SPECIAL FOCUS: EXPLORATION

Will marine seismic data benefit from towed, dual-sensor streamers?

 Having both acoustic pressure and velocity sensors collocated along a towed streamer should offer long-sought benefits. 

Rune Tenghamn, PGS

For decades, it has been known mathematically that if one could simultaneously measure the acoustic pressure and particle velocity aspects of a compressional wave in water, then certain benefits would accrue. Among these benefits is the ability to tow streamers deeper, in a quieter environment, resulting in an improved “weather window,” as well as the removal of the down-going wavefield component that is reflected from the sea surface, thereby recovering significant high- and low-frequency amplitudes normally missing from conventional seismic data. Until now, the actual construction of such dual, collocated sensors on a streamer, as well as understanding how best to use them, had not been achieved. At the June EAGE annual meeting in London, PGS announced that it had made such an achievement.

As one might expect, there was considerable interest in the new technology, as well as the sort of skepticism and questions that would be expected from the exploration community. This article answers questions that were typical of those asked by exploration professionals at the EAGE meeting. It gives examples from the field that support the expected benefits. Also, it shows mathematically why the method should work, and what is meant by a particle of water.

TYPICAL QUESTIONS

Q. What is the main value of this new technology?

A. This technology substantially eliminates the choice between penetration (low frequencies) and high resolution (high frequencies), since it can allow both. This is particularly important in areas where complex shallow structures obstruct deep targets. Having low frequencies for the penetration and still being able to get good resolution of the shallow parts can be valuable, especially in sub-basalt and sub-salt areas. The new technology, called the Next Generation Streamer, could open new exploration and development opportunities in areas affected by poor resolution, poor penetration and/or noisy data.

The streamer can be towed much deeper than conventional streamers, which increases the weather window while decreasing the noise level, normally by 3-7 dB (from measurements during the past three years). Oil companies could also benefit from being able to use both the up-going and the down-going wavefield for processing. Attenuation of multiples can be more effective when using the new streamer. New algorithms will continuously be developed to take full benefit of this new technology.

Q. How does this streamer differ from conventional streamers?

A. There are a number of key elements that differentiate the new streamer, but most importantly, by collocating pressure and velocity sensors, it is possible to separate the recorded wavefield into up-going and down-going signals. When these two signals are properly summed, the receiver ghost cancels out, the signal-to-noise ratio increases, more high and low frequencies are recorded and we have a much clearer image.

One of the major challenges facing marine seismic acquisition is the ghost notch. This notch in the amplitude spectrum is caused by the reflection off the sea surface. Towing the streamer shallower combats this by moving the notch to higher frequencies, but this exposes high levels of weather-related noise. Towing the streamers deeper to get away from weather-related noise moves the notch lower into the bandwidth we wish to measure. Typically, these effects restrict streamer towing depths to a range of about 6-9 m. Oil companies must choose a depth, based on maximizing data quality at the expected target depth, while sacrificing image quality at shallower or deeper targets.

With the new streamer, all that changes. The new technology combines pressure and velocity sensors in the same streamer. Towing the streamers deep decreases weather noise that is typically seen. Figure 1 shows typical weather noise patterns for a 7-m streamer tow and the significantly reduced noise levels from 15-m and 20-m streamer tows.

Fig. 1. Weather noise is reduced at all frequencies with increased streamer depth.

The new streamer is typically towed at 15 m, and while that would normally affect the high frequencies, recording from both pressure and velocity sensors results in notches that are complementary and, after summing, produce a much broader bandwidth signal, typically 5-150 Hz.

Having dual sensors allows measurement of the up-going wavefield as a pressure field and as a particle velocity field. Figure 2 shows two curves that measure peak amplitudes as the wavefields impinge on the cable. These wavefields continue to propagate and reflect off the sea surface, which acts as an inverter and creates a negative pressure field coming back down. This field is recorded as a trough on the pressure sensor. But the sea surface also inverts the particle velocity field, and because the velocity sensor is directional, it records this inverted wavefield as a peak. So when these two are combined, together with the pressure data, we have a peak followed by a trough. When we combine them with the velocity sensor, we have two peaks.

Fig. 2. In 2A, two up-going wavefields are recorded by the cable, one in the pressure (blue) domain the other in the velocity (red) domain (top, right panel). These reflect off the surface and are detected again, where only the pressure wave is inverted (middle panel). The two signals are summed in each domain (bottom panel). In 2B, the up-going and down-going signals are summed across domains, resulting in an up-going wavefield when added, with destructive interference of the ghosts (top, right) or, when subtracted, a down-going wavefield with constructive interference of the ghosts (bottom, right).

If we add the datasets together, the initial peaks will constructively interfere and the trailing ghosts will destructively interfere. Conversely, if we subtract these two datasets, the initial peaks will cancel and the trailing ghosts will constructively interfere. This results in images of an up-going wavefield through the addition, or a down-going wavefield through the subtraction.

Q. Can the new streamer affect 4D seismic? 

A. Conventional seismic data can be re-created from the new streamer. Because the up-going and down-going wavefields can be separated, it is possible to redatum the new streamer to any arbitrary depth and so match any previous survey. To illustrate this, conventional data, shot with an 8-m streamer was compared with data shot at a 15-m depth with the new streamer.

To make the recombination, we performed wavefield separation, took the up-going wavefield and propagated it forward to 8 m, and then took the down-going wavefield and propagated it backwards to 8 m. Then, through the combination, we reconstructed the 8-m streamer data. Figure 3 shows the two sections, which are obviously highly comparable, but the amplitude spectra show further confirmation that we have accurately reproduced the wavefield characteristics.

Fig. 3. Two highly comparable sections, with amplitude spectra, confirm accurate reproduction of wavefield characteristics for legacy 4D work.

Q. How much data have you collected with the new streamers?

A. It is very important to get direct “apple-to-apple” comparisons, to show the advantages compared with a conventional streamer. We mobilized the new streamer on the Ramform Explorer during a port call in Bergen in July and acquired a survey using both conventional and the new streamers simultaneously, to provide a true comparison dataset. We towed the new streamer at 15 m and the conventional at the usual 8 m and acquired about 160 km of data in different sea states. Initial results from early processing indicate that the final results will give significant benefit.

For example, Fig. 4 shows the wavefield separation comparison between pressure wavefield and the particle-velocity wavefield, and their respective amplitude spectra out to 125 Hz. The differences in the character of the notches are easily visible, but also notice the significant differences in the character of the seismic data. The combination of both to derive the up-going wavefield after separation is even more revealing. It provides a much clearer image and demonstrates broader continuous bandwidth and enhanced resolution.

Fig. 4. Comparison between pressure and particle-velocity wavefields, and respective amplitude spectra out to 125 Hz. The differences in the notches and in the character of the seismic data are easily visible. Rightmost panel shows combined (added) wavefields.

Q. What were the main challenges in developing the new streamer, and how did you overcome them?

A. Obviously, placing the velocity sensor in the streamer had many challenges; otherwise it would have been done some time ago. We worked at designing an array that optimizes noise cancellation, and investigated different sensor designs and performance requirements before running field trials. First field trials took place in 2002. These tests showed that we were very close to having a technology that would work in full scale. It took us another five years to get to a production version of the streamer.

Exactly how we designed the streamer is something we don’t want to discuss in detail. It is a combination of a specially developed sensor and how we combined these sensors together in the streamer.

Q. Developing a new streamer normally is very risky and takes years before all problems have been sorted out. When will the new streamer be commercially deployed?

A. The new streamer is built on the same platform as our well-proven RDH-S solid streamers, which were introduced in 2004, and have all the fluid replaced by Buoyancy Void Filler. This filler is an environmentally friendly solid gel and produces a much quieter streamer. For the new streamer, everything is identical, except for the added velocity sensors and new data acquisition electronics to reduce the power consumption in the streamer. There are twice as many channels, and the risk would be that this could limit the maximum streamer length.

The new electronics are based on Ethernet telemetry and will support a 12-km streamer length. During the latest survey, we used a standard and the new streamer, both 8 km long. Even with twice as many channels on the new streamer, the power consumption was only half that of the conventional streamer.

With the streamer based on our existing platform, we don’t see any issues that would prevent introduction of the new streamer into our fleet. The 8-km streamer we are using now is what we see as the production version and is ready for commercial deployment. It has been tested extensively since early this year. The existing streamer is the third version of the new streamer; the first one was built in 2004.

Q. Is there an improvement in resolution using the new streamer?

A. Current experience says that given at least two octaves of frequency bandwidth, resolution is proportional to the maximum frequency that can be usefully acquired. That means that if good signal-to-noise content on the high frequencies can be obtained after deghosting, which we expect with the new streamer, we should get great vertical resolution.

Q. What extra processing is required for the new streamer? 

A. We have invested significant technical resources into this issue, and plan to fully exploit the geophysical advantages of the new streamer. The power of the method lies in the decomposition of the wavefield into separate up-going and down-going components. The up-going wavefield component is free of the receiver ghost, and thus contains significantly greater low- and high-frequency content than conventional hydrophone-only streamer data. It is important to note that the sensitivity differences between hydrophone and velocity sensors are fully compensated for in pre-processing.

Following appropriate pre-processing to preserve sensor fidelity and attenuate noise, wavefield decomposition is used to produce the ghost-free data. Sophisticated implementations of multiple removal are then possible. Otherwise, the up-going wavefield can be considered as “normal” data in that all successive processing technologies are applicable without modification. The caveat, of course, is that the data frequency content and signal-to-noise content is significantly better than other data, and the processing flow must be customized accordingly to preserve and exploit these benefits.

Q. Can you expand on the benefits for multiple attenuation? You mentioned SRME.

A. There are advantages to be gained from multiple attenuation, and a Surface-Related Multiple Elimination (SRME) method has been developed that takes advantage of the dual-sensor data acquired. Multiple prediction is based on the up-going pressure field and down-going velocity field. The fact that this SRME approach needs two data sets from two different recordings is the main difference, from an operational point of view, from procedures based on the feedback theory or on the inverse scattering theory. The latter approaches predict multiples solely from the pressure field.

The key advantage of using the down-going velocity field is that any variations in sea surface level and reflection coefficient are implicitly included. In addition, the use of a velocity field introduces a necessary, angle-dependent scaling into the prediction, which cannot be easily compensated for in other approaches using adaptive subtraction. Field data examples show superior multiple attenuation compared with conventional SRME, Fig. 5.

Fig. 5. Field data examples of (first seafloor) multiple attenuation compared with conventional SRME.

Q. Concerning the particle velocity of the water that the new velocity sensors will pick up: Is this a result of any pressure-to-velocity conversion in the cable itself?

A. No. There is no pressure-to-velocity conversion in the cable itself. For a complete discussion of the mathematics of water particle velocity, see the section on measuring particle velocity. 

THE AUTHOR
	Rune Tenghamn is VP for Innovation and Business Development at Petroleum Geo-Services (PGS). He holds an MSc degree in engineering physics from Chalmers University of Technology. He joined PGS in 1994 and has had various management positions. Most recently, he was VP Marine Technology until May 2007, with responsibility for the new streamer development and other marine technologies being developed in PGS.