Marine seismic data always contain components of noise originating from various sources.
Weather is often the main source of much of the recorded noise and can according toSmith (1999), induce delays that account for up to40%of the total cost of a marine survey. Important work to determine noise characteristics and identifying sources of noise on seismic streamers was done bySchoenberger and Mifsud(1974). Since then, the seismic industry has focused on improving streamer system technology to reduce the effects of many of the identified sources of noise. For instance, innovative engineering solutions and the introduction of new materials have greatly reduced noise from tugging and virtually eliminated electrical interference on modern equipment.Peacock et al.(1983),Bjelland(1993) andDowling(1998) made significant contri-butions to the understanding of noise generation mechanisms on fluid filled seismic streamers.
The primary mechanism under bad weather conditions, was identified to be longitudinal wave propagation inside the streamer that create low-frequency noise. These waves were caused by tugging and strumming from the vessel, paravanes, and tail-buoy together with lead ins, tow and cross cables. Brink and Spackman (2004) andDowle(2006) showed that modern foam filled streamers are less sensitive to such internal bulge-waves as well as vibrations from exter-nal forces like tugging/strumming. A possible noise generation mechanism for flexible slender cylinders was identified byPaidoussis(1966) in the form of buckling and oscillatory instabili-ties. For seismic streamers,Parrish(2005) on the other hand, shows that such oscillations most
Trace number
Figure 3.1: A shot gather where weather related noise show up as vertical stripes.
likely will only be found at frequencies well below one Hz.
In recent work presented byLandrø(2008) the signal remaining from the previous seismic shot is considered as a source of noise. For a seismic line acquired in the North Sea he shows that after 8 s, using a 5.3 Hz low-cut filter, the rms noise from the previous shot is2.5μBar. After 15 s the noise is around1μBar. This is a significant percentage of the overall noise in good weather. However, in a non optimal weather situation it makes up less than5percent of the rms noise level.
External sources of noise may also adversely affect the quality of the seismic data. These in-clude for instance seismic interference, engine and propeller noise and wind and bubbles near the surface. While the latter probably is secondary, the two other can arguably be avoided by careful operational planning.
This work combines applied geophysics and fluid dynamics in an attempt to provide a descrip-tion of some of the complex processes associated with towed streamer arrays used for marine seismic exploration. More precisely it deals with two distinctly different sources of noise. First low frequency hydrostatic pressure fluctuations originating from the wave-induced vertical mo-tion of the ocean. Secondly, dynamical pressure fluctuamo-tions on the surface of the streamer generated within the surrounding turbulent boundary layer. It will be argued that on modern streamers operated during non-ideal weather conditions, these noise sources are significant, and will often dominate most other common types of noise. A typical example of noisy streamer data from a solid streamer recorded during 7s is shown in Figure3.1. This weather related noise has large amplitudes at low frequencies, and it is spatially coherent over a number of hydrophones.
3.2.1 Literature and fluid mechanical background
The flow past a circular cylinder is a classical problem in fluid mechanics. Unfortunately, most of the work performed has been on steady flow normal to the cylinder axis (α = 90◦). For seismic streamers we typically haveα = ±5◦, and the available literature is more limited.
Early work focusing mainly on the average flow on axially symmetric cylinders (α= 0◦) were carried out byWillmarth and Yang(1970),Denli and Landweber (1979) andWillmarth et al.
(1975). They investigated wall pressure fluctuations in conjunction with the mean velocity pro-files. Willmarth and Sharma(1984) andSnarski and Lueptow(1995), performed similar stud-ies where also the turbulent flow propertstud-ies were investigated. In the latter study two different groups of fluctuations in the turbulent boundary layer surrounding a cylinder were identified.
First, low frequency high energy fluctuations originating from the outer parts of the boundary layer, possibly in the form of large scale coherent structures. Secondly, small scale, high fre-quency disturbances related to the so-called burst-sweep cycle. These disturbances probably make a significant contributions to the fluctuating rms-pressure near a cylinder surface.
BothLueptow et al.(1985), and Heenan and Morrison(2002a), as well as a number of other researchers have shown that even small misalignments of the axis of the cylinder relative to the mean flow will cause asymmetry of the boundary layer and induce significant deviations in the fluctuating wall pressure levels around the circumference of the cylinder. Based on mea-surements,Lueptow et al.(1985), andFurey(2005) provide statistics on the distribution of the Reynolds stresses within the turbulent boundary layer atα= 0◦. The Reynolds stresses,uiuj
are the ensemble average product of velocity fluctuations at the same spatial location, where over-bar indicated the ensemble average operator. Their physical significance is thatuiuj repre-sents the average effect of turbulent advection on the mean flow field. Turbulent kinetic energy, for instance, is defined ask= (u2+v2+w2)/2.
InCipolla and Keith(2003) andKeith et al. (2005) details on how boundary layer thickness scale with the cylinder length are presented. Reviews of much of the early work done on wall pressure fluctuations can be found inBull(1996) and inSnarski(1993). Early numer-ical simulations of axial flow were conducted byNeves and Moin(1994b,a) where they also present detailed turbulence statistics for axially aligned flow. Based on wind-tunnel experi-ments,Bull and Dekkers(1993) found that vortex shedding can occur for a limited range of Reynolds number for long cylinders (L/(2a))≈ 3000)at inclination angles as low asα = 1◦. They also suggest that the vortex shedding may be relevant to turbulence-generation (and thus also to noise-generation) in thick axisymmetric turbulent boundary layers. Possible vortex shed-ding at small inclination angles was also reported byAtta(1968). These findings have recently been partly confirmed by Direct Numerical Simulation (DNS) of near axial flow performed by Woods (2006). On the other hand, in experiments performed by Heenan and Morrison (2002a,b), no vortex shedding for inclination between0−6◦was observed. Their hypothesis is that low frequency noise is caused by streamer oscillation or buckling.Furey(2005) also cites a number of experiments which indicate that vorticial structures are shed from cylinders only for inclination angles larger than5◦. It should be noted, however, that cross flow induced vor-tices at small angles can remain attached to the surface and remain within the boundary layer as illustrated in Figure3.2. This could explain the lack of observed shedding in the studies men-tioned above. This phenomenon is sometimes referred to as ’trailing vortices’, seeRamberg (1983) andThomson and Morrison(1971). In a study performed bySnarski(2004), the spec-tral characteristics of flow over a cylinderL/(2a) = 23at different inclination angles were investigated. It was found that forα ≤ 15◦ the energy spectra becomes broad-banded, and the turbulent energy content decreases. He also reports significant Reynolds number effects for small inclinations angles, and suggests that the energy spectra for high Reynolds number
Figure 3.2: Sketch, in side view, of the flow past a yawed circular cylinder in a steady trailing vortex regime. (Positive z point downwards.) [AfterThomson and Morrison(1971)].
flow might be a built up of contributions from both low frequency large scale shed vorticity and higher frequency small scale boundary layer turbulence.