Seismic 21
`
`565
`
`roll with regard to the assigned numbers. the line numbering
`direction (high to low or vice versa). and the station number-
`ing direction within each line.
`
`QC monitor considerations
`With a single line system, selection of a channel or group
`of channels for monitoring could be handled with reference
`to station number or system channel. The 3-D system must
`refer to a station with simply the line number and station
`number. To identify a remote unit, for QC testing or parame-
`ter checks (i.e.. battery voltage). the reference is by line
`number. high or low side of the LIU, and count from the
`LIU.
`Display of the line configurations is done a line at a time
`For example, line 2 may be selected as the monitored line
`which will show the following parameters:
`
`Look behind-Location of the last active geophone
`First channel used-Ground station of first channel of the
`selected group (of line 2)
`Gap location-First and last ground stations of the gap
`Shotpoint location-Ground station at SP or next lower if
`off line
`Recording truck location-Ground station of the recording
`truck
`Last channel used-Last ground station of the selected
`group (of line 2)
`Look ahead-Ground station of the farthest remote unit
`ahead of the recorded group (up to 72 stations may be
`selected)
`
`Extended header information
`A 3-D system must maintain a large record of the line
`configurations and current recording parameters. These data
`must be passed to the data processing center via the seismic
`data tape header. To identify the 3-D recorded data fully
`additional information is inserted in an extended header.
`Data requirements include truck flag number, SP reference
`number, SP location (line and channel reference), number of
`lines, number of Aux channels, multiplex location of the first
`Aux channel, Aux channel parameters (gain, notch, low cut,
`and alias), shorted channels (station number), and detailed
`data about e&f line as follows: CDP step position, number
`of channels, first channel. mux of first channel, flag number
`of first channel, gap size, gap starts after channel, flag
`number offirst gap position, number of look-ahead channels.
`mux of first look-ahead channel, flag number of first look-
`ahead channel, and number of channel sets and parameters
`for channel sets.
`
`Conclusion
`Operation of a multiline 3-D system includes the require-
`ment for identification of every line, SP, and station in the
`survey. With the selection of an appropriate numbering
`scheme, the seismic system will aid the observer by provid-
`ing automated SP and line management. In addition, the
`fiber optic telemetry system discussed will supply all 3-D
`parameters to an extended tape header for use by the
`processing center.
`
`S21.5
`
`The Reality of Trace Binning in 3-D
`Marine Surveying
`W. R. Cotton and J. I. Sanders, Geophysical Service Inc.
`Three-dimensional marine seismic data collection involves
`recording swaths of subsurface reflections from a suite of
`closely spaced parallel lines. The method is necessarily
`complicated because of tides and ocean currents affecting
`navigation of the seismic vessel and the recording streamer.
`A variety of problems may arise which could seriously
`devalue the quality of the final product and these are
`discussed together with ways of compensating for them.
`Two approaches to gathering marine 3-D data have
`emerged: “regular binning” and “dynamic binning”. The
`development of these techniques is described and relative
`merits discussed. Once collection is complete, the process-
`ing geophysicist has some opportunity to minimize adverse
`features which have survived and may regather the data in
`an optimum fashion.
`
`Introduction
`In any 3-D seismic survey, subsurface reflecting horizons
`must be sampled spatially in two orthogonal directions.
`Processing of recorded data is greatly simplified if the spatial
`sampling interval, like the temporal sampling, is constant
`although the spatial interval may be different between the
`two orthogonal directions.
`
`Operational problem
`The marine environment is nonstationary. Tracking a line
`of preplot positions over the sea floor is like aiming at a
`moving target. The regularity of reflection points in space is
`dependent on the success of the navigator in compensating
`for the motion of tides and ocean currents. The problem is
`further complicated by the common-depth-point (CDP) stack
`process which at present is a necessary precursor to the
`migration operation. CDP stack is an approximation which
`breaks down further if the seismic cable does not track along
`the preplot line. In the presence of a cross current, the
`seismic cable streams along a track which is the resultant
`vector of the velocity of the towing vessel in the direction of
`the preplot line and of the cross current. Instead offorming a
`series of superimposed depth point locations, the system
`generates a swath of individual depth points.
`An approximation to the CDP gather process is made by
`defining a small rectangular area called a “bin” and gather-
`ing as a CDP set all traces which fall within it. Where cable
`feathering is minimal, the bin gather resembles a CDP gather
`but where feathering is large, the density of traces within a
`bin is reduced, often seriously. The gather may be further
`complicated by the addition of traces overflowing the loca-
`tion from adjacent lines. Hence, techniques of binning
`swaths of 3-D marine data have been developed and specifi-
`cations for trace assemblage requirements formulated.
`
`Binning techniques
`3-D methods and swath shooting techniques were devel-
`oped on land prior to marine because of the easy solution to
`the navigation problem. The first marine surveys therefore
`used the land technique of swath shooting and gathered
`traces into a set of regular rectangular predefined bins,
`except that a single boat operation was involved for simplic-
`ity and economy (see Figure 1). The seismic vessel is rigged
`
`Downloaded 02/05/14 to 207.119.135.157. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`1
`
`ION 1040
`
`

`

`566
`
`Seismic 21
`
`shooting line
`
`FIG. 1.
`
`conventionally as for 2-D with the addition of magnetic
`compasses in the streamer, a tow point orientation device,
`and an electronic interface to log the output from these units
`and the ships gyrocompass on the navigation tape.
`The line spacing was initially 200 m with the shotpoint and
`depthpoint interval of 25 m, although the bin dimensions
`were 50 m in-line and 100 m cross-line, that is, half the
`preplot line spacing. The term “half-line binning” is applied
`to reflect this relationship. These early surveys enjoyed only
`a qualified success. The coarse line spacing limited to low
`values the frequency at which data could be migrated in the
`cross-line direction without inducing aliasing effects.
`Streamer feathering angles are often inconsistent in both
`magnitude and direction so that the trace assemblage in any
`suite of bins was highly irregular leading to a very variable
`stack response. Occasionally, a line of bins was devoid of
`traces completely. The next step was to close up the line
`spacing and match the cross-line bin dimension exactly to
`this interval.
`A greatly improved quality of 3-D migration results be-
`cause the bins contained near traces, most if not all offsets
`and were dominated by traces from the “prime” line, that is,
`the line being processed. This “rule of three” proved to be
`an important requirement in 3-D migration. So far, the bins
`described have been static or “regular” bins which were
`essentially determined prior to data collection. The rule of
`three could be optimized if the bins were created after data
`collection and forced to include near traces close to one end.
`Then by extending the bin in the direction of feather, a
`continuous range of offsets from the same line would be
`captured. Only in cases of large feather angles would far
`traces be lost and these could be replaced by far traces
`overflowing from an adjacent line if cross currents were
`favorable. This form of binning is termed dynamic binning
`and neatly addresses the problem of binned trace assemblage
`since comparison to a conventional stack is significantly
`improved.
`So far, however, line spacing has not been addressed.
`Each stacked trace submitted to a migration process requires
`an assigned pair of location coordinates. These may be
`derived with varying degrees of precision in numerous ways.
`An initial approach was to average the location coordinates
`of the traces within a bin and assign those means to the
`stacked trace. This method acknowledged unequal spacing
`of the bin-stacked lines (see Figure 2). and through the use of
`a Kirchhoff integral migration routine, maintained the integ-
`rity of lateral offsets throughout. Alternatively, if regular
`binning is applied, the coordinates of the bin centers could
`be assigned to the stacked trace positions and the regular
`grid of stacked data is readily amenable to migration using
`the finite difference or F-K transfrom methods.
`
`FIG. 2.
`
`Depth-point steering
`Two approaches have emerged, one based on a regular
`grid of bins but lacking in consistent trace assemblage, the
`other utilizing dynamic bins to obtain a consistent trace
`assemblage but lacking in a consistent and regular line
`spacing. The necessary improvement to both methods is
`achieved essentially the same way. through the introduction
`of real time on-board binning, and depth-point steering.
`In the case of a grid of regular bins, an on-board computer
`keeps track of trace positions as they are created and logs
`changing trace content of bins. The vessel is steered to
`obtain a maximum trace assemblage which includes near,
`medium. and far offsets and involves most of the range
`available. Where feathering is severe, several passes are
`made to steer a different range of offset traces into a line of
`bins to achieve certain minimum specifications. A high level
`quality assurance display system is necessary to ensure that
`good coverage is being acquired and to expose areas requir-
`ing further work.
`In dynamic binning, the vessel is also steered off-line in
`the presence of cross currents. The on-board software sets
`up the bin, computes the trace assemblage so contained, and
`determines the mean trace location. This location is dis-
`played and the vessel steered to place it on the preplot line
`(see Figure 3). A regular stacked line spacing suitable for
`input to 3-D migration is obtained. The software also logs the
`trace assemblage within each bin and any traces which may
`overflow into the bins of adjacent lines. Ultimately, both
`methods produce the desired result which is a plan of evenly
`spaced, consistently filled bins of traces spanning a full range
`of offsets.
`
`Processing problem
`The data processor has no opportunity to reshoot or to
`correct imperfect collection. However, he does have an
`
`FIG. 3.
`
`Downloaded 02/05/14 to 207.119.135.157. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`2
`
`

`

`Seismic 21
`
`567
`
`opportunity to minimize collection anomalies and an obliga-
`tion to avoid processing induced anomalies. The object of
`the 3-D survey is the production of a valid 3-D migration
`supporting interpretation and lithologic analysis. It is useful
`to analyze some collection and processing anomalies vv hich
`detract from optimum migration.
`
`Amplitude anomalies
`Unusual noise and erroneous stacking velocities may
`induce or fail to diminish amplitude anomalies. A 3-D
`prospect is susceptible to several unusual sources of ampli-
`tude anomaly. The limits on the dimension of processings
`bins create a situation where a regular bin structure may not
`have enough traces to produce respectable amplitude re-
`sponse. One of the realities is that an adjacent bin will
`usually have excess contributors, compounding the anoma-
`ly. The CDP method depends on a continuous distribution of
`offsets among contributing traces. One of the most damaging
`situations is a trace bin composed of a few near and a few far
`traces with nothing between. This results in a poor stack
`response of primaries. and aliasing of multiples. Redundant
`contributions from the same or similar offsets causes the
`same problem. A lower level anomaly may occur when the
`data to be stacked are from different recording lines. Differ-
`ences in sea state, streamer or source configuration all lead
`to irregularities in the set of traces. Additionally a problem
`may be induced by “smear” across the collection bin.
`
`time anomalies
`Again all the standard nemeses of the seismic processor
`must be comprehended by the 3-D marine processor. The
`realities of any binning or collection method are anomalies in
`the relative location of one line of to the next. A location,
`spacing, or navigation anomaly will manifest itself as a time
`discontinuity in sampled wave fields. Since the processor
`has control over bin placement and size, to some degree he is
`able to counteract some collection anomalies. Perhaps the
`most important consideration is the presence of near offsets
`within the bin. A lack will cause stacked primaries to be
`unusually sensitive to velocity variations, perhaps creating
`irreconciliable time anomalies. If the recording vessel is
`operating with limits on deviation from the preplanned line.
`near traces are usually available. The binning algorithm
`should then complement the binning technique employed in
`the field.
`Another consideration in bin size and placement is “prime
`line precedence”. A swath of depth points from the prime
`shooting line is preferred. The gather of traces in a process-
`ing bin should begin with the near trace from the prime line
`and continue selecting traces within the bin from that line
`until midpoint locations creep beyond bin edges. The next
`potential offset is selected within the bin from a secondary
`line if found. This process continues selecting data from
`continuous swaths of coverage until the maximum fold
`criteria is reached, or potential contributors are exhausted.
`Since a feathered cable implies a continuous swath of
`midpoint coverage, it is advantageous to place the near-trace
`midpoint near one edge of the bin. The reality is that it works
`well if there is little feathering, but may be quite poor for
`large, directionally variable feathering. An alternative for
`variable feathering situations is to place the near-trace
`midpoint at the center. Thus the computer process may
`smooth the induced spacing irregularities.
`
`The width of the bin may be chosen slightly larger than the
`prescribed line spacing; about 20 percent. This will minimize
`potential gaps in wave field sampling due to cross-line
`adjustment of bin centers.
`
`Field example
`Figure 4 illustrates the dynamic binning procedures. The
`intended stacking geometry required CDP traces at 12.5 m
`by 75 m. Figure 4a is a map view of one processing bin. The
`bin was centered near the short offset midpoint due to
`variable feathering. The shot spacing was 25 m; therefore,
`every other offset is expected within a 12.5 m bin. The
`midpoint locations are labeled only with the original trace
`numbers and shooting line. Traces 95 through 79 were
`gathered from the prime line, 76 through 42 from the next
`line south, and finally 41 through 2 from the third line. Note
`that location irregularities due to compass and depth trans-
`ducer sections affected the trace number sequence. The
`geometric CDP location and the bin center are also indicat-
`ed. The geometric location would have been the center of a
`regular bin.
`Figure 4b is a display following conventional NMO. There
`are three distinct sets of event times and differences in noise
`content. This is typical of binned data. Figure 4c is the same
`data, but with 3-D NM0
`including location corrections
`based on a 3-D time and velocity model. The event mis-ties
`are almost totally absent between 1 and I .5 sec. At about 2.0
`set the correction is not as evident. attesting to inaccuracies
`of the model or noise content. The 3-D NM0 operation
`incorporates variable stacking velocities due to azimuth and
`time variant statics to account for location. In this case the
`location correction moved the data to the geometric CDP
`location to enhance migration.
`The CDP method is to some extent self-correcting. Veloci-
`ty derived for stacking will automatically compensate for
`some smear. Therefore, shot position has a large effect on
`determining final event time
`
`CONTRIBUTING LINE
`
`CONlRlBUllNG LINE
`
`0 2g249
`
`-L""
`
`.I
`
`4.
`
`‘0
`
`.e .
`
`,’
`
`.
`
`.
`
`.
`
`“0
`. .
`
`.u
`
`.-1
`
`._
`
`.-
`
`’
`
`l
`
`..~
`
`.,,
`
`_.
`
`1’
`IL. .”
`BINCENTER
`ti.v
`
`**
`
`1.
`a,,
`
`_GEOMEl'RIC
`COP
`LOCATION
`
`SHOOTING
`DlRECTlON
`
`l - LINE 248
`
`.m LINE 249
`
`2.5
`
`FIG. 4. (a) Trace bin for 75-m line space by 12.5-m CDP
`space with 3 contributing lines. (b) Binned after standard
`NMO. (c) Binned after 3-D NMO.
`
`Downloaded 02/05/14 to 207.119.135.157. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`3
`
`

`

`568
`
`Seismic 21
`
`Problems
`The most common problems experienced are missing
`coverage (obstacles, etc.) and navigation errors. There is no
`perfect method for generating data where coverage is absent.
`Losses may be minimized using an F-K transform incorpo-
`rating a resampling of K-space and an inverse transform. The
`operation would fill some gaps in the coverage, but at the
`loss of spatial bandwidth. Later methods use a subsurface
`time model to perform discrete location adjustments to
`recorded data and a weighted composition of adjusted traces
`adjacent to a gap. This will provide continuous sampling of
`the wave field, accurate to the extent of the model.
`
`Navigation induced anomalies
`The nature of a 3-D prospect requiring even, consistent
`sampling of the seismic wave field provides the solution to
`navigation errors if they should occur. Routine quality
`control products include cross-line displays. Ideally these
`should have the same appearance as the in-line displays.
`Undiscovered navigation errors will likely be uncovered.
`Further investigation may use horizontal displays which are
`very sensitive to time anomalies.
`
`Conclusions
`3-D marine collection procedures should be complement-
`ed by processing procedures. This is becoming increasingly
`important as more contractors gain 3-D capability and col-
`lection and processing contracts are divided between service
`companies.
`
`S21.6
`
`Dual Vessel 3-D Marine Surveys
`R. LeJlaive, C. G. G.
`Various sorts of obstacles such as production platforms
`may hinder the conventional acquisition of 3-D marine data.
`The dual vessel technique may then be applied. Continuous
`monitoring of the acquisition through RTB (real time bin-
`ning) has been extended to this method. This paper deals
`with the problem of the accuracy of the relative positioning
`of the seismic source and the captors and its effect on the
`seismic quality data. The first arrivals on records resulting
`from one boat operation are first analyzed, so as to create
`maps of superficial refraction markers under the sea floor.
`Statistical comparisons are then made with first arrivals from
`2 boat recordings, using coordinates of sources and captors
`deduced from radio positioning and cable drift measure-
`ments. Conclusions are drawn relating to the possible smear-
`ing effect of inaccurate binning of the data and to the
`maximum resolution of the method.
`
`In offshore oil production, a logical sequence of operations
`would include a 3-D seismic survey after each successful
`exploratory well on a new structure. In real life this is not
`always accomplished, so that the need for a 3-D survey is
`often felt at a time when obstacles like production platforms
`are permanently installed. Furthermore, in many productive
`areas, the obstacles relating to producing fields other than
`that to be studied are close enough to present navigation
`problems for seismic boats towing a 2-mile long streamer.
`An obvious solution to this problem is the use of separate
`
`source and recording vessels, The lateral offsetting of the
`source vessel allows for investigation of those CDP strips
`presenting obstacles. An additional advantage is due to the
`fact that the source vessel can safely get closer to the
`platforms than the cable towing vessel.
`Theoretical calculations. based on statistical studies of
`one-boat operations, showed that the accuracy with which
`the relative position of the two boats and the cable is
`measured should be sufficient for the proper binning of each
`trace and for the NM0 correction within the most energetic
`passband of the seismic spectrum.
`Appropriate radio equipment must be installed on board
`both vessels, so as to transmit in real time the radio
`navigation data, as well as the shooting time and the source
`signature (if necessary for processing).
`For continuous monitoring of operations, the RTB (real
`time binning) system as described in a previous paper
`(Regnaudin et al, 1981) was extended to dual vessel opera-
`tions. Coordinates of the midpoints of each trace are calcu-
`lated by the RTB minicomputer and displayed on the color
`RTC according to a code showing the number of traces
`which fill each line of bins in the grid. In some cases, if the
`shooting rate is not too high, the acquisition may be conduct-
`ed alternately from the two vessels. The progress of the two
`CDP lines is visualized simultaneously on the screen.
`A few 3-D surveys were conducted using this technique in
`the Gulf of Guinea and the Gulf of Mexico; corresponding
`data are being analyzed in order to check the acceptability of
`the above assumptions. First the water breaks and first
`arrivals on short traces are analyzed in order to check the
`accuracy of the relative positioning of the two vessels. The
`figures are compared to that derived from the radio position-
`ing data. This confirms that the relative position is known
`with a standard deviation equal to %&times that of each
`individual positioning, which with Syledis on platform in the
`Gulf of Mexico amounts to 2 m.
`Over the entire survey area, the first breaks of single-boat
`records are analyzed. Several refraction markers are identi-
`fied below the sea floor. Maps of velocity and delays are
`established for each of these markers. The first breaks of
`two-boat recordings are compared to what would have been
`their time on single-boat operations. For this purpose, the
`first arrivals are selected on a sample of corresponding
`records, depending on their offset as calculated from radio
`positioning data and magnetic cable drift measurements.
`Histograms of distance differences are established versus
`various parameters such as lateral offset, longitudinal offset,
`and regional situation on the grid. Conversely. CDP gathers
`are established encompassing both single and dual boat
`recorded traces and corrected for refraction velocities. This
`enables us to evaluate the smearing effect of relative posi-
`tioning inaccuracy for offsets greater than those of single
`boat operation. The problem of the fine binning of the data is
`addressed in view of the above analysis, and a limit is set
`accordingly to the minimum size of binning, depending on
`the cable length.
`The accuracy of the NM0 corrections is discussed: here
`again, a limit is set to the high resolution capacity of the
`method. However. if the direct use of the distances mea-
`sured by radio positioning and cable drift leads to a substan-
`tial fuzziness of the stack, an attempt can be made to use the
`variations in time of the first breaks in computing the NM0
`corrections.
`
`Downloaded 02/05/14 to 207.119.135.157. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`4
`
`