A3. ELECTROMAGNETIC INDUCTION STUDIES IN THE EARTH AND OCEANS
Appendix A3 was largely developed by the following workshop group: John Booker (Group Leader), W. Campbell, A. Chave, C. Cox, A. Duba, G. Egbert, I. Gough, A. W. Green, L. Hirsch, J. G. Kappenman, L. Law, B. Narod, P. Tarits, J. Tyburczy, P. Wannamaker.
Fluids, magmas, and high geothermal gradients result from dynamic processes such as metamorphism, magmatism, convection, and deformation in the Earth. Electrical conductivity (or its reciprocal, resistivity) is the physical property most sensitive to the configuration and chemistry of these fluids, particularly at the low concentrations expected geologically. Temperature and composition strongly influence the mechanical behavior of Earth materials where ionic fluids do not dominate. In such situations electrical conductivity again depends strongly on dynamically important conditions. The possible widespread presence of grain- boundary phases, such as carbon at crustal or mantle depths, has profound implications for the chemical environment, and once again, remote sensing of the electrical conductivity structure is a very relevant tool for understanding the evolution of our planet.
In addition, there are smaller-scale problems of great interest to society—including energy and resource exploration, electric power grid reliability, water quality and waste management—in which electrical conductivity of the target or its surroundings is a diagnostic physical variable. Electromagnetic methods are the only viable way to delineate conductivity structure from the Earth's surface. Thus, basic research in electromagnetic imaging of subsurface structure has potential for many practical benefits.
Electromagnetic studies are traditionally separated into oceanic and land-based studies. Although much of the science and many of the techniques are very similar for these studies, there are substantial technological differences; oceanic measurements are considerably more difficult and more costly to accomplish. In addition, induction caused by water movement in the ocean has no analog on land. The Earth's oceans are important in the long-term storage and transport of environmental heat and thus affect our climate. As a good electrical conductor, ocean water easily generates measurable electric fields as it moves through the geomagnetic field. Furthermore, at periods longer than a day, the electric fields at any point represent a volume average of the motions. Therefore, electromagnetic measurements in the ocean are an extremely useful tool for probing large-scale dynamics and monitoring long-term, climatically important variability of the sea.
Electromagnetic induction studies of the solid Earth involve either simultaneous measurements of time-varying magnetic and electric fields—in orthogonal horizontal directions for the widely popular magnetotelluric (MT) method or array measurements of the time-varying magnetic field—in three components for the geomagnetic deep sounding or magnetic variation (MV) method. Both methods have been employed on land and on the ocean bottom and have increased our understanding of the Earth's crust and mantle. The MT method in particular has seen rapid advances in recent years. By contrast, studies of low-frequency oceanic motions require only measurements of the vector electric field. This can be done with dipoles a few meters long or with grounded cables hundreds to thousands of kilometers long.
Interpretation of electromagnetic data is usually broken into three steps. The first step involves estimating frequency domain transfer functions between measured field components from long time series. It is complicated by the fact that the Earth is almost always multidimensional, external sources may not be ideal, and noise processes are often non-Gaussian. The second step uses these transfer functions to obtain a representation of Earth structure. Besides the nonuniqueness problems shared by all inversions of incomplete and inaccurate data, this inverse problem is both difficult and numerically intensive because it is unstable,
Page 121nonlinear, and often multidimensional. However, progress in all aspects of interpretation has been substantial enough in recent years to constitute a virtual revolution in our ability to image electrical structure.
Laboratory electrical conductivity studies provide the final critical step, linking models of conductivity and the physical and chemical processes occurring within the Earth. Because conductivity is sensitive to environmental parameters, interpretation of conductivity models requires a thorough understanding of the mechanisms that control conductivity under the limited conditions accessible in the laboratory. Oxygen fugacity, pressure, and chemical environment of surrounding minerals can also affect conduction and are particularly important for adequately constraining the thermodynamic environment of the minerals. Critical questions remaining to be addressed relate to longevity and interconnectedness of conducting fluids (pore geometry) and grain-boundary phases—including aqueous fluids, partial melts, and carbon—over geological time.
Laboratory measurements at simulated in situ conditions of the electrical conductivity of rocks primarily composed of olivine and pyroxene—thought to be primary constituents of the upper mantle—are as much as three orders of magnitude lower than conductivity inferred from inversion of electromagnetic induction data for the outer 200 km of the Earth. What is the reason for this discrepancy between observed conductivity and the measured conductivity of the major mineral phases? The mechanisms known to determine electrical conductivity include the following: composition, quantity, and connectivity of the fluid; partial melts; carbon and sulfides; and composition, temperature, and mineralogy of the crystalline matrix. Given this plurality of ways to explain enhanced conductivity, what independent constraints are required to make the explanation unique?
The evidence for a steep rise in conductivity beginning by 400 km is strong. Recent work incorporating very long period electric fields
Page 122appears to improve resolution substantially and shows that the rise may be steplike. Definitive interpretation of this conductivity increase requires additional laboratory studies of the electrical properties of β- and γ-(Mg,Fe)2SiO4 under controlled conditions. The conductivity appears to level out at around 1 siemens per meter below 800 km, but resolution deteriorates rapidly below 1,000 km. How can this depth barrier be broken through? Additional studies of other candidate materials such as high-pressure minerals with varying iron/magnesium contents under controlled thermodynamic conditions are required to interpret conductivities deeper than 800 km.
Can lateral variations in conductivity be delineated at mid-and lower-mantle depths? It is known that there are significant differences in long-period response functions at different magnetic observatories. But how much of this signal is due to deep lateral heterogeneity and how much is due to biases associated with shallow structure such as the ocean and inadequately represented external source field morphology?
Can lateral variations in conductivity in the upper mantle be mapped? Dynamical considerations suggest that temperature and melt gradients should produce order-of-magnitude lateral variations in conductivity, and existing data support their existence but are often susceptible to alternative explanations. Can this situation be improved with more accurate new data collected in a more systematic fashion? Can it also be improved with application of better methods of interpreting data containing potentially distorting multidimensional structural information?
Do fluids persist in the deep crust over long geological times? It is widely accepted that fluid-rich sedimentary rocks and oceanic crust are transported in subduction zones to depths that require dewatering of the rocks and conversions to higher metamorphic grade. But what paths do these fluids take in returning to the surface? Fluids trapped at depth would have enormous importance in the rheology of the lower crust and upper mantle.
What are the mechanical effects of aqueous fluids in the upper crust? High pore pressures have been implicated in large offset horizontal thrusting, in the low strength of such important strike-slip features as the San Andreas fault, and in controlling rupture during earthquakes and
Page 123aftershocks. Data exist suggesting that these fluids produce resolvable conductivity anomalies, but definitive results remain to be obtained.
What role do carbon and graphite play in producing deep-crustal and mantle conductors? Graphite of biogenic origin in metasedimentary rocks may extend to great depths in subduction and continental collision environments.
How is magma segregated from the mantle beneath a spreading midocean ridge? Model studies based on theoretical studies of this segregation process suggest that ocean bottom MT transects across the ridge can discriminate between major alternatives considerably more effectively than feasible seismic studies and will cover a range of spatial scales that seismic methods cannot resolve.
Can the location of magma bodies in volcanic regions be determined? They can involve both imminent volcanic hazard and major energy resources. Seismic methods have inferred many low-velocity bodies in magmatic environments that have been interpreted as possible melt zones, but electromagnetic methods would appear to be much more sensitive to the important variables. The growing ability to collect very high density electromagnetic data and to deal with complicated three-dimensional geometry is beginning to provide answers.
How do electric currents induced in the ocean complete their circuit? Even at very low frequencies for which the ocean is less than one skin-depth thick, strong induced currents are prevented from leaking down into the conductive asthenosphere and deeper mantle by the resistive nature of oceanic lithosphere. Where these currents cross a coast into the more resistive continent, more current will flow at shallow depth than is predicted by the continental conductivity structure. However, the observed coast effect is often less than expected, which may be due to short-circuit conductive paths to the mantle. This can occur at coasts with active subduction or at ancient suture zones. Electromagnetic measurements are very sensitive to the location and structural details of these structures as well as those at ridge crests and active hot spots where additional short-circuiting pathways may exist.
Can motionally induced electric currents be used to sound the oceanic lithosphere? Barotropic tides in the deep ocean drive vertical
Page 124electric currents through the seafloor. If these tides can be accurately specified and their magnetic and electric fields can be measured, one has an electrical sounding technique that is especially appealing because it is preferentially sensitive to the conductivity of the upper lithosphere. The lithosphere is inaccessible to MT because the conductive ocean screens out the necessary shorter-period external source fields.
Electrical conductivity plays an important role in a wide variety of practical contexts, including the following:
Understanding volcanic regimes. Seismic data have frequently been interpreted to imply magma bodies in the crust, but electromagnetic methods are much more sensitive to the important variables. The growing ability to collect very high density electromagnetic data and to deal with complicated three-dimensional geometry has the potential of giving considerably more information than heretofore available.
Assessment of seismic risk. Active faults can sometimes be distinguished from inactive structures by the higher conductivity associated with fluids in the highly fractured material. In addition, there is a significant probability that an electrical boundary coincides with the brittle-ductile transition in the crust. Conceivably temporal variations of conductivity near active faults may prove useful for earthquake prediction.
Geothermal exploration and resource assessment. Electromagnetic methods have proven effective in this context. Fluid-dominated hydrothermal systems always contain high levels of dissociated salts and are thus very conductive targets. However, they tend to be difficult to develop for environmental reasons. The rarer vapor-dominated systems can be distinguished from the fluid-dominated ones by the lower conductivity of their less saline fluids.
Mineral exploration and resource assessment. Again, electromagnetic methods have proven effective in this context. Many economi
Page 125cally important mineral or metal deposits are highly conductive themselves or are associated with highly conducting materials such as carbon.
Hydrocarbon exploration. Many structures that are capable of trapping hydrocarbons are difficult to image seismically, but involve resistive material overlying good electrical conductors. These include crystalline overthrusts, flood basalts, and carbonate reef structures.
Secondary and tertiary oil recovery. Imaging the extent of the zone flooded by steam or chemicals is one of the major practical difficulties associated with these methods for maximizing production. The steam and chemical floods often change the electrical resistivity of the formation so that electromagnetic imaging either from the surface or a borehole is an appropriate tool.
Prediction and amelioration of electric power grid transients. These problems are thought to be due to geomagnetically induced currents in the regional electrical conductivity structure and have been implicated in large-scale blackouts in both the United State and Canada.
Correction of precision aeromagnetic surveys due to anomalous fields produced by induction in electrical structure.
Hazardous waste site characterization. Dumps with acidic, metalliferous, or other conducting toxic materials make good targets for electromagnetic imaging. Targets such as dense nonaqueous liquids are resistive relative to ground water. When such fluids underlie ground water, they may be imaged by appropriately designed systems.
Assessment of saltwater infiltration into an aquifer. This has become an important hydrologic issue in many coastal and desert urban areas. The conductivity contrast between saltwater and freshwater is ideal for electromagnetic imaging.
Measurement of oceanic motions. Oceanic electromagnetism is having an increasing impact in studies of the water-velocity field based on measurements of motionally induced horizontal electric field in the deep ocean. Electromagnetic measurements offer spatial averaging capabilities that are far superior to more direct measurements and may have substantial impact on determining such climatically important quantities as boundary current transport, especially over the long term. Both the
Page 126theory and the technology are well developed and are certain to play a major role in global change research.
MT and MV sensors have generally been deployed for different, but complementary, reasons. An effective deployment might have an MT transect embedded in an MV array to help ensure optimum profile location with respect to a structure of interest and to constrain the three-dimensional context in which the MT interpretation is made. The reasons for this rest in the different nature of the data collected by the two methods.
MV uses an array of simultaneously recording three-component magnetometers and is primarily sensitive to the electric current distribution in the Earth and the source region. Because of the volume-averaging nature of MV data, there will be a sufficiently long period for deployment configurations so that data will be protected against aliasing of small-scale electrical structures within the array. MV data are therefore ideally suited to determining lateral structure. However, the sensitivity of MV data to source current structure requires attention to detecting and eliminating nonplanar source fields or correcting for their effects. Decomposition of MV array cross-spectral density matrices permits isolating uniform sources and synthesizing the response of large arrays from smaller ones. If the source is nonplanar and can be determined from the array (which may require regional or global observatories), response functions closely related to MT may be estimated.
By comparison, MT uses measurements of the orthogonal horizontal electric and magnetic fields and is more closely related to conductivity than current. The electric fields include information about conductivity heterogeneity very local to the measurement site. This effect persists to very long periods and can be aliased if site spacing is too wide. Thus, new deployment strategies such as electric dipoles placed end to end in a continuous profile and new analysis techniques such as spatial filtering of the electric field have been developed to detect and mitigate the distorting
Page 127effect of shallow, three-dimensional structure. MT sensors have been developed that can utilize the background geomagnetic continuum, permitting data to be collected at almost any time. Complete field processing of MT time series in essentially real time now assures that data of adequate quality are being collected. New deployment and processing techniques using remote-reference sensors can mitigate the biases associated with some forms of noise, particularly of cultural origin. Further developments use robust statistical techniques to deal with non-Gaussian aspects of the noise. They produce estimates of transfer functions between the electric and magnetic fields that are more accurate and have more reliable error estimates. Finally, a variety of methods has been devised to stably invert the transfer functions for two-dimensional structure. New computer algorithms are greatly accelerating (by orders of magnitude) the speed with which the necessary computations can be done, and the day seems to be coming when two-dimensional interpretations can be made as the data are collected. Finally, fully three-dimensional inversions are beginning to appear.
Many of the measurement and interpretation developments that are revolutionizing MT can be extended to MV—wideband measurements, remote-reference technique, in-field processing, generalized inverse theory, and so forth. The most important reason that further advances are not being made with MV is that modern equipment in the United States is limited. Another difficulty is that MT fields at a single site are observed to be more highly correlated than MV fields measured between distant sites. Thus, MT measurements result in more precise transfer function measurements. The reason for this needs to be understood and mitigated if possible. It seems likely to be related to stronger effects of nonideal sources on MV than MT. Arrays can characterize the three-dimensional external current distribution that complicates interpretation of the induced response to geological structure. However, the three-dimensional currents in the ground complicate understanding the morphology of external current systems. Thus, MV array experiments complement the needs of the magnetospheric research field. Cooperation between scientists in magnetospheric and solid-Earth induction, particularly with regard to a shared array facility, clearly has much value.
The success of the MT profiling portion of the EMSLAB (Electromagnetic Studies of the Lithosphere and Mantle Beneath [the Juan de Fuca Plate]) experiment in delineating conductivity structure associated with the subducting Juan de Fuca Plate beneath Oregon, and the geologically interesting implications of MT data along all the transects of the Canadian LITHOPROBE project have substantially raised the credibility of MT work. This has resulted in inclusion of an MT component in many of the consortium proposals to the National Science Foundation. Presently, the U.S. academic community has limited access to state-of-the-art wideband MT systems. This situation significantly curtails the development and the application of the technique by university groups, and it needs to be addressed with greater availability of field systems. Furthermore, the complete lack of digitally recording magnetometers for MV work in the United States also means that MT data cannot be collected at periods above 1,000 seconds, although these data are needed to probe the upper mantle. The recent collapse of industrial contractors capable of providing data of the quality required has made it difficult, although perhaps not impossible, to use an industrial alternative for projects amenable to shorter-period data. There has never been a commercial alternative for MV or long-period MT data collection.
The EMSLAB conclusion that fluids are being actively injected into the deep continental crust by the subducted material could not have been reached with data collected entirely on land. As discussed above, the efficient collection of electromagnetic energy by the ocean and the current path to the asthenosphere near the coast have a profound effect on MT and MV data on land. Full use of this opportunity requires data on the deep ocean side to set the boundary condition for the injected current. The United States has one limited facility for ocean bottom MT and MV measurements. These instruments are increasingly being used for long-term experiments of oceanic rather than geological and geomagnetic interest. Moreover, existing instruments are not suitable for the motionally energetic environment of the continental shelves. There are thus severe operational constraints on conducting both onshore-offshore experiments and other MT and MV experiments of marine geological interest; these constraints can be ameliorated by building more instru
Page 129ments. Since the technical issues are primarily repackaging of proven designs and evolutionary incorporation of improvements in components and data storage technology, engineering would be modest. If amortized over a significant number of instruments, the engineering cost would be low, and a new generation of seafloor equipment would be only marginally more expensive than land-based, long-period MT instruments.
A tremendous opportunity for the geomagnetic community will arise over the next decade as hundreds of analog submarine cables are retired from commercial service. They can, for instance, be used to monitor the geoelectric field averaged over planetary spatial scales for years to decades. This will permit extending the long-period limit for MT from 1 day to 10 to 20 days and monitoring the ultra-low-frequency variability of oceanic motions under a variety of circumstances.
Some of the results of motional induction experiments need to be communicated to the wider geomagnetic community. For example, the spectrum of motionally induced horizontal electric fields rises rapidly at periods longer than a few days and is very likely going to determine the precision with which weak signals from Earth's core can be measured. The signal level from the ocean varies strongly with location; it is at least 10 times larger under western boundary currents than in the more quiescent oceanic interior. This has implications for the siting of seafloor observatories. In addition, very little is known about motionally induced magnetic fields, although they are certainly weak compared to external sources at periods of days to months. Whether this relationship continues to longer periods is unknown, yet it clearly impacts the feasibility of secular variation studies with seafloor observatories, and thus the justification for the observatories themselves.
Finally, funding for electrical property research is minimal. Additional support of laboratory studies is necessary for a systematic approach to the problems outlined in this section. While not entirely a geomagnetic issue, encouragement of the National Science Foundation PACEM (Physics and Chemistry of Earth Materials) initiative is appropriate.
Land-based electromagnetic induction research is supported by NSF, DOE, and USGS. The major source of funding from NSF is provided by the Continental Dynamics Program of the Earth Sciences Division, but significant support is also provided by the Geophysics and the Instrumentation and Facilities Programs of this division. Funding of specific research is presently being split by NSF and USGS. USGS also funds several induction research projects carried out by its own personnel. In addition, de facto joint funding is shared between NSF and DOE, because imaging advances supported by the Geosciences Program within the Office of Basic Energy Sciences and instrumental development supported by the Geothermal Technology Division are crucial to the MT profiling projects funded by the Continental Dynamics Program. Support has also been provided by the U.S. Environmental Protection Agency (EPA) and by industry.
For the past decade, most ocean bottom electromagnetic studies have been supported by NSF. This was originally regarded as a marine geophysics topic and supported by the Marine Geology and Geophysics Program of the Ocean Sciences Division. Recent evolution in the specific applications has broadened the support within the NSF to include the Physical Oceanography Program (Ocean Sciences Division) and the Geophysics, the Continental Dynamics, and the Instrumentation and Facilities Programs (Earth Sciences Division). NOAA has supported oceanic electromagnetic research within its Pacific Marine Environmental Laboratory for more than a decade and has supported additional academic investigations through its Atlantic Climatic Change Program.
Currently active laboratories measuring electrical properties of rocks and minerals are being supported by NSF, DOE, and USGS. No interagency funding is presently committed, because the efforts are generally those of individual investigators with few or no graduate students.
For reconnaissance prior to detailed MT profiling and to constrain the three-dimensional context in which the MT interpretation is made, MV arrays should be used. These arrays should additionally be used to understand and correct the effects of external source complications on the geological interpretations. For these purposes, an academic MV array facility (consisting of about 25 digitally recording three-component fluxgate magnetometers) is needed.
To be most effective under the widest range of field conditions, MT instruments need to be mobile and readily deployed, fully remote, referenced for cultural noise cancellation, with complete in-field processing to ensure that quality data are being obtained. Historically the major advances in MT have been made at universities. To guarantee state-of-the-art capabilities and to incorporate new operating modes as they are demanded by new field strategies, maintenance of wideband MT systems at academic institutions are essential—the growth of demand suggests the need for additional systems or that existing systems be upgraded.
To improve our knowledge of mantle conductivity and to understand the constraints it provides on composition, physical state, and dynamics of the Earth's interior require a multifaceted approach. It should include new data, such as ultra-low-frequency, long-baseline MT measurements; improved observatory coverage; better understanding of the effects of source morphology on interpreted conductivity structure; more sophisticated time series processing and inversion methods; and improved laboratory measurements of mantle minerals under controlled thermodynamic conditions.
To expand observatory coverage to otherwise inaccessible areas requires ocean bottom observatories. A small task force should be established to estimate the cost and feasibility of long-term, observatory-quality geomagnetic observation on the ocean bottom. This group should incorporate strong representation
Page 132from scientists and engineers with ocean floor instrumentation experience.
To satisfy the growing interest in utilizing electromagnetic methods for marine geophysical and geological investigations, such as delineating the melt segregation zone beneath spreading ridges and carrying out ocean dynamics studies that can, for instance, constrain long-term climate change, it is necessary that the instrument base be expanded.
To monitor long-term variability of the geoelectric field for both MT and oceanic studies in the deep ocean requires long grounded dipoles. Use of abandoned submarine cables appears to be promising in this context, but it would require close cooperation between scientists and telephone companies.
To extrapolate laboratory results to Earth conditions, it is important to understand point defect chemistry. The influence of minor elements such as hydrogen, nickel, and aluminum on the point defect populations that control solid-state conduction in olivines and pyroxenes must be determined. More extensive use of experimental techniques complementing conductivity, such as measurement of thermoelectric voltages and complex impedance, is required.
To study nonequilibrium electrical properties of water-saturated crustal rocks at elevated temperatures (50 to 500°C) in the laboratory, new experimental techniques are needed. To understand upper crustal conductivities, systematic experimental and theoretical studies of the electrical response of multiphase aggregates and networks are required. Effects of the presence and distribution of other conductivity-enhancing phases such as carbon, magnetite, sulfides, and partial melts must be investigated using advanced experimental techniques that can carefully control thermodynamic variables. Conductivity and complex impedance measurements linked to physical properties such as porosity, permeability, and acoustic velocity in porous water-saturated crustal rocks are needed.
To calibrate apparatus in which new materials are being measured, it is essential that the conductivities of known materials at high temperatures be determined. San Carlos olivine (Fo-90) may be a suitable material. Its conductivity is well known under controlled conditions, it is readily available, it is not excessively resistive, and it is reversibly oxidized or reduced if its stability field is crossed.