastic sediments usually have lower acoustic impedance than surrounding beds and hence show negative polarity for reservoir top reflections and positive polarity for reflections from gas-oil or gas-water interfaces (often called flat spots) (Figure Xc. event D). Ordinarily, apparent polarit) is color-coded magenta and blue for positive and negative, respectively, with the hue intensity graded in five steps according to reflection strength (Figure 6d). Display of attributes Each attribute to be displayed involves a value associated with each sample point. Assimilating and digesting such massesof data pose a major problem. Our usual practice is to color-encode the data and display these in a seismic-section format most

Prograding sedimentary layer patterns and regions of on-lap and off-lap layering often show with special clarity so that phase displays are helpful in picking seismic sequence boundaries (Payton. 1977. p. 310). Phase displays use the colors of the color wheel (Figure 6b) so that plus and minus I80 degrees are the same color (purple) because they are the same phase angle. The cosine of the instantaneousphase angle is also displayed in black and white and is often used as a background for other displays (as in Figures 8 and IO- 13). Instantaneous frequency Instantaneousfrequency, defined by equation (6), is a value associated with a point in time like instantaneous phase. Most reflection events are the composite of individual reflections from a number of closely spaced reflectors which remain nearly constant in acoustic impedance contrast and separation. The superpositionof individual reflections may produce a frequency pattern which characterizes the composite reflection. Frequency character often provides a useful correlation tool. The character of a composite reflection will change gradually as the sequence of layers gradually changes in thickness or lithology. Variations, as at pinchouts and the edges of hydrocarbon-water interfaces, tend to change the instantaneousfrequency more rapidly. A shift toward lower frequencies (low-frequency shadow) is often observed on reflections from reflectors below gas sands, condensate, and oil reservoirs. Low-frequency shadowsoften occur only on reflections from reflectors immediately below the petroliferous zone, reflections from deeper reflectors appearing normal. This observation is empirical and many have made the same observation, but we do not understand the mechanism involved. Two types of explanations have been proposed: (I) that a gas sand actually filters out higher frequencies because of (a) frequency-dependent absorption or (b) natural resonance, or (2) that traveltime through the gas sand is increasedby lower velocity such that reflections from reflectors immediately underneath are not summed properly. These explanations seem inadequate to account for the observations. Fracture zones in brittle rocks are also sometimes associated with low-frequency shadows. Frequency is usually color-coded in 2-Hz steps (Figure 6~). The red-orange end of the spectrum usually indicates the lower frequencies and the blue-

Figure 6. Color codes for attribute values. (a) Reflection strength; (b) phase; (c) frequency; (d) polarity.

Figure 7. Two portions (left and right) of a seismic section for Gulf of Mexico line A. Top: reflection strength; center: instantaneous frequency; bottom: apparent polarity.

Figure

8. Portion of seismic section for Southern North Sea line A. Top: instantaneous weighted average frequency; bottom: apparent polarity.

phase; center:

Reflection time in sec.

Reflection

time in sec.

Raflactinn

tima

in eat

Reflection

time in sec.

Figure

13. Seismic section for Gulf of Mexico average frequency.

line C. Above:

reflection

strength;

below: weighted

Complex Seismic Trace Analysis

familiar seismic nate. often tion to line interpreters. as abscissa that is. location time along a reflection commercial right display shells. shadows the The at 0.650 set is associated The bright with spot

1057
a nonon the of

and reflection attribute

as ordiare secdata thus

gas-reservoir.

Color-encoded superimpo\cd

measurements seismic

(d) is associated gas-reservoir

with

a local deposit low-frequency as shown display no

on a comentional

zones

have them.

so that one can see both attribute see

the conventional simultaneously,

immediately in

underneath the

by (b).

and the color-encoded making encoding range but it easier involv/cs to

yellow-orange the shell

frequency products The polarity

interrelations.

The color

whereas change reservoir color)

deposit (e).

systematic of the gas(blue

assigning This

a color to each v/alue or can bc arbitrary, sequence clear. by the the Many is most

in frequency reflections. that

of values.

assignment in spectral

as seen in (c), deposit

is negative (f)

assigning

colors

and

of the shell

is positive

natural

in making

relative

magnitude

(magenta clear that

color). the

Thus the cnscmble two bright spots

ol displays makes rcpmscnt different

The examples Seis-chrome@ same color distinctly (Figure

in this article process which

were produced produces

exactly

subsurface Most

features. interpretation (Sheriff, hcgins 1976). with the

whenever

the same value

occurs.

stratigraphic

different 6) is usually

colors

can bc used. A color code so that one can detcrwith any sample changes quantita-

interpretation pattcrnx helps

of structure

Attribute of pattcms it is interest

pattern

provjided

aid in correlation, throw

and the otfsct faults.

mine the numerical location tively. Displays I) which and thus

value associated interpret color

establish

across

flowever,

v/ariation

along the bedding interpretation. stratigraphic of

that is of principle
iAtCr l d VarkltiOn in

in attribute have been used include xcalc ( squash and plot ) and suggests the

or other

changes.

Sometimes the

compressed normal ( zooms )

time

meaning

a variation when et al,

is clear. well

but often

horizontal

scale.

enlargement

meaning to seismic

is clear only data (Sheriff are

data are correlated .4s more addi-

of zones of special traces

interest; phase as a area,

1977). the more Tho\c which

2)

conv*entional black-and-white density,

or instantaneous in and

tional

data

assimilated,

interpretable familiar others with miss. Iowis

background modes;

vtariable

are these attribute local geology

measurements. significance

or amplitude

find

3)

blanking conventional

out of samples trace

at zero-crossing white

of the lines

The use of lateral frequency obvtiously zones.

changes of pattern. to define limits ol

especially production

so as to produce structure. often

evlery half cycle Processing 1) phase wavelet 2) prcdctcrmincd the color 3) migration more surface nearly display

to indicate

important. the horizontal section

of data prior to display filtering to convert

involves [en-phase

Compressing length vertical

scale allows a greater

The results changes i\ often which helpful occur

to a nearly

of seismic

to be cornpi-chended.

exaggeration

which

time-dependent

pain to accentuate and

in

delineating

stratigraphic

for the zone of interest;

over long distances. ture, howev*er.

Such a section does distort strucfaults to appcai- steeper, etc.

prior to analysis conform

so that seismic ev*ents of related sub-

causing

to positions

Attribute grated

interpretation

can be made reflection Greene,

on data miand

features.

so as to preserve (Reilly features and

amplitudes 1976).

General interpretation considerations

The they in vJarious attributes rcvcal more often as a set than are anomalous As Figure an 7 do individually. systematic ways Features on

character sharpens tions,

Migration complicadips, so that

and resolves

structural

such as buried foci and conflicting interpretation is more

attribute

meaniiigful. EXAMPLES namely, phase,

various

displays. displays,

example

of the value

of multiple and right)

INTERPRETATION Display weighted of average three

shows two portions tion for Gulf in Taner

(left

of a seismic

sec-

attributes,

ofMexico line
Sheriff, displays (bright

A. (This section is shown Figures and d) X-12.) show The highby the on the the

frequency

and polarity,

of a portion North

and

1977, (a

of a seismic Sea is shown

section

for line A in the Southern 8. Figures 0 and

reflection-strength amplitude events

in Figure

IO show sec-

spots). The bright

:I guz

as indicated spots (red)

reelection tions, The

strength

and instantaneous for a larcper portion been corrected stacking

frequency

red and orange left display (a)

colors. indicate

respectively, input

01 this section. tar variation of

reservoir.

though

data have

source wav/elet @Seiscom Delta Inc. shape

shape before

and the wavelet prior to the

has been corrected

to zei-o pha\c

1058

Taner et al

FIG. 14. Interpretation of seismic section for Gulf of Mexico line B (shown in Figures 11 and 12)

complex-trace analysis. These sections have been migrated by the wave-equation method, and blackand-white phase traces form the background of the sections, except for the phase display (Figure 8a) itself. Interpreted subsurface features are identified on Figure 16. The phase display (Figure Xa) emphasizes continuity and angularities of weak reflections because it is insensitive to amplitude, Thus, the weak-dipping reflections which subcropat the angular unconformity just abovfe0.5 set (A) delineate this unconformity. On a conv,entional section these reflections are so weak that it is difficult to locate the unconformity 50 precisely. Similarly, the unconformities at B and C are made clear by onlap, downlap, and truncation configurations. A Hat spot (D) can be seen associated with the gas resetvoir at the crest of the anticline at I .2 sec. The weighted average frequency display (Figure 8b) should be compared with the instantaneousfrequency display (Figure IO). Laterally constant layer sequences, such as the top of the Danian chalk (E), tend to be characterized by patterns which aid in re-

flection correlation, whereas the patterns change laterally for reflectors such as unconformities (B or C). Orange patterns such as underneath the Danian chalk reflections sometimes seem to be associated with fractured zones (F) (similar patterns are sometimes associatedwith fracture zones in East Texas). There are also low-frequency reflections (orange) in the shadow under the gas accumulation (D). The apparent polarity section (Figure 8c) is interesting mainly for the appearance of the reflection from the gas reservoir. The reflection from the reservoir top has negative apparent polarity (blue) and the reflection from the gas-water interface (D) has positive polarity (magenta). The reflection strengthsection for North Sea line A is shown in Figure 9. Major vertical lithologic changes such as from Tertiary elastics to the chalk (E) or from Triassic elastics to Permian carbonates and evaporites (G) are generally associatedwith high reflection strength. The reflection strength is more constant on the upper of these (Tertiary elastics to chalk), indicating that this lithologic contrast is more constant than the lower contrast where the nature

Complex Seismic Trace Analysis

1059

of the subcropping Permian formations changes laterally. Lateral changes in reflection strength often mark unconformities. The Carboniferous anticline (J) has some amplitude standout. The instantaneousfrequency section for North Sea line A (Figure IO) shows distinctive reflection character associated with the Danian chalk (E) and the Rotliegendes (H), but most of the other reflections change character slowly along,the bedding. Note the low-frequency reflection just below the gas reservoir (D). The block faulting of the Rotliegendes (H) is emphasized by the black-and-white phase background. Figure 11 is a portion of a reflection strength sec- FIG. 15. Interpretation of seismic section for Gulf of tion for line B in the Gulf of Mexico. Several promiMexico line C (shown in Figure 13). nent bright spots are evident (yellow, orange, and red colors). This line is coincident with the crest of a salt ridge and is perpendicular to line C, the section still under development and additional production shown in Figure I3 (C and B at the top of the sections may be established. indicate the intersection). Figure I2 is a weighted Figure I3 shows a portion of a section perpendicuaverage-frequency section for this line. A number of lar to the section shown in Figures I I and 12. Interlowfrequency zones (orange) are in some places pretation of this section is shown in Figure 15. No associatedwith the bright-spotsseen in Figure I I, and drilling has been carried out on the left half of this at other places the low-frequency zones and bright section, but several productive zones have been prespots are not coincident. dicted. Other examples of the geologic interpretation The locations of two wells and information as to of attribute measurements are given in Taner and productive zones in a number of other wells are shown Sheriff ( 1977). on the migrated instantaneous phasesection in Figure CONCLUSIONS 14. The wells are not located on the seismic line but Analysis of seismic traces as part of complex havse been projected perpendicularly onto the line, so some projection errors result. Crossdips are small, (analytic) signals allows the ready determination of so structuralfeatures on this migrated seismic section the amplitude of the envelope (reflection strength), should be nearly correct. The faults have been inter- instantaneous phase, and instantaneous frequency. preted from the seismic data. Color-encoded displays of attribute values aid in Self-potential (SP) logs in the two wells are shown interpretation of seismic data relevant to stratigraphy in Figure I?. Massive shale and interbedded sand and sometimes to hydrocarbon accumulations. The and shale zones, interpreted from the SP logs, corre- reflection-strength portrays reflectivity and hence late well with attribute character, especially phase. information about impedance contrasts. The inLoss of reflections (low reflection strength and de- stantaneous phaseemphasizescoherency and changes crease in phase coherence) and increased noise in dip of successive reflections. The instantaneous (higher frequencies) characterize the supernormal frequency is useful in correlation and sometimes pressure. massive shale zones. Circles indicate pro- appears to indicate hydrocarbon accumulations. duction zones which have been drilled and letters Weighted average frequency aids in identification indicate their order of thickness: A, less than 20 ft of major frequency variations, and apparent polarity thick; B, 20-50 ft thick; C, 50- 150 ft thick. All the sometimes helps in identifying gas accumulations. production is gas. The top of the supernormalpressure Lateral variations in all displays help localize is indicated by SNP. stratigraphic changes. The productive zones generally correlate well with ACKNOWLEDGMEN IS both reflection strength and low-frequency zones, A number of people contributed to the work disalthough a few productive zones do not show as obv,ious anomalies and a few anomalies are not cussed in this paper. Appreciation is especially exD associated with established production. The field is pressed to N. A. Anstey, R. O oherty, and others

1060
in Seiscom Delta, techniques.

Taner et al
the aggregate of whose contribuin the development of these Payton. C. E., Ed.. 1977. seismic stratiyraphy-applications to hydrocarbon exploration: AAPG Memoir 26; Tulsa, Am. Assn. Petr. Geoloprsts. Rabiner. L. R.. and Gold. B.. 197.5. theory and annlication of dr_eital signal processing: Englewood Cliffs. N. J.. Prennce Hall. p. 70-72. Reilly. M. D.. and Green. P. L.. 1976. Wave Equation Mmration: presented at the 46th Annual intl SEC meeting October 27 in Houston. Sheriff. R. E.. 1973. Encyclopedic dictionary of exploration geophysics: Tulsa, Society of Exploration Geophysicists. ~ 1976. Inferring stratigraphy from seismic data: Bulletin of Am. Asan. Petroleum Geologists. v. 60. D. 528p.542. Sheriff. R. E., and Farrell, J.. 1976. Display parameters of marine geophysical data: Dallas. OTC paper no. 2567. Sheriff. R. E., Crow. B. B., Frye. D. W., and Rao, K.. 1977, Hydrocarbon delineation by analytic interpretation: Exploitation Studies. presented at the 47th Annual Intl. SEG Meeting, October 27 in Calgary. Taner, M. T., Sheriff, R. E., Koehler, F.. and Frye, D.. digital computer derivation and applications of velocity functions: Geophysics, v. 34, p, 8599881. Taner. M. T.. and Sheriff, R. E.. 1977. Application of amplitude, frequency, and other attributes to stratigraphic and hydrocarbon determination: in Applicatrons to hydrocarbon exploration, C. E. Payton, Ed.. AAPG Memoir 26: Tulsa, Am. Assn. Petroleum Geologists. p. 301-327. Taner. M. T., Sheriff. R. E.. Koehler, F., and Frye, D., 1976, Extraction and interpretation of the complex seismic trace: presented at the 46th Annual Intl. SEG Meeting. October 28 in Houston.

tions have resulted

N. A. Anstey was the pioneer in both the development of the techniques and in appreciating geological significance. published booklets, The Seiscom 2 7 their He authored two privately and Seiscom 3, 7

which have been important references. assistance of clients who elect to remain especially for their anonymous is also acknowledged,

permission to publish the sections.

REFERENCES
Balch, A. H., 1971, Color sonagrams: A new dimension in seismic data Interpretation: Geophysics, v. 36, p.
1074-1098. Bracewell, R. N., 1965, The Fourier transform and Its applications: New York, McGraw-Hill Book Co., Inc., p. 268-271. Cramer, Harold, and Leadbetter, M. R., 1967, Frequency detection and related topics: Stationary and related stochastic processes, Ch. 14, New York, J. Wiley and

S., 1975, The complex envelope in seismic signal abtalysis: SSA Bull., Y. 6.5. p. 951-962. Gabor, D., 1946, Theory ofcommunication. I: J. In\t. part Elect. Eng., v. 93, pan 111, p. 429-441. Oppenheim, A. V., and Schafer, R. W., 1975, Digital signal processing: Enylewood Cliffs. N. J., Prentice Hall.

Sons Famback.

PROCESSING Let us define a seismic tracef analytic f (t) * F(t) =f (t) + jf (r) * trace F(r),

APPENDIX A APPLICATIONS OF COMPLEX

TRACE -fl(t)fZ(f + 7)ldt; (A-3)

(t) as the real part of trace is

where the quadrature

4(T)

A,(t)A,(t .cos[O,(t) 1

+ 7). - 02(t + 7)ldt + 7). (A-4)

--x

= A(t) [cos O(r) + j sin e(r)].

(A-

I)

Cross-correlation
The cross-correlation and F*(r), is of two analytic traces, F,(r)

+_i

I -x

A,(t)A,(r

.sin[Ol(r) - 02(t + r)]dt.

4(T) =

Ix ~
--r

f-l(~)FS(f + T)df,

(A-2)

Arrival

time measurement

where the bar indicates the complex conjugate;

Phase measurement determines the relative arrival times of signals of similar form. This has important implications in velocity spectra analysis, velocity and I ) dip determination

4(T) = _I[f~(t)fdf + 7) +f:(f)

+_i

, static time corrections,

linear let us

J-SC

cc c[f?

modeling, take f t) (

and in other processing. as a 25.Hz Ricker wavjelet sampled at

(f) .fi(t + 7)

As an example of a timing measurement.

Complex Seismic Trace Analysis

1061

4.msec intervals, and with a maximum value of I at t = - 1 msec. To findf*(t) from the sampled values, we use a 22-point operator designed for the sampling interval of 4 msec. Values off; f , and 0 at the three * sample points closest to the maximum off are t f (1) f* (1) -.49098 O(f) -30.278 degrees IO. 102 degrees 50.362 degrees

and

-= (gJ+ (I@)*
P ut
pin

N .

(A-8)

(f;

+f*f)

k=l

-4msec .84096 Omsec .98159 4msec .59274

.I7489
.71554

We estimate rmax, the time where f(r) h-as its maximum value, in two waya: 1) the time where an interpolating quadratic for f (I) has a maximum; this gives t,,, = -0.938 msec: and 2) the time where 19(t)= 0 by linear interpolation; this gives I,,, = - 1.007 msec, which is in error by only 7 psec. Conjugate of a convolution If we let f(t) = J:, g(7) s (r - ~)d7, the quadrature trace f (t) is given by either of the equiva* lent formulas f*(I) or = Irn g(T)S*(t -rn - 7)dT,

Equation (A-8) can be used for coherence computations such as those involved in velocity analysis. The effectiveness of trace summation (stacking) can be computed on a sample-by-sample basis, eliminating the necessity of averaging over a time window. Product of time series The product of two time series is

F,F, = AI A, [cos(tl, + 0,) + j sin(81 + e,)].

Similarly,

F, E = AlAP [cos(O, - 0,) + j sin(O1 - O,)].

(A-

9)

(A-5) = - g*(T)S(t I --m - 7)dT.

If 0, = e2, F,F, will be real, but if 10, - f12)= 7r/2, F,F, will be imaginary. Consequently. we can deduce the phase differences between complex time series by noting the ratio between imaginary and real parts. If F, and F, are the same except for a phase shift of0,

f (t) *

F,(t) F2(f) = A*(cos 0 + j sine),

and the argument of the product ha\ the constant value 0. If F, and F, are conjugate pairs, then F,(t) F,(t)

When g(t) is a spike sequenceand s(r) is a wavelet, the natural formula to use is the first of these. Sum of time series We can consider simple filtering as a summation and measureits performance by measuringthe outputto-input power ratio. The sum is given by summing real and imaginary parts,

( = -jAt)

cos O(t) sin e(t),

and arg(F,F,) = ST/~, the constantphase difference between a trace and its conjugate.

(A-6)
k=l k=l k=l

Let us consider a complex cross-correlation over the time window 7

Power is given by

$47) =
(A-7)

C F,(f) Fz(t + ~1
T

Pk(t) = Fk(f) Fk(t).

Hence the output-to-input power ratio is

From equation (A-3). following fnrm

we can write this in the

_=
P,,t
pin

N-

c
k=l

(FrcFk)

+ j

x [J (f)fiU T
T

+ T)

1062

Taner et al Table 1. 2.5Hz Ricker wavelet.

.o
,721 ,142

.O

.o
,956 .837 ,671 .494 .336 ,213 ,128 ,075 ,045 .028 ,019 ,014 ,010 ,008 ,006

28.12 27.60 26.53 24.88 22.51 19.54 15.73 11.20 6.81 3.12 1.11 .2l .07 0 0

4 8 12 16 20 24 28 32 36 Al 40 44 A 2 cos (@, e2) 48 52 56 60

.621 ,824 ,590 ,214 -.040 -.I22 -.I08 -.072 -.045 -.028 p.019 -.Ol4 -.OlO -.008 -.006

40 6 80 IO 118 14 154 18 187 22 215 26 238 30 254 34 264 38 268 42 270 46 270 50 270 54 270 58 270

~ ,319 ,445 ,334

~ ,175 0 0 0 0 0 ,069 ,021 ,005 ,001

FIG, A- I. Relations between complex numbers rl and r2.

-fl(r) fT(f

+ 711.

(A- 10)

Thus, complex cross-correlationis composed of four cross-correlationswhich can be computed in a normal manner. If both traces are identical, then the crosscorrelation function is real at zero lag. Crosscorrelation can also be expressed in polar form $(T) = ~&(+M
T

and we can now show the same proportionsby dividing real and imaginary parts by the modulus

6 F,
=

+ 7).

IF,F,I [

(flf2 + fTf$) u: +fT) VI + fZ2)

(fTf2 f .E) 1 I

. {cos[O,(t)

e,(r

7)] +I (f+fT (ff+f: ) )

(A- 12)

sin[O,(t) - O,(r + T)]}.

(A-

11)

We can also show that

F2 F, = (fib + f fi+) T .i (fTf2 f 2*); J

Semblance In the product F,G =A,A,[cos(B, - 0,) +J sin(0, ~ e,)], we can consider that the real part consistsof the product of the modulus of one of the componentswith the projection of the other onto it (Figure A-l). Similarly, the imaginary part is the product of the modulus of the one with the vector component ot the other which is 90 degrees out of phase. In rectangular coordinates, we can write the product in the form

therefore, F,F, + F,F, = (f;f; + fff;), which is real. Note also that F, E = (q F.). Consequently, if we compute the sum of all possible pairwise crossproductsbetween N complex numbers, the result will be real

Complex Seismic Trace Analysis Therefore. the average of the in-phase portion cbais 7 and
/N \o N A1 .I;

1063

zff;;*"+ 2
k=l

lx
k=l

N-l

(A- 14) 64 corresponds to averaging the cross-correlation coefficients between real-valued time series. Note that this equation is for one sample out of each complex time series. Equation (A-8) expressed the ratio between input and output power computed by summing N traces. Let us consider the terms in the numerator, which are squaresof sums of real and imaginary parts of a trace. We know that
2 = $.f:
k=l

Semblance is defined as the power of the sum divided by the average power of the components of the sum (Taner and Koehler. 1969; Sheriff, 1973). Consequently we can compute the semblance coefficient o for a complex time series as

(A-15)
where 1 - ------das1.0 (NI)

+ 2 r %
k=l

i
m=k+l

fkf,n,

APPENDIX B COMPLEX TRACE EXAMPLE OF RICKER WAVELET Let C(w) = (2/rf)1 c02e+ , 2 2 4(w) = 0 in equation (I 1); this defines a Ricker wavelet. Then f(t) = (2,*)r~a~~ 0 e-w212 cos wt do where (2z) = {: F.m(yi _

. .

3) for m ~ 2

= (1 -

t2)

e--t2/2.

The constant factor (2/ n)l in B(w) was chosen so * that f 0) = 1. The conjugate trace is given by (
m

F*(r)

w2e--w212 sinwtdo=

2(2/4112t

The maximum value of B(w) is attained for w = ~ radians per unit time By a suitable choice of the 2 unit of time this maximizing value of o can be made equal to any desired frequency. if WC take the unit of time as 50~-(2)- ~ set, the maximizing frequency / is 25 Hz. Graphs for such a Rickcr wavelet are shown in Figure 5 and data are listed in Table 1. (Since the wavelet is symmetrical, only half of the wavelet is listed.)

p E,

G=!

t2m+l

e-t2/2

(2m + l)!