`
`
`Ex. PGS 1075
`
`
`EX. PGS 1075
`
`
`
`
`
`
`
`291
`
`4
`
`FIG. 3. Stack section of
`data acquired with 400-
`channel sign-bit record-
`ing system.
`
`sweeps techniques (Varisweep and higher-swept bandwidth). The
`sign-bit section has superior shallower data due to no gapping in
`the receiver line and closer channel spacing.
`In conclusion, we compared conventional and sign-bit record-
`ing systems by acquiring data with identical recording parameters
`and with parameters optimized for each system, then compared
`the resulting identically processed stack sections. The sections
`acquired using identical parameters indicates for Vibroseis data
`the results achieved are equivalent. Optimizing parameters for
`each system shows the 400-channel sign-bit instmment gives bet-
`ter temporal and spatial resolution.
`
`Reference
`BNIK, R. H., Hays, D., Sixta, D. P., and Schneider, W. A., 1982,
`Comparison of sign-bit and conventional seismic recording in Eastern
`Colorado: Presented at the 52nd Annual SEG Meeting, Dallas.
`
`POS 2.16
`
`Acoustic and Mechanical Design
`Considerations for Digital Streamers
`L. Peardon, Britoil PLC, Scotland; and N. Cameron, STC Ltd.,
`England
`In the quest for enhanced seismic resolution it is important that
`acquisition keeps abreast of processing. If the bandwidth is not
`present in the recorded field data then no amount of conventional
`processing can retrieve the missing information. It is true that
`certain data inversion techniques can achieve resolution beyond
`the bandwidth of the recorded data but these rely on additional
`information such as well log data. Even these techniques can
`benefit from improvements in the quality of recorded data. To
`this end, both source and receiver technology must be improved.
`In this paper we address the latter problem.
`Three basic criteria for enhanced resolution at the recording
`stage are short group length, small group interval, and low noise
`thresholds. Short group lengths are required to preserve the all-
`important high frequencies. A small group interval is necessary
`to avoid spatial aliasing of low-velocity coherent noise and re-
`fractions which would otherwise contaminate the signal high-fre-
`quency components. High-velocity and incoherent noise must be
`kept to a minimum in order to maintain a reasonable signal-to-
`noise ratio at the higher frequencies. The requirement for a small
`group interval leads to increased channel density, i.e., more
`
`channels per unit length. For example, to maintain a group inter-
`val of 6.25 m in a 3 km streamer requires 500 channels. The
`inherent limitations of analog streamers led to the advent of dig-
`ital streamers employing telemetry communications with the
`shipboard recorder. The ultimate objective of these systems is to
`record alias-free data with good high-frequency content. This
`would prove fruitless if the benefits were destroyed because of
`high noise levels due to poor design.
`In this paper we consider the various noise mechanisms that
`affect a digital streamer and discuss ways in which noise suppres-
`sion may be incorporated either in the design of the streamer or
`in processing data acquired. This is illustrated with data acquired
`from a digital streamer during sea trials.
`
`Introduction
`The trend for seismic streamers is toward increased channel
`capacity. This led to the advent of digital streamers because of
`the inability of analog systems to handle very many channels. It
`is important that the benefits of increased sampling are not offset
`by a degradation in performance due to high noise levels. More-
`over, streamer noise sets the ultimate limit for resolution and,
`therefore, good streamer design is essential. The purpose of this
`paper is two-fold: It presents a detailed analysis of the types of
`noise we expect to see and, in particular, the mechanisms by
`which these noise trains are originated and transmitted in the
`streamer. Secondly, using this analysis we propose guidelines for
`good streamer design. We take cognizance of the fact that some
`excellent papers have been written on streamer noise (see refer-
`ences) but they are with respect to analog systems while this
`paper is biased to digital streamers. Although many features are
`common to both, a major factor is the fine spatial sampling af-
`forded by digital streamers which allows more detailed analysis
`of certain noise trains that are not evident otherwise. Acoustic
`data acquired during recent sea trials of a digital system are used
`to corroborate the theoretical principles.
`
`Noise sources
`Following Bedenbender et al. (1970), the major noise compo-
`nents that occur in a marine seismic streamer may be categorized
`as follows.
`Ambient. This consists of all natural sources of noise such as
`wind, wave action, biological interaction, and noise caused by
`
`Downloaded 08/26/14 to 173.226.64.254. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`
`
`292
`
`Imar
`
`o.ow
`
`0.0100
`
`o.ww
`
`o.wdo
`
`o.w40
`
`c ~coYP*ss
`.
`-NODE
`.BIRD
`0
`.OROUP
`.
`
`o.ww
`
`I
`10
`
`I
`20
`
`I
`30
`
`1
`40
`
`I
`50
`
`I
`00
`
`GROUP NUMBER
`
`FIG. 1, Rms flow noise.
`
`proximity to shore lines. This noise is extremely variable and
`dependent upon conditions prevailing at the time It was investi-
`gated thoroughly by Wenz (1962), who gives upper and lower
`bounds which are heavily dependent on sea state. Swell noise can
`be particularly troublesome, causing bursts of high-amplitude
`noise on shot data.
`Radiated. This is due to propellers and engines of the survey
`vessel and nearby rigs, platforms, and other sea-going traffic. It
`can be a major problem in determining true flow noise character-
`istics. Propagation of this noise can be via direct arrival through
`the sea; reflection from the sea floor and sea surface and seabed
`anomalies; refraction from the subsurface.
`Mechanical. Tugging from the ship and tail buoy causes vi-
`brations which are transmitted down the streamer through various
`paths (e.g., strain members, skin, etc.). This is a major compo-
`nent in the acoustic noise levels of a streamer.
`Flow, This is caused by the stream of water over the skin
`which produces a turbulent boundary layer as the streamer passes
`through the sea. Pressure fluctuations in this layer cause noise to
`be induced in the streamer. Excitation caused by protuberances
`such as nodes and birds (Schoenberger and Mifsud, 1974) gives
`rise to particularly high noise levels. This is evident in Figure 1
`which shows rms values of recorded noise along a streamer. The
`correlation between the peaks and the locations of birds and
`nodes is obvious.
`Electronic. This includes all the electrical system noise in the
`streamer, e.g., A to D convertors, analog filters, and the telem-
`etry system. This noise is relatively easily controlled by careful
`
`Poster Papers 2
`
`design of electronic components and on modem systems should
`not provide a major contribution to the overall noise level.
`Figure 2 is a diagram of an F-K plot containing some of these
`noise trains as measured on a digital streamer. Of special interest
`is the ship noise at 1 650 m/s due to refractions through the sea
`bottom. A possible explanation for the ship backscattered noise
`is that this refracted wave was reflected from a fault (Lamer et
`al., 1981). Also of importance is the bulge wave energy which
`is associated with blocking mechanisms within the streamer. It is
`of very high amplitude but is low frequency and low velocity
`and, consequently, readily removed in processing. We may fur-
`ther classify these noise components into external noise sources
`consisting of the radiated and ambient noise fields and “self”
`noise consisting of electronic, mechanical, and flow noise.
`Radiated noise and coherent ambient noise may be attacked by
`data processing either in real timeor subsequent to acquisition.
`If the data rate exceeds the recording bandwidth it makes sense
`to decimate in an “intelligent” manner by employing on-line
`beam forming techniques. Otherwise such techniques may be
`used off-line. Furthermore, we found that the superior spatial
`sampling afforded by digital streamers allows for very effective
`F-K filtering. Streamer self-noise, on the other hand, can only
`be effectively combatted by proper streamer design. This entails
`a thorough understanding of the mechanisms by which these
`noise components originate and are transmitted throughout the
`streamer.
`
`Streamer self-noise
`Streamer self-noise is initiated by two main exciting forces,
`namely axial vibrations and pressure fluctuations in the turbulent
`boundary layer. The axial vibrations are transmitted within the
`streamer by the strain members and are set up by (a) Input vibra-
`tions consisting of vibrations transmitted from the towing vessel
`along the tow cable, vibrations caused by tow cable strumming,
`and vibrations along the tail rope caused by tail buoy tugging.
`(b) Residual vibrations consisting of vibrations caused by out of
`balance forces on the streamer itself. The noise caused by axial
`vibration is difficult to quantify because it is dependent upon SO
`many variables but in general it is biased to low frequencies
`
`4
`
`-k
`
`0
`
`+k
`
`b
`7
`
`.10 Hz
`
`20
`
`I
`
`f
`
`250 Hz
`
`FIG. 2. Schematic F-K spectrum of streamer noise.
`
`Downloaded 08/26/14 to 173.226.64.254. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`
`
`Poster Papers 2
`
`293
`
`,”
`A
`L
`
`%
`
`-40
`
`-60
`
`-60
`
`-120
`0
`
`100
`
`200
`
`300
`
`400
`
`500
`
`FREQUENCY
`
`(Hz)
`
`z
`A
`t
`
`d
`
`0
`
`-40
`
`-60
`
`-60
`
`-120
`0
`
`100
`
`200
`
`300
`
`400
`
`500
`
`FREQUENCY (Hz)
`
`4.5 Knots
`FIG. 3. Isolator tests. Spectra show input vibration to Isolator (upper curve) and residual vibration (lower curve).
`
`3 Knots
`
`where it dominates the turbulent boundary noise at normal towing
`speeds. The turbulent boundary layer is set up at the streamer
`wall as kinematic forces transcend the viscous forces of the sea.
`This causes noise that is spectrally white and is nonpropagating.
`For this reason it is often referred to as “pseudosound”.
`The exciting forces create noise in the streamer through var-
`ious transfer modes, the main ones being scattering of axial vi-
`brations by connectors and spacers, signal induction by vibration
`of hydrophones, collimated scattering of the turbulent boundary
`layer at connectors and internal components, convective transfer
`of the turbulent boundary layer pressure, and acoustic transmis-
`sions from the turbulent boundary layer. The scattering modes
`create internal pressure fields which set up waves upon which the
`tube acts as a restoring force. These are hose extensional waves
`and breathing waves.
`
`Streamer design
`
`The basic philosophy for the design of an acoustically quiet
`streamer must be to minimize the exciting forces and reduce the
`transfer modes as far as possible. Furthermore, noise fields must
`be attenuated as quickly as possible once they have been initi-
`ated. Input vibrations can be reduced by tuning and isolating
`shipboard machinery and by the use of “soft tows” and fairings.
`However, it is essential also to reduce the input vibration by
`properly designed vibration isolator sections (Bedenbender et al.,
`1970). These can be designed to dissipate vibrational energy over
`a particular frequency range in order to match the input fre-
`quency. Figure 3 demonstrates the effectiveness of well designed
`isolators. The input vibration measured at the head of the isolator
`(upper curve) was reduced at output (lower curve) by some 20
`dB out to 75 Hz at 3 knots towing speed (Figure 3a); this effect
`is enhanced at higher speeds as shown in Figure 3b for a speed
`of 4.5 knots.
`Residual vibration can be reduced via tight manufacturing tol-
`erances on weight and volume in order to lighten ballasting re-
`quirements on the streamer. The outer profile of the streamer
`should be smooth with no radial obtrusions which could cause
`fluctuations in the fluid flow. This is a problem in digital stream-
`ers where the electronic nodes are of larger diameter than the
`streamer sections. This can be remedied to a large extent by re-
`ducing the diameter ratio and achieving a smooth transition be-
`tween the streamer and node diameters. Figure 4 shows spectral
`
`results of an experiment carried out in sea trials where the flow
`noise caused by two different diameter nodes was measured. The
`lower curve is obtained using a node 100 mm in diameter while
`the higher curve is from a node 115 mm in diameter. At frequen-
`cies below 80 Hz, ship-radiated noise dominates both experi-
`ments. However, at higher frequencies the streamer flow noise is
`in evidence. The restriction to 100 mm is obviously desirable
`giving an improvement of some 50 dB.
`Once excitation vibrations have been reduced as far as possi-
`ble, it is necessary to suppress the noise transfer modes. For ex-
`ample, to reduce direct transmissions from the turbulent bound-
`ary layer it is important to distance the hydrophones from the
`skin as far as possible, i.e., maximize the streamer diameter
`within winching constraints. Stiff skins also reduce these trans-
`missions but this has direct implications on the noise fields within
`the streamer. Collimated scattering can be reduced by optimal
`design and matching of materials within the streamer. Scattering
`will occur where there is any physical discontinuity within the
`streamer and is most pronounced near the skin where internal
`pressure is strongest. Therefore, it is important to centralize in-
`ternal components and make them low profile and leave as much
`of the inner volume open to fluid movement as possible. This
`entails the use of specially designed spacers and hydrophone
`mounts.
`Once a noise field has been set up within the streamer it must
`be attenuated or rejected as quickly as possible. Breathing and
`
`Db rel.
`10
`
`-90
`0
`
`50
`
`100
`
`150
`
`200
`
`250
`
`Frequency Hz
`
`FIG. 4. Diameter node tests.
`
`Downloaded 08/26/14 to 173.226.64.254. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`
`
`294
`
`Poster Papers 2
`
`-m.
`FIG. 5. Group responses for bulge wave noise (45 m/s).
`
`2
`
`I,
`
`6
`
`I.4
`
`--.
`
`1.
`
`using the simple velocity filter shown in Figure 6. The applica-
`tion of such F-K filters allows rms noise estimates to be made
`within the useful seismic bandwidth. This has strong implications
`for seismic survey specifications since even very high-amplitude
`noise may not be detrimental to the final processed seismic data.
`Finally, the shot-generated side-scattered noise described by Lar-
`ner et al. (1981) can also be effectively removed with such fil-
`ters.
`
`Conclusions
`In conclusion, with the advent of high fidelity recording sys-
`tems it becomes increasingly important to gain a thorough under-
`standing of noise mechanisms and transmission modes. This un-
`derstanding not only leads to good design principles but also
`enables the efficient use of processing for noise suppression. The
`latter should lead to a relaxation in acquisition constraints thereby
`reducing survey downtime.
`
`References
`Bedenbender, J. W., Johnston, R. C., and Neitzel, E. B., 197q, Electro-
`acoustic characteristics of marine seismic streamer: Geophysq 35.
`Lamer, K., Chambers R., Yang, M., Lynn, W., and Wai, W., 1981,
`Toward an understanding and-suppression of coherent noise in marine
`seismic data: presented at the 51st Annual SEG meetingLos Angeles.
`Schoenberger, M., and Mifsud, J. F., 1974, Hydrophonesreamer Boise:
`Geophysics, 39.
`Weichart, H. F., 1973, Acoustic waves along oilfilled streamer cables:
`Geophys. Prosp., 21.
`Wenz. M.. 1962. Acoustic ambient noise in the ocean: Soectra and
`so&es: Acoust: Sot. Am., 34.
`
`Processing of Wide-Angle
`Vibroseis Data
`F. Wenzel, Geophysical Institute, University of Karlsruhe, West
`Germany
`In 1984 seismic wide-angle VibroseisQ experiments across the
`Rhinegraben, eastern France/southwest Germ
`as a joint French/German venture.
`graben in the Black Forest, signals g
`were recorded with a 200-channel
`sets ranged between 66 and 82 km.
`low signal-to-noise ratio and
`by strong topography. For w
`corrections based on the assu
`no longer appropriate. Ray p
`to be used in order to eli
`
`hit and near-surface in-
`the poor S/N ratio has
`
`hose extensional waves use the tube as a propagating medium,
`so by choosing a tube with a thin wall and low dynamic Young’s
`modulus these waves will be attenuated quickly. However, this
`has direct implications on the amount of noise transfer to the
`system and so a trade-off must be sought.
`Hydrophone group design can be employed to attenuate coher-
`ent noise with specific frequency and velocity characteristics. In
`Figure 5 we see group responses to a bulge wave traveling down
`the streamer at 45 m/s. It can be seen that the 12 m group offers
`greater attenuation than the 6 m group throughout the band 2-6
`Hz which is the frequency band dominated by bulge wave noise.
`In fact, the 12 m group achieves infinite attenuation near the
`center of the band at 3.7 Hz. However, short groups enhance the
`high-frequency response and allow greater flexibility for array
`design. It is also worth noting that low-velocity coherent noise
`may be easily removed in processing without undue effect on the
`seismic data. Therefore, short groups are recommended bearing
`in mind dynamic range considerations and recording constraints.
`
`F-K analysis
`The fine spatial sampling afforded by digital streamers (typi-
`cally 6.25 m) renders it possible to analyze noise trains more
`effectively. This allows greater exactitude in removing noise by
`F-K filtering. Futhermore, since spatial aliasing is dramatically
`reduced there will be less corruption of signal. Referring to the
`example illustrated in Figure 2, refracted ship-radiated noise and
`bulge-wave noise plus all water-borne noise are easily removed
`
`-k
`
`cl
`
`+k
`
`*
`
`_ LOW-CUT
`FILTER
`
`e both problems simultaneously.
`ameters allows the incorporation
`end on the angle of emergence of
`cation of semblance as coherency
`
`IIGH-CUT
`FILTER
`
`t improvement of the S/N ratio.
`
`Data acquisition and characteristics
`In fall 1984 two wide-angle Vibroseis experiments across the
`Rhinegraben were carried out by the French ECORS and the Ger-
`man DEKORP group (Damotte et al., 1985). The measurements
`served as a feasibility study to probe the possibility of recording
`
`FIG. 6. F-K filter to remove ship noise.
`
`@Conoco Inc
`
`Downloaded 08/26/14 to 173.226.64.254. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
`
`