Sie sind auf Seite 1von 9

GEOPHYSICS,

VOL. 56, NO. 1 (JANUARY

1991); P. 9-17. 4 FIGS., 4 TABLES

An example of simultaneous recording where necessary signal separation is easily achieved

William C. Pritchett

ABSTRACT

The key to success in recording multiple data sets simultaneously from multiple sets of sources is being able to separate the multiple data sets satisfactorily. If at a given receiver position the energy level is greatly different from one source than from another source, then we may have difficulty in extracting the weaker trace from the stronger trace. We operated two sets of sources at two adjacent source positions, so that the energy levels from the two sets of sources at any receiver position were not greatly different and satisfactory separation of the traces was relatively easy. Reflections from adjacent source positions as recorded at any receiver position were also similar in that the rays were reflected from adjacent reflection points at depth. Actually, of course, our measurements responded not to rays but to waves which were reflected from rather large areas surrounding the reflection points. Reflections from adjacent reflection points were actually from areas which overlapped substantially; therefore, signals reflected from adjacent reflection points would not have been totally independent

even with conventional recording from one set of sources. Slight, additional contamination of one trace by an adjacent trace because of less than perfect separation of simultaneously recorded data was of little consequence with our technigue. We saved money by speeding up recording, and data quality was excellent. In our application of the technique the reflection targets were rather shallow, but the technique is not limited to shallow exploration. To best meet our exploration objectives, we used simultaneous recording to save field time and money. We used nonlinear sweeps in the field along with effective spectral flattening processing techniques to bring out higher frequencies and improve resolution. We used the stack-array technique (Anstey, 1986) to suppress noise in the stacked data better. Some have objected that our technique cannot possibly work, and have shut their minds against discussion and favorable evidence. However, this technique is based upon nothing more controversial than a valid application of the theorem of superposition. Furthermore, the technique was validated by field results including the recording of 73 mi (117 km) of excellent data in 1984 and 1985 at relatively modest costs.

INTRODUCTION

In 1984, we wanted to record high-resolution data at record times from 0.2 to 0.4 s from rather shallow objectives in northern Ohio. For these shallow objectives we knew our maximum effective offset would be quite limited, and thought we would need rather high CMP fold to achieve satisfactory data quality. Thus we would need to use very short intervals between traces and also between source points, which would result in many records per mile (or per

kilometer). We would also have to use rather short geophone arrays and short source patterns to record frequencies of 100 Hz and possibly higher. We planned to work along roads, with five vibrators in our crew. Five vibrators in a row, bumper to bumper, would constitute a source pattern that was far too long even without any moveup. We chose to use only two vibrators, bumper to bumper, per source point with no moveup between sweeps. We used a group interval and source point interval of 66 ft (20 m) with a split spread of 60 traces, giving

Manuscript received by the Editor November 14, 1989;revised manuscriptreceived July 21, 1990. *Formerly Atlantic Richfield Company; geophysicalconsultant, 1109Janwood Circle, Plano, TX 75075 01991 Society of Exploration Geophysicists. All rights reserved. 9

10

Pritchett points A and B. We define Ml as a simultaneous excitation at points A and B with signals of positive polarity. We define M2 as a simultaneous excitation at points A and B with a positive polarity at A but with a negative polarity at B. Table 1 shows that by the addition of two measurements from both sources simultaneously (Ml + M2), we can extract the earth response to a signal input at point A only. By subtraction (Ml - M2), we can extract the earth response to a signal input at point B only. Theorem of superposition

a CMP fold of 30. This meant recording 80 source points per mile (50 per kilometer). We needed to reduce our costs without sacrificing data quality. We decided to try recording adjacent pairs of source points simultaneously using two sets of two vibrators each to save money by recording more VPs per day. Despite some negative reports from others about simultaneous recording, we were confident that our technique would work well in our application.
SIMULTANEOUS RECORDING TECHNIQUE

General theory with polarity coding and decoding Figure 1 and Table 1 illustrate simultaneous recording using polarity coding and decoding. The basic assumptions are that the earth is linear and that the theory of superposition is valid and applicable. is a first signal applied at surface point A. is a second signal applied at some other surface point B. is the response at another surface point 1 to RI whatever signals are applied. R? is the response at another surface point 2 to whatever signals are applied. KAl(f) is the transfer function from a signal at point A a sensor at point 1. KAZ(f) is the transfer function from a signal at point A a sensor at point 2. K,,(f) is the transfer function from a signal at point B a sensor at point 1. KBz(f) is the transfer function from a signal at point B a sensor at point 2.
sA

SB

The theorem of superposition states that in a linear system the response to multiple input signals applied simultaneously is equal to the sum of the responses to the individual signals applied separately. The theorem of superposition has been used extensively in many fields by physicists and engineers for more than a century. Furthermore, geophysicists often use the theorem either knowingly or unknowingly. Examples of its use in seismology include: (1) forming source patterns by separately recording from one or more surface sources at multiple positions and then summing the set of recordings, (2) using Nigel Anstey (1986) stack-array concept s to suppress source-generated noise, and (3) recording with an ARIS@ source inclined in a vertical plane perpendicular to the seismic line and reversing the angle of inclination and recording again. With the inclined ARIS source, we impart a vertically downward force on the earth both times it is fired. Reversing the angle of inclination reverses the direction of the transverse, horizontal force on the earth. When we sum the two records, the transverse, horizontal forces largely cancel and we obtain the compressional wave plus converted, radial shear-wave response of the earth. When we subtract one Trademark of the Atlantic Richfield Company.

to to to to

For any given frequency component of the input signals, these transfer functions will be of the form K sin (wt + $), where K and + are constants. These transfer functions will be functions of frequency if we consider input wavelets with amplitude and phase spectra. Table 1 lists responses at points 1 and 2 to individual excitations and to simultaneous excitations by signals at

.
FIG.

.
1. A general illustration of simultaneous recording.

Surface of the Earth

Table 1. A general illustration of simultaneous recording. Input signals Response RI Response R2 Remarks

+SB

+sB[K~,(f)l -sB[K,,(.f)l + SB =

+SB[KB#-)l -sB[KB,(J )l

Ml (simultaneously) - SB = M2 +sA (simultaneously) Ml + M2 Ml - M2

-sB +sA

+SA[KA,(~)I

s,[K,,(f)l SB[KBI(.I?I

+SA[KA2(f)l +SA[KAZ(J )I

+ s,[K,,(f)l sB[K~z(.f)l

+SA[KA~(~)I +~SA[KA,(~)I +2S,[K,,(f)l

+2S,[K,,(f)j +2S,[K,l(f)l

Theorem of superposition Theorem of superposition Response to SA only Response to Sa only

Simultaneous

Recording from Pairs of VPs

11

record from the other, the transverse forces are additive, the vertical forces largely cancel, and we obtain the horizontal, transverse shear-wave response of the earth. This technique with the inclined ARIS source is similar in many ways to the technique we used for simultaneous recording, except that the polarity reversal of the horizontal, transverse shear wave is accomplished by a reversal in the angle of inclination of the source rather than by a reversal of the polarity of a sweep. The theorem of superposition is valid for all of these examples, and we applied it properly to the analysis of our technique. Specific technique used Figure 2 and Table 2 illustrate the technique we used to record data simultaneously from adjacent pairs of source points. We recorded 60 traces simultaneously from each of two source points (total of 120 signals). We used two vibrators at each A position and two more simultaneously at each B position (a total of four vibrators). After each series of sweeps, both sets of vibrators moved up two source positions and the roll-along switch was advanced two positions before recording the next pair of source points. Our vibrators were equipped such that we could reverse the polarity of alternate sweeps. Table 2 shows how we used the polarity reversing equipment for coding and decoding the simultaneously recorded signals. The B set of vibrators used the polarity reversing option, whereas the A set of vibrators did not. After the fourth sweep of each set of eight sweeps, we reversed the polarity of the power line cancellation box (COS system marketed by Mountain System Service) before proceeding to cancel power-line interference for both sets of simultaneously recorded signals. To extract the traces from the A set of vibrators, we simply summed the data from all eight sweeps. To extract the traces from the B set of vibrators, in processing we reversed the polarity of the data from sweeps 2,4,6, and 8 before summing. We recorded and stored data from individual sweeps to extract the data from

the B set of vibrators. We also summed in the field to extract the data from the A set of vibrators which gave excellent monitor records for quality control. Before the field work, we were absolutely confident that our technique would work, but we were quite aware of concerns of others, including the supervisor of our contract crew. These concerns centered on the fact that the vibrator pad of the forward vibrator of the A set was only 33 ft (10 m) from the pad of the rear vibrator of the B set. On even numbered sweeps the polarities of these vibrators were 180 degrees out of phase in time Those concerned were worried about the strange stresses applied to the earth by the out-of-time phase operation of sets of vibrators set relatively close together. However, the theorem of superposition plus simple arithmetic provided an easy way to analyze the net results, and we simply felt no need to dwell on the stresses. This theorem is not restricted by any limits on the spacing between sources or sets of sources so long as the earth response is linear. Of course, there is interaction between the two sets of vibrators, but linear interaction does not invalidate the theorem. From previous field measurements, we were confident that on reasonably firm surfaces any nonlinearity associated with the vibrator sources would not extend beyond a distance exceeding 1 ft (113 m) from the vibrator pads. Certainly vibrator pads separated by a minimum distance of 33 ft (10 m) would not result in any nonlinear interactions. Why necessary signal separation was easily achieved Our purpose in operating the two sets of vibrators very close together was the resulting ease of achieving the necessary separation of the individual records from simultaneous excitations by the two sets of vibrators. This close spacing made the amplitudes of the responses from the two sets of vibrators approximately equal for any given trace and at any given time Therefore, we avoided the need to extract a relatively weak response from one set of vibrators from a

PosItIonnumbers 90 92 94 96 96 100 102 104 106 106 110 112 114 116 116 120 Position numbers

PPPPPPPPPPPPPPPPPPPPPPPPPPPPPPP 204 0 206 0 2: 0 300 xx xx A0


;"2 0 204 0 206 0 208 0 ; xx xx A0

39

305

307 0

3:

49

0 4:

0 405

VPS numberslor Trace100.5 a 101.5

39

3:

0 305

307 0

3:

4:

0 403

VPS numbersfor Trace102.5 8 103.5

200

0 202

Cl 2:

206 0

208 0

3;

xx xx Al3 vibrators

31 00000000000 33

35

37

39

41

VPS numbersfor Trace104.5 8 105.5

1008101 102 B 103 1048105

l-30 & 31-60 l-30 8 31-60 l-308 31-60

67- 96 8106-135 69- 96 8108-137 71-100 8 110-139

FIG. 2. Illustration of the line layout used to record pairs of VPs simultaneously. The polarity of the sweep for the set of vibrators at A is always positive, but for the set of vibrators at B the polarity is negative for even-numbered sweeps. The figure illustrates only the center portion of a sixty trace cable. Following each simultaneous recording with VPs at A and B the observer advances the roll-along switch two positions and both sets of vibrators move up two positions.

12

Pritchett
METHODS USED FOR HIGH SIGNAL-TO-NOISE AND HIGH RESOLUTION RATIO

relatively strong response from the other set. Also, the reflections from the two sources on any given trace were very much alike because the areas on the reflecting interfaces from which these reflections took place overlapped substantially. Indeed, the reflection points were only 33 ft (10 m) apart as compared to quite large reflection areas on each objective interface. We took half of the radius of the first Fresnel zone as an indicator of the effective reflection area, approximately expressed as O.SRFz = 0.5[(v/4f)2 + rv*/4f1.5, where O.SR,z = the radius of the effective reflection area, u= the average velocity of the earth down to a reflector, f= a frequency component of a reflection wavelet, and t= the two-way reflection time for the reflector. We calculated effective reflection area radii of about 150 ft (46 m) for our shallower objective at 100 Hz and 200 ft (61 m) for our deeper objective at 100 Hz. Obviously the overlap of the reflection areas for both objectives was quite substantial for each trace from the two sets of sources. Since the reflection areas overlap substantially for adjacent traces, the signals from adjacent traces are not independent. Even if half the Fresnel zone radius had been small compared to the distance between the reflection points from the two sources, the reflection areas would still have overlapped substantially because of the length of the geophone arrays and the length of the source pattern. Whenever we utilize the stack-array technique, traces from a given geophone string as recorded from adjacent sourcepoints will not be independent. Instead, reflections on such traces will be from areas which overlap extensively. In a real sense the reflection data from such traces are substantially mixed in the earth. With our simultaneous recording technique any slight additional mixing of the reflection signals due to less than perfect separation in decoding would be of little consequence. Therefore, our simple polarity coding and decoding system was quite adequate.

Methods used to achieve a high signal-to-noise ratio (SIN) and high resolution included (1) use of the stack-array concept (Anstey, 1986), (2) use of a CMP fold of 30, (3) use of a logarithmic upsweep of 25 to 140 Hz to boost the higher frequency components of our reflected signals, and (4) use of a spectral whitening program to flatten the amplitude spectra of our reflections.
Stack array with a CMP fold of 30

Table 3 gives the recording parameters we used. Our field technique took advantage of what Anstey (1986) calls a stack array. Each vibrator point was centered halfway between trace position centers. For each source point the front and back cable offsets were very slightly asymmetrical. Figure 3 illustrates that despite this slight asymmetry, in each CMP gather the offset distances from the front and back cables interlace and the differential offset distance is equal to one group interval. Therefore we met a requirement for the compliance with the stack-array concept. We used a linear array of equally spaced and equally weighted geophones with an effective length equal to the distance between trace position centers. We used a group interval geophonearray to maximize the recording of the higher frequency components of our reflections. In other high-resolution projects, we used a double group interval array despite some loss of higher frequency components in order to suppressmore effectively

Table 3. Recording parameters.

Parameter VP and group interval VPs per mile VPs per kilometer Number of traces Geophone array: Number and type Effective length Source pattern: Number of vibrators Vibrator pad spacing (within each pattern) Moveup (between sweeps) Sweeps per VP Sweep: Length Frequency band Type Cable: Type Gap Offsets for A vibrators Offsets for B vibrators 66 ft (20 m) :: 60 12 in line, equally spaced One group interval 2 per VP (4 per pair of VPs) 33 ft (10 m)
0

Table 2. Coding used to record and separate the earth response to two sets of vibrators.

Polarity of vibrators Sweep no. 1 A B

Polarity for Summing A Record B Record + _ + to the + + -

8 6 s (+ 2 s listen time) 25 to 140 Hz (upsweep) Logarithmic Split with 30 traces to each side 8 positions 2145-231-x-297-2211ft (654- 70-x- 91-674 m) 2211-297-x-231-2145ft (674- 91-x- 70- 654 m)

+ + + + + + + + 4 + + Then we changed the polarity of the power-line cancellation box without triggering any sweeps prior next sweep + + + _ + :, + + + + _ + ;: +

Simultaneous

Recording from Pairs of VPs


RESULTS

13

troublesome, source-generatednoises (Pritchett, 1990, sets. 2.2, 4.2, and Appendix D).
Use of a logarithmic sweep

Asymmetry of the first breaks

A logarithmic sweep (or some other, suitable nonlinear sweep that would boost the higher frequencies substantially) was important for two reasons: (1) to provide enough energy at the higher frequencies without spending an unnecessary amount of field time sweeping lower frequencies, and (2) to suppress ground roll and other noises which were dominated by the lower frequencies. With a nonlinear sweep we soughtto provide an adequate ratio of signal to background noise over the sweep band in the minimum amount of sweep time We needed to provide more sweep time at the higher frequencies to offset higher earth attenuation and to offset inefficiencies in the injection of higher frequencies by our vibrators (Pritchett, 1990, sets. 6.3-6.6, and Appendix J). We did not need to spend nearly as much sweep time at lower frequencies where earth attenuation is quite low and where our vibrators were more effective. As compared to normal practice with linear sweeps, we spent much less sweep time at the lower frequencies. Consequently, we suppressed much of the low-frequency, source-generated noise so that relatively high-frequency reflections from our objectives were visible in the field records. With linear sweeps it is probable that ground roll and other low-frequency noises would have dominated the field records and obscured objective reflections. It is advantageous to be able to see good reflections from objectives on field monitor records for quality control and other purposes such as determining in the field the effective upper frequency limit for our objectives (Pritchett, 1990, sets. 5.1, 6.6, and 7.1).
Use of a spectral whitening program in processing

The first breaks always occurred earliest on trace 30 of the field monitors which were the responses to the vibrators at the A positions. When we extracted the responses to the vibrators at the B positions, the first breaks always occurred earliest on trace 31. This asymmetry of the first breaks corresponded to the geometrical asymmetry which was different for vibrators at the A and B positions. These asymmetrical responses confirmed that we did extract the signals that we wanted and that separation of the two sets of signals was quite adequate.
Data quality achieved and costs

By providing the processors with field data with an adequate signal to background noise ratio over the sweep frequency band, we made it possible for the processor to produce a final seismic section with relatively flat amplitude spectra from 25 Hz to at least 120 Hz (and in some line sections up to the maximum sweep frequency of 140 Hz). We began recording data with nonlinear sweepsin January 1983 on other projects. Immediately we were quite pleased with both the field monitor records and the final processed sections, but we soon learned that even better final sections resulted when processors used a spectrum whitening (flattening) program we had licensed (for our own use) from Western Geophysical Company. The problem with the earlier sectionswas that our processorswere not used to getting data with an adequate signal to background noise ratio at the higher frequencies and they had not pushed their deconvolution programs hard enough to flatten the spectra sufficiently. Our data processor learned to use their deconvolution techniques more effectively with our improved data, but we continued to use the Western program as our standard because it ensured that we got the benefit of all of the resolution available from our data.

Figure 4 compares a short portion of a seismic line recorded in 1984using simultaneous recording with the same line portion recorded in 1985 using the conventional technique with only one set of vibrators. Unfortunately, in 1985 we came back too soon and the comparative data were recorded during the Spring thaw. As a result, the conventionally recorded data are not as good as the simultaneously recorded data. The processing of the two sets of data was essentially the same; however, the processor should have muted the 1985 data more so as to cut out shallow refractions. If the two sets of data had been recorded under nearly identical circumstances, I am sure the conventionally recorded data would have been just as good as, but no better than, the simultaneously recorded data. Table 4 shows the daily progressof the work in 1984. We used a 25 to 140 Hz logarithmic sweep with group and VP intervals of 66 ft (20 m), and recorded 3698 VPs over 46.23 miles (74.4 km) in 154 hours of 14 winter days which included a moderate crew move. We recorded 80 VPs per mile (50 VPs per km) with an overall acquisition cost of $2872 per mile ($1785 per km). We averaged 264 VPs per day and 3.3 miles (5.3 km) per day. At this rate, in a 215 hour crew-month we would have recorded 64.5 miles (104 km) of this very high-density data. For our objectives, data quality (including resolution, bandwidth, continuity, and S/N ratio) was excellent. Without simultaneous recording of pairs of VPs the estimated acquisition cost would have been about $4500 per mile ($2800 per km). The work was done by an excellent crew that we had used continuously for about five years under a monthly contract. The low costs were not the result of a low bid by a contractor who was desperate for work. We also tried to save money in data processingby making an approximate normal-moveout correction and then summing the A and B records which halved the number of traces for further data processing. The sections with the summed data were good, but full-fold processing of the separate traces yielded significantly better results. We tried extracting and applying a static correction before summing the A and B records, but full-fold processing still gave superior results.

Pritchett

A B 0 .

Source positions for vibrators without polarity reversals Sourca positions for vibrators with polarity reversals Trace positions (illustrated with only 10 Vams per record) Midpoints between s~urca positions and traca positions
0 0

. ..

cl

.. .. . .. .. .. . .. .. .. . .. ..
Otfsets in Group Intervals SbIEiL6.5, 6.5. 4.5.5.5. 7.5 9.5, 7.5, 5 5. 6.5. 6.5 6.5. 6.5. 5.5. 7.5, 9.5 7.5, 5.5. 4,5, 6.5, 6.5 6.5, 6.5, 4.5, 5.5, 7.5 9.5, 7.5, 5.5, 6.5, 6.5 6.5,6.5.5.5.7.5.9.5 7.5. 5.5, 4.5, 6.5, 6.5 6.5, 6.5, 4.5. 5.5, 7.5 9.5.7.5,5.5.6.5.65 6.5, 6.5, 5.5. 7.5, 9.5 7.5, 5.5, 4.5, 6.5, 6.5 B
0 0 0

.. .

.. .
A

.. .
A

. B .

.. .. .. .. .. . .. .. .. . .. .. ..
B . . . 9 . . . . A . . B A

.. , .
. . . . .
0

.. .

..
A

A . . . . . B .
0 0

.. .

A
0 0

. . B 0 0 cl

FIG. 3. Offsets in CMP gathers with simultaneous recording. Note that in each CMP gather that the offset distances from the front and back cables interlace and the differential offset distance is equal to one group interval. Therefore the technique does take advantage of the stack-array concept if we use linear, equally spacedand equally weighted geophone arrays with effective lengths equal to a group length (or equal to an integral number of group lengths).

Simultaneous
MINIMIZING FIELD time

Recording from Pairs of VPs

15

Minimum total sweep time per VP Aside from economic and certain other considerations, there is a minimum amount of sweep time required for each VP. There is a practical limit on the minimum sweep length and on the number of sweeps per VP. When we use nonlinear sweeps that boost the signal energy at the higher frequencies, the rate of change of frequency is quite rapid at the lower frequencies particularly if the sweep length is very short. I do not recommend sweep lengths of less than about 6 s for such nonlinear sweeps, but very short sweeps are not very efficient anyway. Since we usually record for a period of a few seconds beyond the end of each sweep, our signal

injection time becomes a smaller fraction of our total recording time when the sweep time is very short (and this low duty cycle limits the advantage of using very short sweeps). With short sweeps we must use very short tapers, particularly at the start of nonlinear upsweeps. The noise edit program for individual sweeps depends upon having enough sweeps that after summation the edited sections of one sweep are filled in with good data from other sweeps. We probably need at least four sweeps per vibrator points for good noise editing. Therefore, with nonlinear sweeps the practical minimum sweep time per VP is the product of about 6 s times four sweeps, or a total of 24 s (plus four times the listening time after each sweep). With linear sweeps we might conceivably use sweeps as short as

SECTIONRECORDEDSIMULTANEOUSLY
0 SEC.

0 SEC.

0.5

1. i-1 MILE-

REPEATSECTIONRECORDED CONVENTIONALLY (UNFORTUNATELYDURINGSPRINGTHAW)


FIG. 4. Comparison of simultaneously recorded data with conventionally recorded data. The section for the simultaneously recorded data (top) was recorded in late 1984. The conventionally recorded data (bottom) was recorded in early 1985 under unfavorable circumstances during the spring thaw. The data processing was essentially the same except the mute on the conventionally recorded data should have cut deeper to eliminate shallow refractions that contaminate the data near 0.14 s. Under identical circumstances the simultaneously recorded data would be as good as, but no better than conventionally recorded data.

16

Pritchett recording might or might not more than offset the cost of the additional vibrators and operators.
CONCLUSIONS (1) This method of simultaneous recording from adjacent pairs of vibrator points worked quite well. (2) With our method of simultaneous recording we were able to use four vibrators effectively along a road even though we needed a very short source pattern and could not operate vibrators side by side. (3) By recording simultaneously from pairs of vibrator points we effectively speeded up recording and saved money. (4) Even if we could have used four vibrators at each position with conventional recording, the data could not have been acquired any faster than we were able to acquire it with simultaneous recording. (5) If we had been able to use four vibrators at each position, we could have then used two sets of four vibrators each with simultaneous recording and increased productivity even more. This is particularly applicable for deeper objectives.

2 s and a total sweep time of 8 s (plus the total listening time). If the listen time were 2 s, then with nonlinear sweeps we would need a minimum of 32 s (four sweeps times the sum of 6 + 2 s) per VP and with linear sweeps possibly a minimum of as few as 16 s (four sweeps times the sum of 2 + 2 s). In our 1984 project in Ohio if we had been able to operate vibrators side by side, could we have worked faster with conventional recording? We could have either reduced the sweep length or reduced the number of sweeps per vibrator point, or both. Our nonlinear sweep from 25 to 140 Hz of 6 s duration was about as short as is practical, however. We could have cut our number of sweeps from eight to four and still have had an effective editing means for suppressing noise bursts. Therefore, with conventional recording with four vibrators per VP we could have achieved about the same recording rate as with two sets of two vibrators and simultaneous recording of pairs of VPs. If we could have used two sets of four vibrators per VP and simultaneous recording, we could have reduced our total sweep time per VP and reduced the total field time somewhat. Optimum vibrator power per VP In projects with deeper objectives we can often use four or more vibrators effectively at each vibrator point. By doubling the number of vibrators at each vibrator point, we might be able to use one-fourth as much total sweep time for the same ratio of signal to background noise. Under these circumstances we cannot decrease field time by a factor of four because of layout time etc., but often we can reduce field time significantly. So where it is feasible and practical to use more vibrators or more powerful vibrators per vibrator point, we should consider the economic benefits versus the economic costs. Where we can effectively use four vibrators per position, it would be counterproductive to split four vibrators into two sets of two each and use simultaneous recording. However, in such cases we should consider using two sets of four vibrators per set with simultaneous recording from pairs of vibrator points. The increased rate of

Beginning in 1984 several papers presented at technical meetings reported disappointing results with simultaneous recording. Why were we so successful? In devising a solution to our problem we provided ourselves with two advantages. First, our geometry and recording system were such that the amplitude of the combined signal plus the sourcegenerated noises were nearly the same from both sources at all receivers. This was true at early record times when the energy level was relatively high and still true at late record times when the energy level was quite low. Therefore, the separation of the two sets of traces (of signal plus sourcegenerated noise) was less difficult than for cases where one signal plus noise is much larger than the other signal plus noise at some time during the recording. Second, our geometry was such that the two signals were very much alike in

Table 4. Production rates for simultaneous recording of pairs of VPs.

Day

No. of VPs 248 262 290 240 302 226 112 338 300 182 354 284 300 260 3698 264

Miles 3.1 :::,

Km 5.0 5.3 Z

Crew Hrs 11.5 11.5 11.5 11.5 11.5 10.0 11.0 11.0 11.0 9.5 11.0 11.0 11.0 11.0 154.0 11.0

Milhr 0.27 0.29 0.32 0.26 0.33 0.28 0.13 0.38 0.34 0.24 0.40 0.32 0.34 0.30 0.30

km/hr 0.43 0.47 0.51 0.42 0.53 0.45 0.21 0.61 0.55 0.39 0.64 0.51 0.55 0.48 0.48

14 Totals Daily avgs.

:::: 2.8 1.4 4.2 3.8 2.3 4.4 3.6 3.8 3.3 46.2 3.3

2:: 2.3 6.8 6.0 :.: 5:7 6.0 5.2 74.4 5.3

Notes: Laid out line 2 and began recording on 1215184.Cable was laid off road on 12/g/84. End of line 2 and start of line 1 on 12/l l/84. Rain on 12114184.End of line 1 and picked up on 12/18/84.

Simultaneous

Recording from Pairs of VPs

17

that the centers of the reflection points were only 33 ft (10 m) apart; consequently, there was substantial overlap between the two areas from which the two signalswere reflected from the various interfaces of interest. Therefore the separation of the two signals did not have to be perfect. If the A signal were slightly contaminated with the B signal, it would make very little difference. Use a good tool the right way and it works. Use it carelessly and you may get hurt.
ACKNOWLEDGMENTS

Richfield Company for supplying the data and for their permission to publish this paper.
REFERENCES

Anstey. N. A.. 1986, Whatever happened to ground roll?: The


Leading Edge, 5, no. 3, 40-45. Deluchi, L., M. R., and Werner, H., 1987.3-D seismic survey of the Gaggiano Oil Field with quatemary encoding and dual-source Vibroseis technique: Presented at the 57th Ann. Internat. Mtg., Sot. Expl. Geophys., Expanded Abstracts, 570-573. Garotta, R., 1984, Simultaneous recording of several Vibroseis seismic lines: Presented at the 54th Ann. Internat. Mtg., Sot. Expl. Geophys., Expanded Abstracts, 308-310. Khan, T. A., and Olson, E. Lynard. 1987, Field productivity: a challenge to manufacturers of field systems: Presented at the 57th Ann. Internat. Mtg., Sot. Expl. Geophys., Expanded Abstracts, 576576. Pritchett. W. C., 1990, Acquiring better seismic data: Chapman and Hall, Ltd.

James B. Peterson contributed substantially to the planning of this project and to instructing the excellent field crew in the implementation of the technique. I thank the Atlantic

Das könnte Ihnen auch gefallen