Sie sind auf Seite 1von 8

Magnetotellurics

Encyclopedia of Geomagnetism and Paleomagnetism 2007


Martyn Unsworth

Introduction
Magnetotellurics (MT) is the use of natural electromagnetic signals to image subsurface
electrical conductivity structure through electromagnetic induction. The physical basis of the
magnetotelluric method was independently discovered by Tikhonov (1950) and Cagniard (1953).
After a debate over the length scale of the incident waves the technique became established as an
effective exploration tool (Vozoff, 1991; Simpson and Bahr, 2005).

Basic method of magnetotellurics


Electromagnetic waves are generated in the Earth's atmosphere and magnetosphere by a range of
physical processes (Vozoff, 1991). Below a frequency of 1 Hz, most of the signals originate in
the magnetosphere as periodic external fields including magnetic storms and substorms and
micropulsations. These signals are normally incident on the Earth's surface. Above a frequency
of 1 Hz, the majority of electromagnetic signals originate in worldwide lightning activity. These
signals travel through the resistive atmosphere as waves and when they strike the surface of the
Earth most of the signal is reflected. However, a small fraction is transmitted into the Earth and
is refracted vertically downward, owing to the decrease in propagation velocity (Figure M168).
The oscillating magnetic field of the wave generates electric currents in the Earth through
electromagnetic induction, and the signal propagation becomes diffusive, resulting in signal
attenuation with depth. The signals diffuse a distance into the Earth that is defined as the skin
depth, , in meters by

where is the conductivity (S m1), f is the frequency (Hz), and is the magnetic permeability.
The skin depth is inversely related to the frequency and thus low frequencies will penetrate
deeper into the Earth. The impedance of the Earth is defined as

where Ex is the horizontal electric field and Hy is the orthogonal, horizontal, magnetic field. From
this impedance, apparent resistivity can be defined as

and the electric and magnetic field components will have a phase difference
In general, the impedance is written as a tensor,

which relates the horizontal components of the electric and magnetic fields. The impedance
relates the applied magnetic fields to the resulting electric fields and can be considered a transfer
function. Note that the apparent resistivity depends on the ratio of the electric and magnetic field
components. This makes the MT method simple to apply by combining values of Ex and Hy
recorded at different times. The other useful characteristic is that the direction of the incident
wave does not affect the value of apparent resistivity. The apparent resistivity can be considered
an average value of the Earth's resistivity over a hemisphere of radius . Thus, by computing
apparent resistivity as a function of frequency, the variation of resistivity with depth can be
determined. This is illustrated in Figure M169. At high frequency (100010 Hz) the apparent
resistivity is equal to the true resistivity of the upper layer. As the frequency decreases, the skin
depth increases and the MT signal penetrates further into the Earth and the apparent resistivity
rises. The MT phase (xy) is the phase delay between the electric and magnetic fields at the
Earth's surface. The apparent resistivity and phase are related through

where xy is in degrees. Thus, when the apparent resistivity increases with decreasing frequency,
the phase will be less than 45. Similarly, a decrease in resistivity will correspond to a phase
greater than 45. At the lowest frequency, the apparent resistivity asymptotically approaches the
true resistivity of the lower layer, and the phase returns to 45. Note that the phase is sensitive to
changes in subsurface resistivity with depth. For a multilayer model, MT data can reliably
determine the conductance of a layer. Conductance is the vertically integrated conductivity, and
for a uniform layer the conductance is the product of conductivity and thickness. A consequence
of the inverse problem of electrical conductivity is that MT data cannot individually determine
the conductivity and thickness of a layer. Thus layers with differing values of conductivity and
thickness, but the same overall conductance cannot be distinguished with MT.

Figure M168 Propagation of electromagnetic waves from a distant lightning strike to the location
where MT data is recorded. The resistive atmosphere forms a waveguide between the conductive
Earth and ionosphere. The electric field (E) and magnetic field (H) are both orthogonal to the
direction of propagation (k). Note that the electromagnetic energy travels as a wave in the
atmosphere, but diffuses in the Earth. This type of signal propagation occurs above 1 Hz.

Figure M169 Variation of


apparent resistivity and phase
that would be measured at the
surface of a twolayer Earth
model. Note that the depth
sounding of resistivity is
achieved by varying the
frequency of the signal. The
dip in apparent resistivity
below 10 m at 1 Hz is a
resonance phenomenon.

Early studies analyzed MT data in terms of a onedimensional (1D) conductivity model. In this
class of model, conductivity only varies with depth. This approach is sometimes valid in
locations where the geoelectric structure does not change rapidly in the horizontal direction.
However, it is usually necessary to consider at least a twodimensional (2D) Earth model. In this
case, the apparent resistivity computed from Ex and Hy will differ from that derived from Ey and
Hx and the application of a 1D MT analysis can give misleading results. For a 2D Earth, Ex is
dependent only on Hy and Hz, and these three field components comprise the transverse electric
(TE) mode with the impedance (Zxy) computed from Ex and Hy. The transverse magnetic (TM)
mode comprises the Hx, Ey, and Ez field components, with the impedance (Zyx) computed from Ey
and Hx (Figure M170). In a 2D Earth with the xaxis parallel to the geoelectric strike direction,
the impedance tensor can be written as:

Figure M170 Geometry of electromagnetic field components over a twodimensional Earth. The
transverse electric (TE) mode is also called the Epolarization. Similarly, the transverse magnetic
(TM) mode is also called the Bpolarization.

The TE mode is most sensitive to alongstrike conductors. In the TM mode the electric current
flows across the boundaries between regions of differing resistivities, which causes electric
charges to build up on interfaces. Thus the TM mode is more effective than the TE mode at
locating interfaces between regions of differing resistivity.
If the subsurface structure is threedimensional (3D) then all four elements of the impedance
tensor are nonzero. Progress has been made in the last decade in 3D MT modeling and inversion.
However, if a single profile of MT stations is available, and 3D effects can be shown to be small,
then a 2D analysis can be used. If the subsurface conductivity structure exhibits electrical
anisotropy, this will influence the measured impedance tensor. However, it can be difficult to
convincingly distinguish heterogeneity from anisotropy in MT data. Smallscale, nearsurface
bodies can generate electric charges on their boundaries. If the body is small, then insignificant
electromagnetic induction occurs and the only effect is galvanic distortion. This changes the
magnitude of the electric field at the surface and can cause a static shift, which is a
frequencyindependent offset in the apparent resistivity curve (Jones, 1988). The phase curve is
not affected. Static shifts are an example of spatial aliasing. A range of techniques is used to
remove static shifts, and include external measurements of surface resistivity and estimation of
the static shift coefficient in modeling and inversion.

Magnetotelluric data collection and time series processing


MT data are recorded in the timedomain, with the electric fields measured using dipoles 50200
m in length that are connected to the ground with nonpolarizing electrodes.
Audiomagnetotelluric (AMT) data (100001 Hz) typically sample the upper 12 km and are
often used in mineral exploration (see EM, industrial uses). Magnetic fields are measured with
induction coils, and in noisy environments the natural signals are supplemented with a
transmitter. This modified technique is termed controlledsource audio magnetotellurics
(CSAMT). Broadband MT data (10000.001 Hz) are used for sounding to midcrustal depths.
Induction coils are generally used and a recording time of one day is required. In the presence of

excessive cultural noise additional recording may be needed. Noise can originate in a wide range
of sources, including power lines, cathodically protected pipelines, railways, water pumps, and
electric fences. Longperiod magnetotelluric (LMT) data measure very low frequencies (1
0.0001 Hz) and are used for imaging the lower crust and upper mantle. A specialized LMT
instrument is used with a fluxgate magnetometer, solar panels, and low power electronics. MT
data can also be collected on the seafloor.
MT time series data are processed to yield frequencydomain estimates of apparent resistivity
and phase. Modern processing schemes compute fast Fourier transforms of subsections of the
time series and then utilize robust statistical techniques to average the multiple estimates of the
impedance. The application of robust statistics has dramatically improved the quality of
responses and allowed many types of noise to be effectively suppressed (Jones et al. 1989;
Egbert, 1997). In MT data collection, time series data should be recorded simultaneously at
several locations to allow for the removal of noise at the measurement location through the
remotereference method (Gamble et al. 1979). This is important even in locations with minimal
cultural noise (Figure M171). In this example, ground motion from ocean waves caused
oscillations of the magnetic sensors and resulted in the apparent resistivity being artificially low
in the band 30.3 Hz. When the data were processed with a remote reference, the bias was
removed.

Figure M171 Estimates of apparent resistivity and phase at an MT station in central California in
1999. The open circles were derived from remote reference processing, while the black circles
were derived from local data only. The MT data are contaminated by magnetic noise due to
ocean waveinduced ground motion. Note the downward bias in the apparent resistivity in the
frequency band 30.3 Hz when local MT data processing is used.

Magnetotelluric data interpretation


Before modeling or inverting MT data, it is vital to understand the dimensionality of MT data.
Tensor decomposition is a common approach and several techniques exist (Groom and Bailey,
1989; Bahr, 1991). Each method determines how well the measured MT impedance data can be
fit to a 2D geoelectric model and gives an estimate of the geoelectric strike direction. It is
common for a welldefined, consistent geoelectric strike to only be defined in a subset of an MT
dataset. Decomposition can also determine if shallow conductivity structures are causing
galvanic distortion of the surface electric fields. Galvanic distortion is a more general case of the
process responsible for static shifts, and changes both the magnitude and direction of the electric
fields. In extreme cases, nearsurface structure can also cause distortion of the magnetic field
through intense current channeling (Lezaeta and Haak, 2003).
Once the dimensionality has been understood, and distortion addressed, MT can be forward
modeled or inverted in 1D, 2D, or 3D to recover a model of subsurface electrical conductivity.
The inverse problem of electrical conductivity is nonunique (Berdichevsky and Dmitriev, 2002),
which implies that a finite set of MT data containing noise can be reproduced by an infinite
number of geoelectric models. To overcome this nonuniqueness and select a preferred model,
additional constraints must be imposed on the solution. One of the most successful methods is to
require that the geoelectric model derived from the inversion satisfies both the MT data and
some additional requirements (regularization). In the absence of any other geoelectric
information, the most common requirement is that the resistivity model should be as spatially
smooth as possible in the horizontal and vertical directions (Constable et al., 1987). Widely used
inversion algorithms for MT data include those of Rodi and Mackie (2001) and Siripunvaraporn
and Egbert (2000). The fit of the model is usually measured in term of the rootmeansquare
(rms) misfit of the predicted model response to the measured data. An rms misfit significantly
greater than one indicates that the inversion is incapable of fitting the MT data, and usually
indicates excessive noise in the data, or 3D effects that cannot be physically reproduced by a 2D
inversion algorithm. A misfit significantly less than one indicates that either the error bars were
too large or that the data is being over fit. In this second scenario, the resistivity model usually
appears spatially rough, with the appearance of a checkerboard.
The MT method is now routinely used in both commercial exploration and in research.
Commercial applications include exploration for minerals, hydrocarbons, and geothermal
resources (see EM, industrial uses). Researchers use MT to study the structure of the continents
and the dynamics of plate boundaries (Brown, 1994) and also in EM, regional studies. MT
measurements are also made on the seafloor for both commercial and academic investigations
(see EM, marine controlled source).

Crossreferences
Anisotropy, Electrical EM Modeling, Forward EM Modeling, Inverse EM, Industrial Uses EM,
Marine Controlled Source EM, Regional Studies Galvanic Distortion Magnetometers,
Laboratory Periodic External Fields Storms and Substorms, Magnetic Transfer Functions

Bibliography
Bahr, K., 1991. Geological noise in magnetotelluric data: a classification of distortion types.
Physics of the Earth and Planetary Interiors, 66: 2438.

Berdichevsky, M.N., and Dmitriev, V.I., 2002. Magnetotellurics in the Context of the Theory of
Illposed Problems. Tulsa, OK: Society of Exploration Geophysicists.
Brown, C., 1994. Tectonic interpretation of regional conductivity anomalies. Surveys in
Geophysics, 15: 123157.

Cagniard, L., 1953. Basic theory of the magnetotelluric method of geophysical prospecting.
Geophysics, 18: 605635.

Constable, S.C., Parker, R.L., and Constable, C.G., 1987. Occam's inversion: a practical
algorithm for generating smooth models from electromagnetic sounding data. Geophysics, 52:
289300.

Egbert, G.D., 1997. Robust multiplestation magnetotelluric data processing. Geophysical


Journal International, 130: 475496.

Gamble, T.B., Goubau, W.M., and Clarke, J., 1979. Magnetotellurics with a remote reference.
Geophysics, 44: 5368.

Groom, R.W., and Bailey, R.C., 1989. Decomposition of magnetotelluric impedance tensors in
the presence of local threedimensional galvanic distortion. Journal of Geophysical Research,
94: 19131925.
Jones, A.G., 1988. Static shift of MT data and its removal in a sedimentary basin environment.
Geophysics, 53: 967978.

Jones, A.G., Chave, A.D., Egbert, G.D., Auld, D., and Bahr, K., 1989. A comparison of
techniques for magnetotelluric response function estimates. Journal of Geophysical Research,

94: 1420114213.
Lezaeta, P., and Haak, V., 2003. Beyond magnetotelluric decomposition: induction, current
channeling, and magnetotelluric phases over 90. Journal of Geophysical Research, 108,
doi:10.1029/2001JB000990.
Rodi, W., and Mackie, R.L., 2001, Nonlinear conjugate gradients algorithm for 2D
magnetotelluric inversion. Geophysics, 66: 174.

Simpson, F., and Bahr, K., 2005. Practical Magnetotellurics. Cambridge: Cambridge University
Press, p. 270.
Siripunvaraporn, W., and Egbert, G.D., 2000. An efficient datasubspace inversion for
twodimensional magnetotelluric data. Geophysics, 65: 791803.

Tikhonov, A.N., 1950. Determination of the electrical characteristics of the deep strata of the
Earth's crust. Doklady Akademii Nauk, SSSR, 73(2): 295297.
Vozoff, K., 1991. The Magnetotelluric method. In Nabighian, M.N. (ed.), Electromagnetic
Methods in Applied Geophysics, Vol. 2, Chapter 8. Tulsa, OK: Society of Exploration
Geophysicists.

Das könnte Ihnen auch gefallen