Skip to content

Magnetotellurics and Geomagnetic Depth Sounding

MT and GDS for CSIR NET geophysics: natural sources, impedance, apparent resistivity and phase, Bostick depth, TE and TM modes, static shift, induction arrows.

  • 21 min read
  • Updated
On this page · 16 sections
  1. Prerequisites
  2. The Hook: Listening to the Sky to See 100 km Down
  3. Sources of the Natural Field
  4. The Basic Theory
  5. Field Practice
  6. Dimensionality and the Impedance Tensor
  7. The Tipper and Induction Arrows
  8. Geomagnetic Depth Sounding (GDS)
  9. Applications
  10. Memory Tricks
  11. Don’t Confuse
  12. Common Mistakes
  13. Practice
  14. Quick Revision Table
  15. Key Takeaways
  16. What’s Next

Prerequisites

The Hook: Listening to the Sky to See 100 km Down

How CSIR NET asks it. Part C Geophysics: apparent resistivity from E and B in field units, phase above or below 45°, skin and Bostick depths, the source of MT signals by frequency, TE and TM modes, static shift, dimensionality from the impedance tensor, and the direction of induction arrows.


Sources of the Natural Field

Period rangeMain sourceNotes
short periods (frequencies above about 1 Hz)lightning worldwide (sferics) in the Earth–ionosphere waveguidethe audio band used by AMT
long periods (below about 1 Hz)solar wind–magnetosphere–ionosphere interaction: micropulsations, substorms, daily variationthe band of broadband and long-period MT and GDS
near 1 Hz, and near a few kHzweak natural signal: the dead bandsdata are noisier there

The method rests on the plane-wave assumption of the Tikhonov–Cagniard theory: the source field is far away and laterally uniform, so it arrives as a vertically travelling plane wave, refracted almost exactly vertically into the conducting Earth. It fails near the auroral and equatorial electrojets and for very long periods, where the source has structure.


The Basic Theory

Diffusion and skin depth

At the periods used, displacement currents are negligible in the Earth (\( \sigma \gg \omega\varepsilon \)), and Maxwell’s equations reduce to a diffusion equation:

$$\nabla^2\mathbf{E} = i\omega\mu_0\sigma\,\mathbf{E}\qquad (e^{i\omega t}\ \text{time dependence}).$$

In a uniform half-space the fields decay as \( e^{-kz} \), with \( k = \sqrt{i\omega\mu_0\sigma} = (1 + i)/\delta \), and

$$\boxed{\delta = \sqrt{\frac{2\rho}{\omega\mu_0}} \approx 503\sqrt{\rho T}\ \text{m}}\qquad(T = \text{period in s}).$$
\( \rho \) (\( \Omega\,\text{m} \))\( T \)\( \delta \)
100.01 s159 m
5010 s11.3 km
100100 s50 km
1,00010,000 s1,590 km

Impedance, apparent resistivity and phase

The impedance is the ratio of orthogonal horizontal electric and magnetic fields:

$$Z_{xy} = \frac{E_x}{H_y}.$$

Over a uniform half-space,

$$Z = \frac{i\omega\mu_0}{k} = \sqrt{i\omega\mu_0\rho} = (1 + i)\sqrt{\frac{\omega\mu_0\rho}{2}}\quad\Longrightarrow\quad\rho = \frac{\lvert Z\rvert^2}{\omega\mu_0},\ \ \varphi = 45°.$$

Over any Earth, the same formulas define the apparent resistivity and phase:

$$\boxed{\rho_a = \frac{1}{\omega\mu_0}\left\lvert\frac{E_x}{H_y}\right\rvert^2 = 0.2\,T\left\lvert\frac{E_x}{B_y}\right\rvert^2}, \qquad \boxed{\varphi = \arg\frac{E_x}{H_y}}$$

The practical form uses E in mV/km, B in nT and T in s (Cagniard’s formula). It follows from \( H = B/\mu_0 \): \( \rho_a = \mu_0 T\lvert E/B\rvert^2/2\pi \), and the unit conversions give exactly 0.2.

Reading the phase

PhaseMeaning at the depth sensed at that period
\( \varphi = 45° \)uniform resistivity
\( \varphi > 45° \)resistivity decreasing with depth (a conductor below)
\( \varphi < 45° \)resistivity increasing with depth (a resistor below)
\( \varphi \to 90° \)perfectly conducting substratum
\( \varphi \to 0° \)perfectly insulating substratum

The phase responds a little before (at shorter periods than) \( \rho_a \) to a change at depth: it behaves roughly like the slope of the \( \log\rho_a \) curve. Because phase is unaffected by static shift (below), it is the more robust quantity.

Two-layer behaviour (computed)

\( T \) (s)0.0010.010.11101001,000
A: 1 km of 100 over 10 \( \Omega\,\text{m} \): \( \rho_a \)100102.783.627.114.211.210.4
A: phase (°)45.044.261.062.153.348.046.0
B: 500 m of 10 over 1,000 \( \Omega\,\text{m} \): \( \rho_a \)10.010.18.439.2205551822
B: phase (°)45.045.033.312.619.331.739.9

In A the phase rises above 45° near 0.1 s, before \( \rho_a \) has fallen much; in B it drops well below 45° as \( \rho_a \) rises. Both curves approach the basement value only slowly.

Depth transforms: Niblett–Bostick

A quick approximate conversion of \( \rho_a(T) \) and \( \varphi(T) \) to resistivity against depth:

$$\boxed{D = \sqrt{\frac{\rho_a T}{2\pi\mu_0}} \approx 356\sqrt{\rho_a T}\ \text{m}}, \qquad \boxed{\rho_B = \rho_a\left(\frac{\pi}{2\varphi} - 1\right)}\ \ (\varphi\ \text{in radians}).$$

At \( \varphi = 45° = \pi/4 \), \( \rho_B = \rho_a \). At \( \varphi = 60° \), \( \rho_B = \rho_a/2 \) (a conductor below); at \( \varphi = 30° \), \( \rho_B = 2\rho_a \). The Bostick depth is \( \delta/\sqrt{2} \).

Conductance from the low-frequency asymptote

If a conductive layer of total conductance \( S \) sits on an effectively insulating basement and is thin compared with the skin depth, the impedance tends to \( Z \to 1/S \). Then

$$\rho_a = \frac{1}{\omega\mu_0 S^2} = \frac{T}{2\pi\mu_0 S^2}, \qquad \varphi \to 0°,$$

so \( \log\rho_a \) rises with slope +1 against \( \log T \) (the S-line), and any point on it gives \( S = \sqrt{T/(2\pi\mu_0\rho_a)} \). For \( \rho_a = 400\ \Omega\,\text{m} \) at \( T = 100 \) s on the S-line, \( S \approx 178 \) S: for example, about 1.8 km of 10 \( \Omega\,\text{m} \) sediments. This is the MT counterpart of the Schlumberger 45° branch: like DC soundings, MT resolves a thin conductor’s conductance much better than its thickness or resistivity separately.

Over a perfectly conducting substratum at depth \( h \), the opposite limit holds: \( Z \to i\omega\mu_0 h \), \( \rho_a = \omega\mu_0 h^2 \) falls with slope −1 and \( \varphi \to 90° \).


Field Practice

  • Sensors. Two pairs of non-polarisable electrodes (lead–lead chloride or silver–silver chloride), each pair tens to a hundred or more metres apart, for \( E_x \) and \( E_y \). Induction-coil magnetometers for the higher frequencies; fluxgate magnetometers for long periods. Three magnetic components (\( H_x, H_y, H_z \)).
  • Processing. Time series are cut into windows and Fourier transformed; impedances are estimated by (robust) least squares from cross- and auto-spectra at each frequency, with coherence as a quality measure.
  • Remote reference. Noise in the local magnetic channels biases the impedance. Recording the magnetic field simultaneously at a distant quiet site and using it as a reference (whose noise is uncorrelated with local noise) removes this bias. Remote reference processing is standard practice.
  • Cultural noise. Electrified railways, power lines, pipelines with cathodic protection and electric fences can swamp the natural signal; long-period MT near cities is particularly affected.

Dimensionality and the Impedance Tensor

In general each horizontal electric component depends on both magnetic components:

$$\begin{pmatrix} E_x \\ E_y \end{pmatrix} = \begin{pmatrix} Z_{xx} & Z_{xy} \\ Z_{yx} & Z_{yy} \end{pmatrix}\begin{pmatrix} H_x \\ H_y \end{pmatrix}.$$
EarthImpedance tensor
1-D (layered)\( Z_{xx} = Z_{yy} = 0 \); \( Z_{xy} = -Z_{yx} \), the same in every direction
2-D (strike along x), in strike coordinates\( Z_{xx} = Z_{yy} = 0 \); \( Z_{xy} \ne -Z_{yx} \): two independent modes
2-D, rotated off strikeall four non-zero; rotating the axes to strike makes the diagonal vanish (with a 90° ambiguity in which axis is strike)
3-Dall four non-zero in every rotation

Swift skew, a rotation-invariant check:

$$\kappa = \frac{\lvert Z_{xx} + Z_{yy}\rvert}{\lvert Z_{xy} - Z_{yx}\rvert}.$$

It is zero for 1-D and ideal 2-D data; large values suggest 3-D structure (or distortion). Galvanic distortion by small near-surface bodies can produce non-zero skew even over a regional 2-D Earth; decomposition methods (Groom–Bailey) and the phase tensor (Caldwell, Bibby and Brown), which is unaffected by galvanic distortion, were developed to deal with this.

TE and TM modes

For a 2-D Earth with strike along \( x \):

TE mode (E-polarisation)TM mode (B- or H-polarisation)
Fields\( E_x \) (along strike), \( H_y \), \( H_z \)\( H_x \) (along strike), \( E_y \), \( E_z \)
Impedance\( Z_{xy} = E_x/H_y \)\( Z_{yx} = E_y/H_x \)
Currentsflow along strike, never cross boundariesflow across strike, cross boundaries, build up charges on them
Vertical magnetic fieldyes: gives the tippernone
Behavioursmooth across contacts; sensitive to conductors and their conductance; affected by distant conductors (inductive)sharp at contacts (E jumps); sensitive to resistive structures and lateral boundaries; more affected by near-surface galvanic effects

Static shift

Small near-surface heterogeneities (smaller than the shallowest skin depth) accumulate charges on their surfaces, which distort the measured electric field by a real, frequency-independent factor. The result:

  • the whole \( \log\rho_a \) curve shifts up or down by a constant (parallel shift),
  • the phase is unaffected,
  • interpreted directly, a shift by factor \( s \) in \( \rho_a \) scales depths by \( \sqrt{s} \) and resistivities by \( s \).

Diagnosis: \( \rho_a \) curves of TE and TM (or of neighbouring sites) are offset at high frequency although their phases agree, or \( \rho_a \) is offset from what the phase implies. Corrections: tie the high-frequency \( \rho_a \) to an independent measurement at the same site (a central-loop TEM sounding is the usual choice, since TEM uses no electric-field measurement), spatial averaging or filtering over many sites, or solving for shift factors during inversion.


The Tipper and Induction Arrows

Lateral conductivity contrasts make a vertical magnetic field \( H_z \), which over a 1-D Earth is zero. It is linked to the horizontal field by the tipper (geomagnetic transfer function):

$$\boxed{H_z = T_{zx}H_x + T_{zy}H_y}$$

\( T_{zx} \) and \( T_{zy} \) are complex and depend on period. The real and imaginary parts are drawn as induction arrows at each site:

  • Parkinson convention: real arrows point toward zones of concentrated current, i.e. toward conductors.
  • Wiese convention: the real arrows point away from conductors.
  • The length of the arrow measures the strength of the lateral contrast; arrows are short over 1-D structure, and longest at the edges of a conductor.
  • In 2-D, the real arrows are perpendicular to strike.
  • The period at which the arrows are largest indicates how deep the responsible conductor lies.

The coast effect. Sea water is a very strong conductor. Near coastlines, induced currents concentrate in the ocean, and induction arrows (Parkinson) point toward the sea at long periods. This must be modelled before inland anomalies are interpreted.


Geomagnetic Depth Sounding (GDS)

GDS uses only the three magnetic components of natural variations, usually recorded simultaneously by an array of magnetometers.

ApproachWhat is measuredWhat it gives
Magnetometer array studies\( H_x, H_y, H_z \) at many stationsmaps of tipper and induction arrows; Fourier maps of the anomalous fields; location and strike of conductivity anomalies in the crust and mantle
Single-station transfer functionstipper at one sitedirection and rough depth of nearby conductors
Global / long-period GDSvariations such as the daily (Sq) variation and magnetic storms (Dst), recorded at observatoriesseparation of the field into external (inducing) and internal (induced) parts by spherical harmonic analysis; the internal-to-external ratio gives the variation of conductivity with depth in the mantle

From the internal/external separation, or from the ratio of vertical to horizontal fields over large areas, one obtains a C-response (Schmucker–Weidelt transfer function):

$$C(\omega) = \frac{Z}{i\omega\mu_0}\qquad(\text{units of length}), \qquad \rho_a = \omega\mu_0\lvert C\rvert^2.$$

The real part of \( C \) is approximately the depth to the “centre” of the induced currents. It carries the same information as the MT impedance but is obtained from magnetic data alone, which is how global studies show that the mantle becomes much more conductive with depth, especially through the transition zone.


Applications

TargetWhy MT or GDS
Sediments below volcanic cover (for example sub-basalt sediments)basalt is resistive and defeats seismic imaging; MT sees the conductive sediments beneath
Geothermal systemsthe clay cap above a reservoir is a strong conductor
Crustal conductors (graphite, fluids, partial melt) along sutures and fault zonesmapped by tipper and 2-D/3-D MT
Lithosphere–asthenosphere boundary, lithospheric thicknesslong-period MT
Mantle conductivity profileglobal GDS and long-period MT
Groundwater and mineral explorationAMT and CSAMT at shallower depths

Memory Tricks

Don’t Confuse

PairFirstSecond
MT vs GDSE and H; absolute \( \rho \) vs depthH only; lateral conductors, global mantle profile
TE vs TME along strike; \( H_z \) exists; inductiveH along strike; charges on boundaries; galvanic
Skin depth vs Bostick depth\( 503\sqrt{\rho T} \): field at \( 1/e \)\( 356\sqrt{\rho_a T} \): approximate depth of the transformed value
Static shift vs real layeringparallel \( \rho_a \) offset, phase unchangedchanges both \( \rho_a \) shape and phase
Parkinson vs Wiese arrowspoint toward conductorspoint away from conductors
Swift skew vs phase tensorsimple dimensionality check, affected by distortiondistortion-free description of the regional structure
MT vs CSAMTnatural plane-wave sourcegrounded-dipole transmitter, valid only in the far field

Common Mistakes


Practice

All questions below are original practice questions written for this page. They are not from any CSIR NET paper.

Level 1: Foundation

  1. What is the MT phase over a uniform half-space?
  2. Which natural source dominates MT signals above about 1 Hz?
  3. In which 2-D mode does a vertical magnetic field appear?
Answers
  1. 45°.
  2. Worldwide lightning (sferics).
  3. TE mode (E-polarisation, E along strike).

Level 2: Part B Style

Practice Q1. At a period of 10 s, \( E_x = 10 \) mV/km and \( B_y = 2 \) nT. The apparent resistivity is:

(a) 5 \( \Omega\,\text{m} \) (b) 10 \( \Omega\,\text{m} \) (c) 50 \( \Omega\,\text{m} \) (d) 250 \( \Omega\,\text{m} \)

Answer: (c) 50 Ω m

\( \rho_a = 0.2T\lvert E/B\rvert^2 = 0.2\times10\times25 = 50\ \Omega\,\text{m} \). (b) forgets to square the ratio. (d) drops the 0.2. (a) drops the period: \( 0.2\times25 = 5 \).

Practice Q2. For that sounding (\( \rho_a = 50\ \Omega\,\text{m} \), \( T = 10 \) s), the skin depth and the Niblett–Bostick depth are approximately:

(a) 11.3 km and 8.0 km (b) 8.0 km and 11.3 km (c) 3.6 km and 2.5 km (d) 1.6 km and 1.1 km

Answer: (a)

\( \delta = 503\sqrt{500} \approx 11.25 \) km; \( D = 356\sqrt{500} \approx 7.96 \) km \( = \delta/\sqrt{2} \). (b) swaps them. (c) uses \( \sqrt{\rho T} \) with \( T = 1 \) s. (d) uses \( \sqrt{T} \) alone and omits the resistivity.

Practice Q3. At one period, \( \rho_a = 50\ \Omega\,\text{m} \) and \( \varphi = 60° \). The Bostick resistivity and its meaning are:

(a) 25 \( \Omega\,\text{m} \): resistivity decreasing with depth (b) 100 \( \Omega\,\text{m} \): resistivity increasing with depth (c) 50 \( \Omega\,\text{m} \): uniform (d) 75 \( \Omega\,\text{m} \): resistivity increasing with depth

Answer: (a)

\( \rho_B = \rho_a(\pi/2\varphi - 1) \), with \( \varphi = \pi/3 \): \( \pi/(2\pi/3) = 1.5 \), so \( \rho_B = 0.5\rho_a = 25\ \Omega\,\text{m} \). A phase above 45° means a conductor below. (b) is the result for 30°.

Practice Q4. Over a long north–south coastline, with the sea to the west, Parkinson induction arrows at long periods on the land side point:

(a) east (b) west (c) north (d) they vanish

Answer: (b) west

Parkinson real arrows point toward concentrations of induced current, here the conductive ocean to the west (the coast effect), and perpendicular to the coastline (the 2-D strike). (a) is the Wiese convention. (c) is parallel to strike. (d) would hold only over a 1-D Earth.

Level 3: Part C Style

Practice Q5. An MT sounding gives:

\( T \) (s)0.0010.010.11101001,000
\( \rho_a \) (\( \Omega\,\text{m} \))10.010.18.439.2205551822
\( \varphi \) (°)45.045.033.312.619.331.739.9

Which statements are correct?

  1. The near-surface layer is about 10 \( \Omega\,\text{m} \).
  2. The low phases at 0.1–10 s indicate a much more resistive substratum.
  3. At \( T = 1 \) s, \( \rho_a \) is still close to 10 \( \Omega\,\text{m} \) while the phase has already fallen, showing that phase responds to the deeper layer at shorter periods than \( \rho_a \).
  4. If all \( \rho_a \) values were multiplied by 3 by a near-surface body while phases stayed the same, the inferred depth to the resistor would increase by a factor of about 1.7.

(a) 1, 2 and 4 only (b) 1 and 2 only (c) 2, 3 and 4 only (d) all four

Answer: (a) 1, 2 and 4 only

(The data were computed for 500 m of 10 \( \Omega\,\text{m} \) over 1,000 \( \Omega\,\text{m} \).)

  1. True. The short-period plateau with 45° phase gives the top layer.
  2. True. Phases well below 45° mean resistivity increasing with depth.
  3. False. At 1 s, \( \rho_a = 39.2 \), already four times the surface value. The phase leads \( \rho_a \) (look at 0.1 s: \( \rho_a = 8.4 \), barely changed, while phase is 33.3°), but the statement describes 1 s wrongly.
  4. True. That is static shift. Depths scale as \( \sqrt{\rho_a} \): \( \sqrt{3} \approx 1.73 \). Phase-based or TEM-tied interpretation avoids the error.

Practice Q6. A long-period MT site lies on a sedimentary basin. At \( T = 100 \) s, \( \rho_a = 400\ \Omega\,\text{m} \), on a branch where \( \log\rho_a \) rises with slope +1 against \( \log T \) and the phase is near 0°. Which is correct?

(a) The basin fill has a total conductance of about 178 S above a resistive basement. (b) The basement resistivity is 400 \( \Omega\,\text{m} \). (c) The basement is a perfect conductor at about 178 m depth. (d) The sediment thickness is exactly 1.78 km.

Answer: (a)

On the S-line, \( \rho_a = T/(2\pi\mu_0 S^2) \), so \( S = \sqrt{T/(2\pi\mu_0\rho_a)} = \sqrt{100/(7.90\times10^{-6}\times400)} \approx 178 \) S. (b) mistakes a point on a rising asymptote for a plateau. (c) has the wrong limit: a perfect conductor gives slope −1 and phase 90°. (d) is one of many equivalent possibilities: 1.78 km of 10 \( \Omega\,\text{m} \), or 0.89 km of 5 \( \Omega\,\text{m} \), give the same S. Only the conductance is fixed.

Practice Q7. At one site and period, the impedance tensor (in arbitrary units) is \( Z_{xy} = 1 + i \), \( Z_{yx} = -0.9 - 0.9i \), \( Z_{xx} = 0.05 + 0.02i \), \( Z_{yy} = -0.03 + 0.01i \). At another site, \( Z_{xx} = 0.6 + 0.4i \) and \( Z_{yy} = -0.1 + 0.5i \) with the same off-diagonal terms. Which statements are correct?

  1. Swift skew at the first site is about 0.013.
  2. Swift skew at the second site is about 0.38.
  3. The first site is consistent with 1-D or 2-D structure; the second suggests 3-D structure or distortion.
  4. At the first site, \( \lvert Z_{xy}\rvert \ne \lvert Z_{yx}\rvert \) proves the Earth is 3-D.

(a) 1, 2 and 3 only (b) 1 and 3 only (c) 2, 3 and 4 only (d) all four

Answer: (a) 1, 2 and 3 only

\( \lvert Z_{xy} - Z_{yx}\rvert = \lvert 1.9 + 1.9i\rvert = 2.687 \).

Site 1: \( \lvert Z_{xx} + Z_{yy}\rvert = \lvert 0.02 + 0.03i\rvert = 0.036 \); \( \kappa = 0.013 \). Site 2: \( \lvert 0.5 + 0.9i\rvert = 1.030 \); \( \kappa = 0.38 \).

1 and 2: True. 3: True, small skew allows 1-D or 2-D; large skew points to 3-D or galvanic distortion. 4: False: unequal off-diagonal terms with small diagonal terms are exactly what a 2-D Earth (TE and TM differing) gives.


Quick Revision Table

IdeaFact or formula
Sourceslightning above about 1 Hz; magnetosphere below; dead bands near 1 Hz and a few kHz
Governing equationdiffusion \( \nabla^2\mathbf{E} = i\omega\mu_0\sigma\mathbf{E} \)
Skin depth\( 503\sqrt{\rho T} \) m
Apparent resistivity\( \lvert Z\rvert^2/\omega\mu_0 = 0.2T\lvert E/B\rvert^2 \) (mV/km, nT, s)
Phase45° uniform; >45° conductor below; <45° resistor below
Bostick\( D = 356\sqrt{\rho_a T} \); \( \rho_B = \rho_a(\pi/2\varphi - 1) \)
S-lineslope +1, \( \varphi \to 0 \): \( S = \sqrt{T/2\pi\mu_0\rho_a} \)
Perfect conductor belowslope −1, \( \varphi \to 90° \)
1-D tensor\( Z_{xx} = Z_{yy} = 0 \), \( Z_{xy} = -Z_{yx} \)
Swift skew\( \lvert Z_{xx} + Z_{yy}\rvert/\lvert Z_{xy} - Z_{yx}\rvert \)
TE / TME along strike, \( H_z \), conductors / currents across strike, charges, resistors and contacts
Static shiftparallel shift of \( \log\rho_a \); phase unchanged; fix with TEM
Tipper\( H_z = T_{zx}H_x + T_{zy}H_y \); zero in 1-D
Induction arrowsParkinson toward conductors; Wiese away; coast effect
GDSmagnetic-only; arrays for lateral conductors; Sq and Dst for mantle profile; \( C = Z/i\omega\mu_0 \)

Key Takeaways

What’s Next

Advertisement
Advertisement