Would you like some towed-streamer EM with your seismic?

The development of the towed-streamer electromagnetic (EM) acquisition system is now seeing its 10-year anniversary. The technology has finally reached maturity, and the service is commercially available. The impetus behind the design is obvious. It facilitates efficiency similar to seismic acquisition where only one vessel is required, quality control is performed real-time on the source signal as well as the recorded data, and onboard processing provides quick-look results similar to a brute stack in the world of seismic. The EM acquisition software is conveniently based on the same modules as the seismic data. Simultaneous acquisition of seismic also is facilitated, all at a sailing speed of 4 knots to 5 knots.

The engineering of this system was considered impossible to achieve due to the EM noise generated when the receiver dipoles are towed in the water under the influence of the Earth’s magnetic field, but the acquisition system now produces data of excellent quality. The residual uncertainty in the processed data is normally within 2% to 3% for most of the offsets and typically within 5% even for the longest offsets and lowest frequencies.

The layout of the towed-streamer EM acquisition system with simultaneous acquisition of 2-D seismic is shown in Figure 1. The 800-m (2,625-ft) long bipole source is towed at 10 m (33 ft), and the source current is 1,500 Amps. The streamer is 8,000 m (26,250 ft) long with configurable receiver dipoles, and it is towed at a nominal depth of 100 m (328 ft). The bipole source and the seismic streamer also are laterally separated by 100 m to prevent crosstalk from the powerful source to the low-level signals in the seismic streamer.

FIGURE 1. This image shows the layout of the EM and seismic acquisition systems. The EM streamer has 72 configurable receiver electrode pairs that are short (about 200 m) at the near offset and long (about 1,100 m) at the far offset. (Images courtesy of PGS)

The typical streamer configuration consists of 72 receiving electrode pairs distributed along the streamer to facilitate densely spaced offsets from 0 m to 7,700 m (25,262 ft). At the near offsets the electrode pair is 200 m (656 ft) long. The lengths are successively increasing to 1,100 m (3,610 ft) for the most distant offset. This increases the signal-to-noise (S/N) ratio since the signal decays with increasing offset. Additional stochastic noise reduction is achieved by stacking and low-rank approximation by means of singular value decomposition. Correlated noise originating in sudden tugs of the streamer also can be reduced by an algorithm based on Wiener filtering. The maximum nominal water depth is 400 m (1,312 ft) since the water column absorbs part of the transmitted energy. Larger water depths are acceptable if the lateral extent of the resistive structure of interest is large and/or has a very high transverse resistance.

The source signal is referred to as an optimized repeated sequence (ORS) that generates a denser set of discrete frequencies than the conventional monochromatic square wave. Typically the ORS has twice the density of discrete frequency peaks compared to the conventional square wave. The signal is typically transmitted for 100 seconds followed by a 20-second window of no transmitted signal that is used for background noise estimation and noise reduction processing. The dense spatial sampling combined with the dense set of discrete frequencies facilitates improved resolution and a more data-driven inversion with less dependence on a priori information. The dual-sensor seismic streamer is towed at 20 m (66 ft), and the seismic source is a conventional airgun array. The EM acquisition is not as weather-sensitive as seismic is, and the deep tow of the dual-sensor seismic streamer lowers the wave-induced noise and extends the weather window.

FIGURE 2. The acquisition grid covers the Fastnet Basin located in the Irish sector of the Celtic Sea. Most of the program involved simultaneous acquisition of EM and seismic data.

Multiclient project

An extensive multiclient acquisition program was undertaken in September 2013 in the Fastnet basin in the Irish sector of the Celtic Sea (Figure 2).

The program involved simultaneous acquisition of towed-streamer EM and 2-D seismic as well as 2-D seismic only in areas where the water depth exceeds the nominal maximum of 400 m. In total, 2,800 line km (1,740 line miles) of EM and seismic data were simultaneously acquired in 35 days for an average of 80 line km (50 line miles) per day.

Anisotropic inversion case study

Towed-streamer EM data facilitates anisotropic inversion in spite of the fact that only the inline electric field component is measured. The main reason this is possible originates in the shallow tow of the EM source and receiver, which increases the amount of energy that travels through the atmosphere. This so-called airwave propagates horizontally through the atmosphere, and it will couple with the horizontal conductivity in the anisotropic subsurface.

PGS considers anisotropic inversion algorithms to be mandatory since the overburden tends to be at least mildly anisotropic and the hydrocarbon-charged reservoirs tend to be strongly anisotropic, especially when the high-resistivity sands are interbedded with layers of low-resistivity shale. Assuming the subsurface is isotropic is likely to introduce artificial banding in the inversion that creates an effective anisotropy due to the poor vertical resolution. The basement always seems to be isotropic, and this fortuitous fact offers an interesting advantage because this allows surveyors to identify a charged reservoir located immediately on top of the basement simply by displaying the resistivity ratio.

Currently PGS uses two different anisotropic inversion algorithms. The first is an open source program that can be described as a regularized non-linear 2.5-D anisotropic inversion built around a parallel adaptive finite element approach that only requires a sparse horizon model from depth-converted seismic as constraint. The second algorithm is based on the 3-D contraction integral equation method using a reweighted regularized conjugate gradient technique to minimize the objective functional. The low computational cost for the 3-D inversion is unprecedented within the world of EM and requires very little a priori input.

FIGURE 3. A 2.5-D inversion is shown in color as an overlay on a seismic 2-D section. The two fields are seen in yellow and red at a depth of 1.2 km, with Bressay on the left and Bentley on the right.

Bentley and Bressay are heavy oil fields in the North Sea that were originally discovered in 1977 and 1976, respectively. So far they have not seen any production due to the low value of the oil combined with a low recovery factor. A further complication is that these are injectite sandstone reservoirs with sills and dykes extending upwards from the top of the reservoir. However, production is about to commence with a renewed interest in characterization of these reservoirs. A number of towed-streamer EM survey lines were acquired in the area, and an example of the 2.5-D anisotropic inversion is shown in Figure 3.

The EM acquisition sail line runs from north-northwest on the left to south-southeast on the right. The resistivity is shown as a color overlay on a depth-converted seismic section. Only a sparse seismic horizon model was used to constrain the inversion. The top of basement is seen at about 1.4 km (0.9 miles) to the left and dipping gently to the right. The two oil fields are seen in yellow and red, with Bressay at about 1.2 km (0.7 miles) depth and at 7.5 km (11 miles) distance along the survey line. Bentley is located at approximately 1.2 km depth and at 17 km to 21 km (13 miles) along the survey line. The depth, lateral extent, and resistivity agree very well with what is known about the two reservoirs.

Towed-streamer EM now provides a fully mature service from acquisition to 2.5-D and 3-D inverted resistivity volumes. With simultaneous acquisition of EM and 2-D seismic at an acquisition speed of 4 knots to 5 knots, the efficiency is unprecedented. The dense subsurface sampling combined with the dense set of discrete signal frequencies results in high resolution and data characterized by excellent S/N ratio that requires very little a priori information for the inversion to resistivity.