Page 137 - Acquisition and Processing of Marine Seismic Data

128 2. MARINE SEISMIC DATA ACQUISITION

instance, up to 250 and 500 Hz frequencies can filters, are applied to the data before digitizing.
be recorded with 2- and 1-ms sampling rates, Their low cut frequency is approximately
respectively. Normally, ghost notches deter- 3–4 Hz, while high-frequency cut-off is generally
mine the upper frequency limit of the seismic 80% of the Nyquist frequency. In conventional
signal spectrum, and it is 125 Hz for a 6-m depth seismic surveys, 4–200 Hz (18–300 dB/Oct) cut-
streamer (Fig. 2.36). If we use a sample rate of off values are typical, whereas a wider pass-band
1 ms with a 500-Hz Nyquist frequency, then is designed for high resolution acquisition, such
we have an oversampled dataset between 125 as 5–412 Hz (12–300 dB/Oct).
and 500 Hz, which considerably increases the
processing time since sampling rate directly 2.5.2.4 Dynamic Range and Seismic Sample
affects the total volume of the recorded data. Format
In conventional 2D or 3D acquisition, the Dynamic range is defined as the ratio
sampling rate is selected at either 2 or 4 ms between the highest and the lowest amplitude
depending on the resolution required. For most that can be recorded by an instrument without
seismic surveys, 2 ms is preferred. In high- any distortion and is usually expressed as deci-
resolution surveys, however, 1-ms sampling is bel units (dB). For instance, if a recording system
typical since higher frequencies can be recorded can receive and convert analog data between 1
by towing the source and streamer at shallower and 100 amplitude units, then the dynamic
depths. range of this system is 20 log(100/1) ¼
40 dB. For a seismic recording system, dynamic
2.5.2.3 Recording Filter Cut-Off range is important since it defines the range of
Frequencies the input signal amplitude (maximum and min-
The maximum signal frequency that can be imum signal amplitude values) that can be reli-
reconstructed after digitizing the reflected seis- ably converted to the digital form during
mic signal is determined by the Nyquist fre- sampling. If the dynamic range of the recording
quency (Section 4.10). In other words, we system does not satisfy the amplitude range of
cannot record the frequency components higher the seismic signals, then the high signal ampli-
than Nyquist. However, this does not imply that tudes are clipped, as in the case in Fig. 2.80.
there are no higher frequency amplitude compo- A seismic system with a high dynamic range
nents in the water. Indeed, there is always a can record both extremely small and excessively
static noise component, which spreads all over high signal amplitudes at the same time without
the available spectral bandwidth, including the distortion. Direct arrivals and sea floor reflec-
frequencies beyond the Nyquist frequency. tions in relatively shallow water surveys, espe-
Our sampling rate is always high for these cially in the case of hard sea bottom, can
higher frequency random noise components, generate high amplitude arrivals. The ampli-
since these noise amplitudes are of frequencies tude difference between the largest signal
higher than our Nyquist value. As a result, our amplitude and the ambient noise level may be
predetermined sample rate downsamples these more than 100 dB, which indicates an amplitude
noise components, which ultimately results in ratio of 100,000/1 (Shirley et al., 1985). In 1991,
them being aliased onto the amplitudes of lower delta-sigma type analogue-to-digital (A/D) con-
frequency components in the spectrum. verters introduced 24-bit recording systems into
In order to prevent aliasing of the high- the seismic industry, which provides a 144.48 dB
frequency noise amplitudes beyond the Nyquist dynamic range.
frequency during the recording, wideband fil- Received and digitized seismic amplitude
ters, termed antialiasing filters or recording samples are recorded onto the tape drives or