Marine magnetotellurics for sub salt exploration 2. A marine equipment system

S. Constable
Scripps Institution of Oceanography

A. Orange
Arnold Orange Associates

G.M. Hoversten and H.F. Morrison
University of California at Berkeley


(This is an abridged version of a paper in revision for publication in Geophysics based on an abstract presented at the November 1996 SEG meeting in Houston.)

Summary

Induction in electrically conductive seawater attenuates the magnetotelluric (MT) fields and, coupled with the red nature of the natural magnetic fieldspectrum, leads to a dramatic loss of electric and magnetic field power on the seafloor at periods shorter than 1000 s. For this reason the marine MT method has traditionally been used only at periods of 10^3 to 10^5 s to probe deep mantle structure; rarely does a seafloor MT response extend to 100 s period. To be useful for mapping continental shelf structure at depths relevant to petroleum exploration, however, MT measurements need to be made at periods between 1 and 1000 s. This may be accomplished using AC-coupled sensors, induction coils for the magnetic field and an electric field amplifier developed for marine controlled source applications. The electrically quiet seafloor allows the attenuated electric field to be greatly amplified before recording, and in deep (1 km) water, motional noise in magnetic field sensors appears not to be a problem. In shallower water, motional noise does degrade the magnetic measurement, but seafloor magnetic records can be replaced by land recordings, producing an effectively sea-surface MT response. Field trials of such equipment in 1 km deep water produced good quality MT responsesat periods of 3 to 1000 s; in shallower water responses to a few hertzcan be obtained. Using an autonomous seafloor data logger developedat Scripps Institution of Oceanography, marine surveys of 50 to 100 sites are feasible.

Introduction

The magnetotelluric (MT) method has been used map sedimentary structure as an aid to petroleum exploration for several decades (e.g. Vozoff, 1972; Orange, 1989). The reliability and usefulness of the MT method has improved greatly over the past few years as a result of progress in several areas. These include improvements in data acquisition technology, the introduction of a remote reference to reduce bias associated with noise in the magnetic field measurements,`robust' response function estimation methods, and improved 1D, 2D, and even 3D forward and inverse modeling codes. As a consequence, the MT method represents an important non-seismic exploration tool, particularly for reconnaisance surveys and in areas where the seismic reflection method performs poorly. The latter include buried salt, carbonate, and volcanic horizons that efficiently reflect and scatter acoustic energy. In the companion paper (Marine Magnetotellurics for Petroleum Exploration 2, Numerical Analysis of Subsalt Resolution ), 2D and 3D modeling is used to demonstrate the utility of the MT method in delineating subsalt structure.

Many petroleum prospects are offshore, but because seawater attenuates the magnetic source-field at frequencies below 0.01 Hz, the marine MT method has traditionally been considered sensitive only to great depths. Indeed, the focus of marine MT instrumentation has been drift-free measurement of EM fields at essentially DC frequencies (e.g. Filloux, 1987). Deployed mainly on the deep seafloor in persuit of mantle conductivity structure, these instruments rarely provide responses at periods shorter than a few hundred seconds. Investigations of shallow conductivity have been undertaken usinga controlled source method, replacing the lost power at high frequencies using a man-made transmitter positioned on or close to the seafloor (e.g. Cox etal , 1986). However, the use of a marine controlled source method is technologically challenging, and the method favours the more resistive, hard-rock seafloor of the deep ocean over the conductive sediments of petroleum targets on the continental shelf. Modeling a 3D source field also presents a greater difficulty than modeling the MT plane-wave source.

The companion paper demonstrates that if instrumentation and processingare optimized for the frequency band 0.001 to 1 Hz, then the marine MT method is indeed viable for delineating subsalt structure. The work described here resulted from field trials conducted for the express purpose of providing the petroleum industry with a usable offshore electromagnetic explorationmethod.

Seafloor magnetic fields

Simple 1D theory demonstrates that for downward propagating energy, electric and magnetic fields measured on the seafloor provide an MT impedance for the subsea section that is independent of the overlying conductive layer. It appears, then, that one may neglect the effect of the seawater layer and approach seafloor MT just as one would for land MT. However, although the field ratio is unchanged, the fields themselves will be attenuated by induction in the seawater layer, so the instrumental and logistical impact of smaller fields must be considered. Also, wave and current motion in shallow water makes measurement of magnetic fields difficult, and it can be desirable to use nearby land recordings of the magnetic field for the MT impedance on the grounds that B is spatially coherent over large horizontal distances (at least several hundred kilometers). The effect of taking seafloor E with the equivalentof sea-surface B needs also to be considered. Attenuation of the electric and magnetic fields through seawater depends on the electrical structure of the seafloor, and so cannot be estimated in detail prior to carrying out an electromagnetic survey. However, general predictions can be made. The effect of the resistive seafloor is to slightly enhance the magnitude of the electric fields over their half-space values. The effect of the seafloor on the magnetic field, on the other hand, is much larger, and increasing the resistivity of the seafloor greatly attenuates the fields over their half-space values.

With the appropriate equipment the seafloor electric field is relatively easy to measure, because although it is much smaller than normally encountered on land, there is no cultural contamination and the isothermal, isosaline environmentis ideal for non-polarizing electrodes. However, as intimated above, motional noise in the magnetic sensors can degrade the quality of the B field measurement, particularly in shallow water. The use of a land magnetic sensor then becomes attractive, by taking the ratio of seafloor E to land B to get a magnetotelluric impedance. Because the sea's effect on the E field is small, the correction can be ignored or predicted quite accurately using an estimated seafloor resistivity. One then has the equivalent of a sea-surface MT response, and the first layer of the model will be the seawater itself. Since sea depth and resistivity are well known, this layer can be fixed during modeling or inversion to recover accurate estimates of seafloor structure.

Instrumentation

Marine MT studies have traditionally used fluxgate or torsion fiber magnetometer sensors and DC coupled electric field sensors. On land, however, MT exploration for shallower targets is conducted with induction coils which respond to the time derivative ofthe magnetic field. This AC coupling removes the 50000 nT main componentof the field, but because the natural field spectrum is very red at frequencies above 1 Hz coils can be used effectively to periods in excess of 1000 s. We use the BF-4 magnetometer coil manufactured by Electromagnetic Instruments Incorporated (EMI). It follows that AC coupling is also appropriate for measuring electric fields, and the amplifier developed for seafloor controlled source applications is well-suited to seafloor MT (Webb et al ., 1985). This amplifier is designed to take advantage of the low noise and low impedance of the marine silver-silver chloride electrodes.

The MT method requires time series of electric and magnetic fields that are many times longer that the longest period of interest; a typical recording time of 10 hours would be used to obtain reliable MT responses at 1000 s period or longer. It is impractical to moor a ship for this length of time, and somewhat difficult to do so in deep water, and so we choose autonomous data recorders rather than wireline instrumentation. Seafloor data loggers have been developed for a variety of applications at Scripps Institution ofOceanography (SIO). Remote reference processing requires that data at the base and reference sites be acquired simultaneously with synchronized acquisition systems. More importantly, processing seafloor data with land magnetometer components requires precise timing; phases accurate to 5 degrees at a frequency of 2 Hz requires timing to 7 ms or better. Since all the seafloor instruments are autonomous, and beyond the reach of radio communication, accurate timing must be accomplished by on-board quartz clocks. The on-board clocks are started using a GPS time standard, with initial timing accurate to 1-10 microseconds. After recovery, clocks are again checked against the GPS standard to estimate drift or error. Drift rates are typically less than 1 ms per day.

Offshore testing

In order to evaluate the operation of the MT instrument system, we have conducted testing off San Diego in various water depths. The first test was carried out from the 7th to the 16th of April 1994 by fitting three orthogonal BF-4 to coils one of the SIO controlled source electric field receivers. This instrument was deployed in 1000 mwater in the San Diego Trough, approximately 30 km offshore and due west of SIO. A second seafloor data logger was `deployed' at Pinyon Flat Geophysical Observatory about 150 km NE of the offshore site. This instrument was connected to 2 horizontal BF-4 sensors and to 75 m electric dipoles terminated by conventional lead-lead chloride electrodes. An EMI electric field signal conditioner was used with 40 dB of gain and 10 Hz low-pass filtering. Crystal oscillators in both instruments were started from a precision clock at SIO prior to deployment. Sampling rate was continuous at 8 Hz during the 10-day deployment.

The data were re-formatted from the SIO binary format used in the seafloor instruments to EMI time series format, and written to optical disk using the ISO 9660 standard. This allows considerable volumesof data (up to 700 Mbytes at a time) to be read directly into MTR-93, a commercial MT software package for computation of MT response functions. A total of 9 days data were processed in this way. Coherency is high at periods longer than 10 s, showing not only that the magnetic fields are spatially homogeneous over these distances, but also that the orientation of the seafloor instrument has been correctly estimated and that the seafloor magnetic field is being measured with fidelity. A dramatic drop in coherency at periods shorter than 7 s comes from a combination of the red source-field spectrum and seawater attenuation.

Also shown are response computed using seafloor E and remote B (rather than seafloor B), to illustrate both the improved accuracy at higher frequencies (i.e. smaller error bars) and the attenuation associated with seawater. High frequency attenuation is associated with diminished seafloor electric field; if one corrects for this the high frequency is again asymptotic to seawater conductivity. At periods longer than about 100 s the response is depressed because the larger land magnetic field has been used. The XY component agrees with the seafloor response to within a factor of 2, because the lower apparent resistivity produces only minor changes in the magnetic field, but the YX mode differs by an order of magnitude because the higher seafloor apparent resistivity produces more attenuation of seafloor B.

Shallow water deployments (around 100 m) have also been carried out. In these deployments the quality of the electric field measurements is still good, but motional noise associated with wave activity and water currents degrades the quality ofthe magnetic field significantly. The use of land magnetic recordings becomes essential, but the difference between the (effective) sea-surface response and the seafloor response is less dramatic for the thinner seawater layer. Excellent MT responses to frequencies of 2 Hz have been obtained in this way.

Concluding remarks

Using a novel combination of induction coil sensors and AC-coupled electric field amplifiers, good quality seafloor magnetotelluric responses at periods of 3 s to 1000 s have been obtained in water 1 km deep. In the past, a short period limit of 300 s would be typical for marine MT in water of this depth. In shallower water, responses to several hertz can be obtained. This increased bandwidth allows the MT method to be used on the continental shelf to map sedimentary structure as an aid to petroleum exploration. While MT lacks the resolution of the seismic reflection technique, it can be used in situations where the seismic method performs poorly, such as imaging beneath salt, volcanics or carbonates. Also, because electrical conductivity is a strong function of porosity, it can be used alone or in conjunction with seismic velocities to back out porosity and permeability (e.g., Evans, 1994).

In the companion paper (Hoversten et al., 1996), 10 to 20 stations of synthetic data are inverted to recover accurate 2D structure for a variety of geometries. In practice, several times this number of stations might be needed to characterize a prospect, and a modern MT land survey can consists of up to 50 or 100 stations. The new SIO MT system has been designed with operations of this size in mind. Oncea field project is mobilized, it takes about an hour to prepare an instrument for deployment, and perhaps another hour to recover a deployed instrument (depending on water depth and station spacing). Thus about 5 instruments can be recovered and deployed in a 10-hour working day and left overnight to record. At this rate a 100-station survey can be completed in three weeks. Loss rates will vary between about 1 and 5% per deployment, and so a fleet of 10 instrumentsat the start should ensure completion of the marine operation. While the costs of the ship and instrument replacement do not have to be borne by land surveys, the daily productivity is greater for marine MT, and it is estimatedt hat the per station cost of a marine survey would not greatly exceed the cost of an expensive land operation. Indeed, both small and large proprietary surveys have already been conducted during the development of the equipment and methodology.


Acknowledgements: An engineering team consisting of T. Deaton, C. Hollinshead, D. Jacobs, J. Lemire, A. Nance, D.Willoughby, and P. Zimmer helped develop the seafloor MT instrument. We thank captain L. Zimm and the crew of the R.V. Sproul for assistance with the marine experiment, and Electromagnetic Instruments Incorporated for the loan of equipment.


References:

Cox, C.S., S.C. Constable, A.D. Chave, and S.C. Webb, 1986: Controlled source electromagnetic sounding of the oceanic lithosphere, Nature 320 (52-54)

Evans, R.L., 1994: Constraints on the large-scale porosity and permeability structure of young oceanic crust from velocity and resistivity data, Geophysical Journal International 119 (869--879)

Filloux, J.H., 1987: Instrumentation and experimental methods for oceanic studies, in Geomagnetism, editied byt J.A. Jacobs. Academic Press (143-248)

Orange, A.S.,1989: Magnetotelluric exploration for hydrocarbons, Proceedings of the IEEE, 77 (287--317)

Vozoff, K., 1972: The magnetotelluric method in the exploration of sedimentary basins, Geophysics 37 (98--141)

Webb, S.C., Constable, S.C., Cox, C.S., and Deaton, T.K., 1985: A seafloor electric field instrument, Geomagnetism and Geoelectricity 37 (1115-1129)