Page 189 - Acquisition and Processing of Marine Seismic Data

180 3. NOISE IN MARINE SEISMICS

approximately 1500 m/s. The primaries arrive (Fig. 3.10A). While wave troughs cause low
at the same recording time as multiples; how- ambient pressure, wave peaks result in high
ever, they almost always have higher velocities pressure on the hydrophones, which ultimately
than multiples. Multiple reflection hyperbolas creates successive positive and negative high
cross-cut the primary reflection hyperbolas on amplitude bands on the shot records during
shot and CDP gathers, because the propagation the surface wave passage. The noise amplitude
velocity of the signal directly affects the curva- becomes significant during rough weather con-
ture of the reflection hyperbolas (Section 1.3.4). ditions, especially when the streamer is towed
It is not possible to avoid recording of the at relatively shallow depths for high-resolution
multiples during acquisition, and therefore a surveys.
number of specific processing methods have Swell noise generally obscures underlying
been developed to remove the multiple ampli- primary reflection amplitudes on the raw shots
tudes, such as surface-related multiple elimina- because of its relatively higher amplitudes with
tion (SRME), wave equation multiple rejection respect to the reflections, and degrades overall
(WEMR), predictive deconvolution, etc. The data quality. Although this type of noise is
theoretical basis and application restrictions of extremely dominant due to its higher amplitude,
these methods along with their specific cha- we can easily discriminate swell noise and
racteristics are explained in detail in Chapter 7. reflection amplitudes in terms of their different
frequency contents. Amplitude and the fre-
quency band of the swell noise increases as the
weather conditions become rough (Berdenben-
3.4 SWELL NOISE der et al., 1970), and most of the noise ampli-
tudes are below 10 Hz, which allows us to
Swell noise is the most dominant noise type in remove almost all swell noise using a suitable
raw marine seismic data. The main characteris- band-pass filter. Elboth et al. (2009) suggested
tics of the swell noise are its large amplitude and that noise amplitudes in the 0–2 Hz range are
low-frequency content, and it sometimes from wind-driven surface waves, whereas those
induces delays or temporary suspension of data above 2 Hz originate from hydrostatic pressure
acquisition in marine surveys (Dondurur and fluctuations.
Karslı, 2012). It occurs either from (i) wind- Fig. 3.11A illustrates a raw shot gather and its
driven longitudinal sea surface waves (which amplitude and f-k spectra, while Fig. 3.11B
cause hydrostatic pressure fluctuations due to shows its 14 Hz low-pass filtered version which
the vertical motion of the ocean), and (ii) has only amplitudes lower than 14 Hz. This fre-
dynamic pressure variations along the streamer quency band almost completely contains only
(turbulence effect of hydrostatic pressure fluctu- swell noise amplitudes, especially for the shal-
ations due to a turbulent layer surrounding the low parts of the data.
streamer) when the streamer bends in the water. The turbulence effect along the streamer
Both agents produce significantly large ampli- becomes significant if shooting is initiated
tude variations on the seismic data (Fig. 3.10). before turning of the streamer is not completed
The first type is seen as positive and negative during the line change. This type of noise is dis-
low frequency noise bands with high ampli- tinguished on shots as vertical high-amplitude
tudes. This is because the hydrophones are sen- stripes. Fig. 3.12 shows a number of successive
sitive to pressure changes, including the shots with prominent swell noise from the tur-
ambient pressure, which is continuously modi- bulence effect formed by streamer turning at
fied by surface wind-driven wave activity the beginning of a survey line, which produced