Appendix 4 Magnetotelluric Electromagnetic Methods

APPENDIX 4 MAGNETOTELLURIC ELECTROMAGNETIC METHODS

A4.1 Introduction methods where an artificially created electromagnetic field is used, but the principles of the method remain the same. As described in Section 5.7.1.1, frequency domain electromagnetic (FDEM) methods are now rarely used in mineral exploration. They still find use in applications requiring A4.2 Natural source magnetotellurics shallow conductivity information, usually surveys associated with environmental monitoring. Exceptions are EM As described in Atmospheric noise in Section 5.4.2.1, there fi methods that rely on natural electromagnetic fields produced are natural time-varying electromagnetic elds originating by natural sources, i.e. the magnetotelluric (MT) methods from the magnetosphere, the region around the Earth (Chave and Jones, 2012; Simpson and Bahr, 2005) and the which includes the atmosphere and the ionosphere, and audio-frequency magnetic (AFMAG) method (Ward et al., at higher frequencies lightning strikes. The time variations 1966). In the case of MT methods, very deep penetration is occur over a wide frequency range (see Fig. 5.26), although ‘ ’ possible, greater than is achievable with the resistivity and there is a dead zone around 0.1 Hz and another around fi fl other EM methods (see Chapter 5), making them attractive 2 kHz. These elds induce circular current ow systems, ’ for exploring for deep targets. As these are passive geophys- several thousand kilometres across, within the Earth s crust fl ical methods (see Section 1.2), only receivers are required. and mantle. The currents ow as horizontal layers and are The relatively simple survey equipment and logistics is an known as telluric currents. fi fi advantage of these methods. However, as with all passive The electric eld (E- eld) of the telluric currents is fl geophysical methods, the use of a natural ‘transmitter’, horizontal, so the currents ow as large horizontal sheets fi fi whose characteristics vary with time and may not always and their associated magnetic eld (H- eld) is also hori- be ideal for the survey objectives, can be a disadvantage. zontal. In accordance with the skin effect phenomenon (see Figure 5.26 shows the natural EM spectrum, which for Section 5.2.3.1), lower frequencies penetrate deeper, the active source electromagnetic methods is a source of noise. actual depth range depending on the conductivity vari- For natural source EM methods this is the ‘primary’ mag- ations. The amplitude, phase and directional relationships fi fi netic field. The full range of frequencies shown in Fig. 5.26 between the E- eld and H- eld are measured for may be exploited (see Fig. 5.1). The lowest frequencies, individual frequencies over a range of frequencies, which down to 0.0001 Hz, are used in academic studies of the allows electrical soundings to be obtained, i.e. variations in Earth’s deep interior and for regional exploration targeting. electrical properties with depth. By making recordings at At higher frequencies extending into the audio range multiple locations, pseudosection and/or pseudovolume (10 Hz to 20 kHz), the method is referred to as audio datasets can be produced. magnetotellurics (AMT) and is used to map shallower features, from tens of metres to several kilometres depth. A4.2.1 Survey practice The airborne AFMAG method makes measurements at fi similar frequencies. The two horizontal components of the electric eld, EX and

Most natural source measurements are made on the EY, are measured using two electric dipoles, each of up to ground, but airborne AFMAG measurements are also 1000 m in length, although a few tens of metres is more possible, speciﬁcally in the form of an MT-related method common. Non-polarising electrodes are used (see Section called ZTEM. Here we describe the natural source EM 5.4.1). The two or possibly three components of the mag- ﬁ methods and also describe controlled source magnetotelluric netic eld, i.e. the perpendicular horizontal HX and HY 2 Magnetotelluric electromagnetic methods

components and the vertical HZ component, are measured A4.3.1 Acquisition of CSAMT data with coils or, for lower-frequency measurements, with The source of the artificial signal is a large electric dipole magnetometers. In MT/AMT the natural ‘transmitter’ is produced by a pair of current electrodes typically 1–3km in effect located at an infinite distance from the survey apart (see Section 5.6.3). The current is transmitted over a area. The electromagnetic wave at the survey area is, there- range of frequencies, usually powers of 2 in the range 8 to fore, a plane-wave, i.e. the field lines are straight and 8192 Hz. The dipole is located a large distance, typically parallel to each other, unlike the curved field found closer 5–12 km, from the survey area so that the current flow in to the source (typical of other electrical and EM methods). the survey area is chiefly horizontal. The transmitter dipole This simplifies the mathematics of calculating the resistiv- is usually orientated perpendicular to the geological strike ity of the subsurface and simplifies interpretation of the (Fig. A4.1) to orientate the electric field, and therefore the survey data (see Section A4.4). current flow, across strike. This is known as the transverse Time series (see Section 2.2) of E-field and H-field magnetic (TM) mode. Orientating the dipole parallel to the variations are recorded. Lower frequencies, i.e. greater geological strike so that the electric field is parallel to strike depth of investigation, require longer recording time. For is known as the transverse electric (TE) mode. AMT surveys, recording periods extend from tens of min- The receiver measures two signals simultaneously: the utes to a few hours. MT measurements typically require horizontal electric field in the direction parallel to the a recording period of a few days in order to investigate transmitter dipole (E ), and the horizontal component of down to mantle depths, and may continue for weeks or X the magnetic field in the direction perpendicular to the even months for studies of the deep interior. transmitter dipole (H ). The electric receivers are dipoles MT measurements are made on a regional scale with Y comprising two non-polarising electrodes (see Section 5.4.1) measurements forming an approximately regular traverse and typically 10–300 m in length. The magnetic receivers or grid, depending on site accessibility. AMT measure- are coils, i.e. dB/dt sensors (see Section 5.7.1.5). It is possible ments are made at the smaller prospect scale and usually to measure all the electric and magnetic components of comprise measurements made at regular intervals along the transmitted field with two perpendicular orientations survey traverses spaced at several times the station interval. of the transmitter dipole, i.e. EX, EY, HX, HY and HZ to obtain the full tensor (see Section 2.2.3). Full-tensor A4.3 Controlled source audio-frequency CSAMT measurements provide information about the magnetotellurics electrical anisotropy of the subsurface (see Section 5.3.1.4). In the commonly used broadside configuration (TM AMT measurements using a man-made source are known mode, Fig. A4.1), a survey comprises a series of readings as controlled source audio-frequency magnetotellurics at stations along one or more traverses parallel to the (CSAMT). The data provide information from depths of transmitter dipole, perpendicular to the regional strike. a few tens of metres to about 2 to 3 km. A detailed descrip- Ideally, measurements are made in the region where the tion of the CSAMT method, with several case studies, is E-field is approximately parallel, and the H-field perpen- provided by Zonge (1992). dicular, to the transmitter dipole. This is a roughly

Transmitter (Tx) AMPS 45° - 60° Strike direction

4 or more Near-field skin depths zone

Receivers E E E E

Survey H Far-field Y H traverse E zone X Figure A4.1 CSAMT survey arrangement. The transmitter E field H dipole is oriented perpendicular to the geological strike field fi (TM mode). Electric (EX) and magnetic (HY) elds are measured in far-field zone. A4.3 Controlled source audio-frequency magnetotellurics 3 triangular zone whose apex is at the centre of the transmit- and parallel to each other (cf. MT/AMT surveys), which ter dipole and whose sides trend 45–60° to the dipole. Note greatly facilitates interpretation of the data. The responses that the length of the survey lines can increase with are nearly equivalent to those recorded in MT/AMT, and increasing distance from the transmitter as the width of so too are the methods of interpretation. When surveying this zone increases outward. The largest distance from the on homogeneous ground, the measured apparent resistiv- transmitter dipole at which readings can be made is ity is independent of transmitter–receiver geometry, and determined by the signal strength necessary to obtain a depth of penetration depends on the frequency and the stable reading at the receiver. resistivity of the subsurface. In the near-field these param- – The magnitude and phase of EX and HY are measured. eters are a function of transmitter receiver geometry, but The magnitudes of both parameters are normalised by the not frequency. In the transition zone the response is transmitter current ,and phase-shift measurements are ref- complex and dependent on the resistivity contrasts and erenced to the signal source by synchronising the receiver their complexity. Clearly, far-field data provide useful timing to that of the transmitter. information about the electrical properties of the sub- Resistivity variations in the subsurface mainly affect the surface; so as far as is possible, measurements should be fi fi fi E- eld, whilst the HY- eld varies quite smoothly over a made in the plane-wave far- eld region of the transmitter large distance. This allows for considerable simplification dipole for each frequency. in the measurement procedure, because for a group of The distance to the far-field from the transmitter dipole neighbouring E-field measurements it is usually only depends on frequency and the resistivity of the ground, necessary to make one central H-field measurement. and is related to skin depth (δ) (see Section 5.2.3.1). When Modern multichannel electrical/electromagnetic receivers making measurements broadside to the transmitter dipole, can be configured for CSAMT measurements and allow the ideally the transmitter dipole should be located a distance responses from multiple E-field dipoles to be recorded of about four skin depths from the survey area so that simultaneously with one associated H-field measurement. measurements are made in the plane-wave far-field. The receiver stacks and averages a series of measurements Typically this is between 5 and 12 km in most geological to improve the signal-to-noise ratio (see Section 2.7.4.1) environments, but is often determined by logistical and filters the data to attenuate powerline and telluric noise constraints. To ensure that this is the case the electrical (see Section 5.4.2). The whole measurement cycle usually conductivity of the ground must be known, which is, of takes less than one hour. course, the object of the CSAMT survey; another example of the geophysical paradox (see Section 1.3). Since skin depth varies with frequency, in practice most datasets A4.3.2 Near-field and far-field measurements contain both transition-zone and near-field measurements. The electromagnetic field about the transmitter dipole is At a particular station in a particular geo-electrical envir- described in terms of a near-field and a far-field. In the onment, a near-field response may be measured at low near-field (close to the dipole), the amplitude of the electric frequencies, transitioning into a far-field response at higher field decreases with distance from the centre of the trans- frequencies. The range of transition frequencies where mitter dipole as 1/distance3, and that of the magnetic field the field pattern changes depends on distance from the as 1/distance2. Both the E- and H-fields are strongly transmitter dipole. curved. In the far-field (far from the dipole), amplitude of Correct interpretation requires recognition of when the electric field decreases by 1/distance3, as does that of responses are near-field, transition zone or far-field for the magnetic field. There is a transition zone between the each measurement frequency, an important issue discussed near- and far-fields where the rate of decrease in the in Section A4.5.1. strength of the magnetic field varies from 1/distance2 to 1/distance3 moving away from the dipole. This places A4.3.3 Survey design restrictions on survey layout (see Section A4.3.3) and raises several issues when interpreting the data. The transmitter dipole needs to be located in a technically Importantly, in the far-field, the electric and magnetic and logistically appropriate location with respect to the fields penetrate nearly vertically into the ground and are survey area and the regional geology, with stringent approximately plane-waves, i.e. the field lines are straight consideration for near- and far-field effects. Computer 4 Magnetotelluric electromagnetic methods

modelling can assist in determining the optimum location A4.3.3.2 Resolution of the transmitter dipole. Also, the response of the Vertical resolution is usually about 5 to 20% of the depth of expected electrical structure of the survey area can be investigation, depending upon the complexity of the sub- modelled so that the optimum length of the receiver’s surface electrical variations and the local noise levels. E-field dipole and the measurement interval can be deter- Lateral resolution is controlled by the length of the mined, and the resolution and depth of investigation of the receiver’s E-field dipole, with resolution approximately frequency range available from the transmitter can be equal to the dipole length. However, the amplitude of the investigated. Like other electrical methods, the survey line measured signal also depends on the dipole length, so the spacing is determined by the strike length of the target with actual length will be a compromise between achieving the spacing set so that at least two, and preferably three, maximum resolution and maintaining an adequate meas- survey lines pass over the target zone. Note that uniform urement signal-to-noise ratio. Lateral resolution at depth is conductivity/resistivity is the usual assumption of the also dependent upon array size and frequency, with reso- target, but anisotropic and heterogeneous conductivity/ lution decreasing with decreasing frequency (increasing resistivity structures are common. wavelength). In horizontally layered ground where a grid of widely spaced discrete soundings may be more appropriate than a A4.4 Reduction of MT/AMT and CSAMT data series of closer-spaced soundings along widely spaced survey lines, the choice of station interval should take into Reduction of survey data involves editing for noise- account the resolution needed to resolve expected fault contaminated readings and various forms of filtering may offsets or lateral changes in the thickness of the layering. be applied (see Section 2.7.4.3). Remote reference process- ing can be used to account for cultural electrical noise. It is A4.3.3.1 Depth of investigation based on comparison of simultaneously recorded data For far-field measurements the depth of investigation from different stations. Noise sources are largely the same δ (Zinvestigation) is related to skin depth ( ) (see Section as those for other types of EM surveys (see Section 5.4.2). 5.2.3.1) and is given by: rffiffiffi δ ρ pffiffiffi A4.4.1 Resistivity and phase difference Zinvestigation ¼ ¼ 356 ðA4:1Þ 2 f Impedance of the subsurface is determined from the where δ is in metres, ρ is the true resistivity of the perpendicular components of the horizontal electric and ground in Ω m and f is the frequency in Hz. As with all magnetic fields. Impedance (Z ¼ E/H), in ohms (Ω), is a frequency domain EM techniques, an inverse relationship measure of the opposition to the flow of alternating electric between depth of investigation and frequency applies; currents (see Section 5.2.1). An MT/AMT impedance also greater depth penetration is achieved in resistive tensor (see Section 2.2.3) containing four ‘transfer func- ’ fi environments. tions (ZXX etc.) can be de ned as follows: In practice, depth of penetration is usually limited to Z Z H E about 3 km because of constraints set by the lowest fre- XX XY X ¼ X ðA4:2Þ Z Z H E quency that avoids the near-field zone for the transmitter– YX YY Y Y receiver separation used. Also, signal-to-noise levels meas- It is from these quantities that the conductivity structure of ured at the receiver ultimately limit the weakest signal that the subsurface can be interpreted. If the subsurface elec- can be detected; recall that the fields fall off as 1/distance3 trical structure is 1D (see Section 2.11.1.3), i.e. consists of in the far-field zone, which is a very rapid rate of decline. horizontal electrical layers, or is 2D with the X-direction ¼ ¼ CSAMT also has a minimum depth penetration because parallel to strike and Y perpendicular, then ZXX ZYY 0. the top ten metres or so are transparent, depending upon If the strike direction is known the data can be mathemat- ‘ ’ near-surface resistivity and the frequencies used. Very ically rotated if required. In this case, the non-zero ZXY

shallow variations may be inferred from static effects (see and ZYX impedance can be used to determine apparent Section A4.4.2), but in general the shallow subsurface is resistivity (see Section 5.6.2.1) as a function of frequency ρ ρ better investigated using time-domain EM systems (see using Eq. (A4.2) where XY uses EX and HY and YX uses

Section 5.7). EY and HX as follows: A4.5 Display and interpretation of MT data 5

2 charges to accumulate on the surface of local near-surface ρ ¼ 1 EX ð : Þ XY TE mode A4 3 electrical features. It affects the magnitude of the E-field ωμ HY data and the calculated apparent (Cagniard) resistivity 2 fi ρ ¼ 1 EY ð : Þ (Eq. (A4.5)); the H- eld data and phase difference are YX ωμ TM mode A4 4 HX unaffected. The charge accumulation shifts the resistivity sounding curve parallel to the resistivity axis, either up or and where ω (¼ 2πf) is the angular frequency of the fields down depending upon the polarity of the charge. It is a and μ is the magnetic permeability of the subsurface (see static offset across all frequencies and is known as the static Section 3.2.2.3. For most rocks μ ≈ μ ; see Zhdanov and 0 effect (Fig. A4.2a). Importantly, the shape of the resistivity Keller (1994)). Equations (A4.3) and (A4.4) show that curve is not affected. Failure to account for this will lead to apparent resistivity is related to the square of the imped- incorrect estimates of resistivity and depth during data ance, which is volumetrically averaged over the penetration modelling. depth of the signals. After rotation, XY data are referred to Static shift is dependent upon the size and depth of the as transverse electric (TE) mode, and YX as transverse body causing it, its resistivity contrast, and the wavelength magnetic (TM). In the TE mode the electric field is parallel of the fields with respect to body size. It is also dependent to strike and the magnetic field is perpendicular to strike. on the length of the E-field receiver dipole relative to the In the TM mode, the magnetic field is parallel to strike and size of the body producing the effect, and the location of the electric field is perpendicular to strike. the dipole with respect to the body. Large dipoles spatially For CSAMT measurements, where strike is usually filter near-surface inhomogeneities and minimise static known and used to orientate the transmitter and receiver effects, but at the cost of reduced lateral resolution. Shallow dipoles, and most commonly E and H measurements X Y features produce the greatest effect. are made, it is usual to calculate the Cagniard resistivity There are various approaches to accounting for static (ρ ). For E-field in millivolts/km and H-field in Cagniard shift. For example the shift may be treated as an unknown nanotesla, Eq. (A4.3) can be written as: variable in the modelling of the data, by averaging or

2 filtering the data from adjacent stations, and using the ρ ¼ 1 EX ð : Þ Cagniard A4 5 phase measurements to estimate apparent resistivity. Alter- 5f HY natively, static shifts can be estimated using time-domain The Cagniard resistivity is only valid for a plane-wave (TD) electromagnetic (see Section 5.7.4.3) or resistivity source, i.e. for measurements in the far-field in the case soundings (see Section 5.6.6.1), if available. The data are of CSAMT. In homogeneous ground it represents the inverse modelled to obtain a two- or three-layered 1D true electrical resistivity of the ground. In heterogeneous Earth model that is consistent with the observed TD data. ground it is the apparent resistivity. In the near-field the This model is then used to forward model an MT response, Cagniard resistivity over-estimates the ground resistivity, which is compared with the observed MT data. The MT and a correction can be applied. curves are translated so that they overlie the time-domain For both natural source and controlled source data, data. To achieve this with accuracy requires the time phase difference (sometimes referred to as phase, ϕ) is the domain-derived and MT data curves to overlap. Figure lead of the electric field with respect to the magnetic field A4.2d shows corrected apparent resistivity data, in this and is usually expressed in milliradians. Phase difference is case the correction being determined from the phase related to changes in resistivity at a particular frequency. It measurements. is insensitive to static effect (see Section A4.4.2) and in the case of CSAMT it is sensitive to the transition zone. A4.5 Display and interpretation of MT data Magnetotelluric data are commonly displayed on logarith- A4.4.2 Static effect mic scales, with apparent resistivity or phase shift plotted MT, AMT and CSAMT data are prone to ‘static shift’ due against period (1/frequency; see Appendix 2), the latter to heterogeneous electrical properties in the near surface at being used as a proxy depth parameter. For CSAMT data, a scale smaller than the resolving capability of the data frequency decreasing downwards is commonly used as the (Dennis et al., 2011). The electric field causes electric vertical axis, also to reflect increasing depth of 6 Magnetotelluric electromagnetic methods

a) investigation. In both cases a series of soundings measured Apparent resistivity along a survey traverse are used to construct a pseudosec- rr 12 1 2 3 tion (see Section 2.8.1). Calculated parameters for a par- 3 1 2 ticular frequency or pseudo-depth can be displayed as 2D Static Conductive shifts zone contour plots and coloured images (see for example Figs. A4.2b–d). r 1 All measured parameters, i.e. E-field magnitude and phase, and H-field magnitude and phase, as well as r 2 apparent resistivity and phase difference can provide some Log frequency Section information to assist geological interpretation. Changes in rr 12> resistivity of the ground are reflected predominantly in 0 100 changes in the E-field data, whereas the H-field data b) Metres resolve large-scale resistivity layering across the survey 0.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5 16.5 area and large-scale 2D and 3D electrical structures. Static 4096 High fi fi 2048 50 effects can be identi ed by comparison of E- and H- eld 50 20

1024 63 data, and if present approximate corrections can be 40 13 512 6 40 79 1 applied. The most common types of CSAMT data used

32 16 32 256 10

0 for interpretation are variations in Cagniard resistivity and 10 20 128 13 phase, so these are emphasised here. However, as with 8 25 Frequency (Hz) 64 25 Low High induced polarisation and resistivity data (see Section 32 5 10 16 5.6.6.3), interpretation of pseudosections can be a challen- ging task and there is an increasing reliance on the results c) of inverse modelling (see Section A4.5.4). 0.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5 16.5 4096 Crucial to interpretation of CSAMT data is the recogni- High 2048 100 fi fi 0 tion of non-far- eld zone responses in both the E- eld and 700 70 1024 0 H-field data. This is not a problem for MT/AMT data

90 512 950 900 600 0 which are all measured in the ‘far-ﬁeld’.

256 0 850 800 85 128 0 Low 900 55 750 Frequency (Hz) 64

32 A4.5.1 Recognising far-ﬁeld responses 16 in CSAMT data

d) Figure A4.3 shows Cagniard resistivity and phase difference 0.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5 16.5 – 4096 versus frequency for different transmitter receiver separ- 2048 ations over a 1D, three-layer resistivity model with the

40 25 25

3 50

1024 0 63 5 middle layer having the lowest resistivity. These are elec-

32 32 20 512 40 16 79 trical sounding curves like those described in Section

3 16 256 13 5.6.6.1. The AMT data are a far-ﬁeld response, whilst the

128 100 20 13 ﬁ Low CSAMT data show the near- eld and transition zone Frequency (Hz) 64 High responses deviating from this at low frequencies. The 32 10 5 16 AMT data are easy to understand, recalling that frequency is a proxy depth parameter. For the model shown, at high Figure A4.2 Illustration of static effects in CSAMT data. frequencies (shallower depths) the apparent resistivity (a) A surface zone of higher conductivity causes static shifts, of approaches that of the upper layer, 100 Ω m. As frequency opposite sign, in the soundings obtained at locations (2) and (3), relative to the true sounding (1). (b) Apparent (Cagniard) resistivity (Ω m) pseudosection containing static shifts. Note the highly station 11.5, suggesting it is due to static shift. (d) Data in (b) after resistive dyke-like feature near station 11.5 and a deeper resistive correction for static shifts. The feature near station 11.5 is no longer feature near station 6. (c) Equivalent phase difference (mrad) data present. (b to d) Redrawn, with permission, from Zonge and Hughes to (b). Note the lack of evidence for the resistive feature near (1991). A4.5 Display and interpretation of MT data 7

f r Ω Phase ( ) Cagniard ( m) identifying the zone. The notch is best developed when 010(p /4) 1 102 103 104 there is a low-resistivity layer in the section, especially when it overlies a more resistive basement. 1 32 km Phase responses are characterised by values close to zero 2048 2 16 km in the near-field, with an abrupt transition to far-field 3 8 km 512 4 4 km 100 W m responses. Phase difference data can reveal the nature of 100 m 5 2 km 10 W m fi 200 m the electrical layering of the survey area. In the far- eld, 128 1000 W m π 5 phase difference for homogeneous ground is /4 radians, 4 i.e. the E- and H-fields are out of phase by 45°. Phase 32 3 Notch 5 difference greater than π/4 indicates that the ground is 2 8 3 getting more conductive with depth or, if less than π/4, 4 1 45° Slope 1 getting more resistive. 2 2 Varying the transmitter–receiver separation has the AMT AMT Frequency effect of changing the frequencies at which the changes (Hz) between the various zones occur. For the maximum separ- fi Figure A4.3 The apparent (Cagniard) resistivity (ρ) and phase ation of 32 km the far- eld exists at frequencies higher than difference (ϕ) frequency soundings for a range of CSAMT about 8 Hz. At the shortest separation (2 km) the far-field transmitter–receiver separations for a three-layer 1D resistivity responses occur at frequencies higher than about 256 Hz. model. The AMT data represent an entirely far-field response. Note In pseudosections, near-field and transition zone how the AMT resistivity sounding curves mimics the actual electrical responses produce horizontal contours/elliptical shapes, structure of the subsurface. The CSAMT curves for different which may be misinterpreted as electrical layering; for transmitter to receiver separations show the onset of non-far-field responses occurring at increasingly higher frequencies for smaller example see Figs. A4.10a and b. It is advisable to study separations. All the CSAMT apparent resistivity curves show a data from individual stations to identify the range of fre- well-developed transition zone notch with the near-field response quencies comprising the far-field zone. developing at lower frequencies. The near-field response is characterised by a near-constant slope to the curves and values of apparent resistivity that are higher than the true resistivities of A4.5.2 Model responses the model. Figures A4.4 and A4.5 show CSAMT pseudosections for three simple subsurface conductivity models. The data are decreases (increasing depth) the apparent resistivity for an E-field/transmitter/survey traverse perpendicular to decreases towards that of the next deeper layer, but does geological strike (TM mode) and in the far-field zone for not reach the actual value of 10 Ω m. At the lowest fre- all frequencies. quencies the apparent resistivity curve converges towards Figure A4.4 shows apparent resistivity and phase the true resistivity of the lower part of the model, 1000 Ω m. responses of a vertical contact. The pseudosections resem- Considering the CSAMT response for 8 km separation, ble the distribution of physical properties with the lateral note how the far-field response occurs at frequencies position of the contact being fairly accurately resolved. The higher than 64 Hz, recognised by its good correlation with horizontal contours within each zone do not depict the the AMT (far-field data) curve. The transition zone occurs actual structure. Instead these reflect changes in the elec- between about 8 and 64 Hz and the near-field responses tromagnetic field with frequency, cf. Fig. A4.3. In Figure occur below 8 Hz. Measuring in the near-field causes the A4.4 true values of subsurface resistivity are measured at calculated resistivity to double in value as the frequencies about 4 Hz. are halved, creating the false impression of high resistivities The responses of a compact source with long strike extent, at depth. On a plot of log-resistivity versus log-frequency, a with either a greater or lesser resistivity than its surrounds, slope of –45° indicates near-field data. Apparent resistivity are shown in Fig. A4.5. The presence of the anomalous body pseudosections show many closely spaced, near horizontal, is clear in every case with better definition of its top surface contours. The behaviour of the apparent resistivity curve in than of its base, creating the impression of a dyke-like the transition zone is variable, but a common response is a source. As with the contact model, the lateral extent of the low-resistivity ‘notch’ or shoulder, which is useful for source is particularly well resolved – acharacteristicofthe 8 Magnetotelluric electromagnetic methods

a) Location (X ) a) Location (X ) 0 0 50 W m Source 500 500 100 W m 10 W m Depth (m) Depth (m) 100 W m 1000 1000

b) Location (X ) b) Location (X ) 8192 8192 2048 50 50 2048 40 512 63 32 512 25 100 128 79 20 128 8.9 16 6.3 100 100 32 13 32 5.0 8 10 8 100 4.0 Frequency (Hz) Frequency (Hz) 2 8 2 6 0.5 0.5

c) Location (X ) c) Location (X ) 8192 8192 725 2048 950 2048 775 512 675 512 1025 128 700 975 128 32 725 925 32 775 875 900 8 8 00 8 Frequency (Hz) Frequency (Hz) 850 2 2 775 0.5 0.5

0 500 Metres d) Location (X ) 8192 Figure A4.4 Pseudosections of the far-ﬁeld TM mode CSAMT 2048 responses of a resistivity contact. The survey traverse and the 512 ﬁ 100 transmitted E- eld are perpendicular to the contact. (a) Model, 128 Ω (b) apparent (Cagniard) resistivity ( m) and (c) phase difference 32 100

(mrad). Redrawn, with permission, from Zonge (1992). 8 0 126 10 158 Frequency (Hz) 2 0.5

e) CSAMT method, subject to the controls on lateral resolution Location (X ) 8192 described in Section A4.3.3.2. The true values of resistivity 2048 are not obvious in the apparent resistivity data. 512 128 700 32 800 800 750 A4.5.3 Interpretation pitfalls 8 Frequency (Hz) 2 Static effects cause significant interpretational errors if not 0.5 recognised. Figure A4.2b shows static shift in resistivity 0500 data presented as pseudosections. The shift causes large Metres fl lateral changes in apparent resistivity, re ected by the Figure A4.5 Pseudosections of the far-field TM mode CSAMT closely spaced vertical contour lines. If these were due to responses of a localised source whose strike is perpendicular to the lateral variations in the geology then equivalent features survey traverse and the transmitted E-field. (a) Model, (b) and (c) Ω would be observed in the phase data (Fig. A4.2c). apparent (Cagniard) resistivity ( m) and phase difference (mrad) when the resistivity of the source (10 Ω m) is less than its surrounds. In addition to issues related to the CSAMT near-field (d), (e), As for (b) and (c) with the source more resistive (5000 Ω m) and transition zone responses, the interpretation of than its surrounds. Note the excellent definition of the position and CSAMT and AMT soundings suffer all the same problems top of the source, but not so its base. Redrawn, with permission, of ambiguity as for the other electrical methods. For from Zonge (1992). A4.5 Display and interpretation of MT data 9 example, MT/AMT data respond to the ‘conductance’ of a A4.5.4 Modelling source, i.e. the product of its conductivity and its thickness. Like multifrequency EM data, the soundings and pseudo- The two variables cannot be separated. sections from MT measurements can be transformed to depth models by inversion methods. Like all inverse A4.5.3.1 Topographic effects modelling techniques (see Section 2.11.2.1), simplifying Topography local to the receiver dipoles affects CSAMT assumptions must be made about the local geology, which measurements in the same way it affects other electrical can adversely affect the results. Inversion methods may methods (see Section 5.6.7.3). The effect of topography work only with far-field data, requiring other responses depends on its orientation relative to the electric and to be manually edited out of CSAMT data. Alternatively, magnetic fields. When the topography strikes perpendicu- near-field measurements can be corrected so as to mimic lar to the electric field (TM mode), hills disperse the far-field responses. The corrections are approximate equipotential and current-flow lines whilst valleys focus because they require simplifying assumptions to be made them (Fig. A4.6a). This produces deep resistivity highs about the subsurface, and the resulting errors propagate under valleys and lows under hills with overshoots at the into the inversion. More sophisticated algorithms work anomaly edges (Fig. A4.6b). Moderate topography can have with the uncorrected data from the near-field and asignificant effect. When the strike of the topography is transition zones as well as the far-field, e.g. Routh and parallel to the electric field (TE mode), the effects on the Oldenburg (1999). data are smaller and less complicated, with the artefacts Inversion algorithms may model the data from each being shallower and of opposite polarity to those of TM station independently, i.e. produce a series of 1D models, mode (Fig. A4.6c). If the topography is known its effects can with the results plotted side-by-side to construct a be included in the computer modelling of the survey data. parasection. More sophisticated are 2D inversions which CSAMT is insensitive to the topography between the combine all the available data from a traverse and can transmitter and receiver dipoles because the apparent include topography in the model. As would be expected, resistivity and phase difference are affected only by the the 1D approach works well when its assumptions, i.e. flat geology local to the receiver, and not that between the ground surface and horizontal electrical property layering, transmitter and receiver dipoles. CSAMT therefore finds are valid. Three-dimensional inversion methods are also application in mountainous terrains. available but are computationally expensive. In MT data

a) Hill Equipotential surface Bluff Ground Current flow line surface Valley

Current Current Current dispersion focusing dispersion

b) H H H L H L

Figure A4.6 Effects of topography. (a) Effects of

Frequency (Hz) H Frequency (Hz) Frequency (Hz) topography on current flow for the transmitted E-field oriented perpendicular to the strike of the topography c) (TM mode). Topographic effects in E-field measurements for the E-field oriented (b) as in (a), and (c) parallel (TE mode) to the strike of the topography. H – high, L – low. Redrawn, with H L L H L H L H L Frequency (Hz) Frequency (Hz) Frequency (Hz) permission, from Zonge and Hughes (1991). 10 Magnetotelluric electromagnetic methods

sets, the low frequencies used mean that a large volume of A4.7 Examples of magnetotelluric data crust is ‘averaged’ when calculating apparent resistivities The CSAMT and AMT methods can be used at the etc. This means that assumptions of 1D, 2D etc. are less prospect scale for target detection. AMT has successful likely to be valid. It is possible to estimate where the data detected deep conductive targets such as massive nickel can be adequately represented by, for example, a 2D model. sulphide mineralisation at Voisey’s Bay (e.g. Zhang et al., A commonly used approach is based on the ‘phase tensor’; 1998). The mapping of a kimberlite is described in Section see Caldwell et al. (2004). The ‘dimensionality’ of the data A4.7.1. Another use of AMT relies on the association is estimated for defined frequency bands and those data between unconformity-type uranium deposits and graphite- that are ‘3D’ can be excluded from modelling. Unfortu- bearing conductive shear zones below the unconformity. nately, the ‘3D’ data may be a significant proportion of the The area of the McArthur River Mine, Athabasca Basin, whole dataset. Saskatchewan, Canada, has been studied in detail as part of a project (EXTECT-IV) to develop exploration methods for this style of mineralisation (Jefferson and Delaney, 2007). A4.6 MT versus other electrical and Several case study papers describe the use of AMT in this EM methods context. Tuncer et al. (2006) describe data that map CSAMT has the following distinct advantages over basement conductors to a depth of 2–3 km, and also that conventional resistivity and induced polarisation methods. possibly contain a response from the zone of silicification The measured parameters are free from electrode geomet- associated with the mineralisation. rical effects characteristic of the pseudosection plots of the CSAMT can be used to detect conductive targets at various electrode arrays (see Section 5.6.6.3) since the depth, typically massive sulphides or hydrothermal alter- receiver in CSAMT samples only the ground nearby which ation zones (e.g. Takakura, 1995; Basokur et al., 1997), but is independent of the distant transmitter; CSAMT is rela- importantly it can also detect resistive targets such as tively insensitive to cultural conductive features; CSAMT silicified zones associated with gold mineralisation (see provides greater penetration; and it is logistically more Section A4.7.2). Other applications include mapping fea- efficient than electrical resistivity methods where smaller tures such as faults, the weathered zone, lithological dipoles are required to increase lateral resolution, which changes in the basement and geothermal zones. reduces depth penetration. Importantly, compared with A use of MT by the diamond industry is identifying other EM methods, CSAMT can, through the measure- regions of thick lithosphere and mantle regions above the ment of the electric field, discriminate resistive as well as graphite–diamond stability field that possibly contain sig- conductive zones whereas EM responds only to conductive nificant quantities of carbon (Jones and Craven, 2004). features. CSAMT responds best to resistivity contrasts Also, based on recent ideas about the importance of deep rather than the discrete conductors which respond well to crustal and mantle processes in the formation of mineral other forms of EM surveying. deposits and, in particular, the prospectivity of regions CSAMT is logistically simpler when surveying local at the edge of major cratonic blocks (Begg et al., 2010; grids, but in broad area reconnaissance surveying, the need McCuaig et al., 2010), the method is being used at terrain to move the large distant transmitter dipole makes it scale to locate deep zones of alteration and major fault/ inefficient. CSAMT is more efficient than other EM and suture zones. Wannamaker and Doerner (2002) describe galvanic electrical methods when surveying small areas in work from the Carlin Trend in Nevada; Juanatey et al. hilly terrains and densely vegetated areas as there is no need (2013) describe work from the Skellefte district in Sweden; to move the transmitter loops and dipoles. On the other and Dentith et al. (2013) describe data from the southern hand, in flat open areas EM and galvanic resistivity Yilgarn Craton, Western Australia. Figure A4.7 shows the methods are faster, and they can produce higher resolution. 2D inversion results of an MT survey that passed close The main advantage of MT/AMT methods is their large to the Olympic Dam Cu–U–Au–Ag–REE deposit in depth penetration compared with active EM methods. The South Australia, described by Heinson et al. (2006). The use of only receivers means surveys are relatively cheap. data demonstrate mapping of the edge of a block of (resist- However, they offer less resolution than active resistivity ive) Archaean crust and also show that there is a deep zone and EM methods. of more conductive crust under the deposit itself. The A4.7 Examples of magnetotelluric data 11

SW NE Olympic Dam X MT stations deposit Location ( ) 0 Proterozoic sediments Proterozoic sediments Granite batholith

10 Resistive middle crust

Fault/suture Resistive Archaean Conductive crust - Depth (km) 30 crust due to mineralising fluids? Resistivity ( W m) 1826 40 0 20 350 Kilometres 20 50

Figure A4.7 Resistivity cross-section derived from 2D inversion model of an MT dataset from near the Olympic Dam Cu–U–Au–Ag–REE deposit. Vertical exaggeration ¼ 2. A large zone of conductive lower crust extends upward to underlie the deposit. Based on a diagram in Heinson et al. (2006).

reason for the higher conductivity change is uncertain, but deposited in the crater near the surface with diatreme Heinson et al. suggest it may be due to graphite precipi- facies below. tated in the deep crust by metal-bearing hydrothermal ﬂuids. A4.7.2 CSAMT response of the Golden Cross epithermal Au–Ag deposit

This example illustrates the application of CSAMT to map A4.7.1 AMT response of the Regis zonation in a hydrothermal alteration system. The Golden kimberlite pipe Cross epithermal Au–Ag deposit is located in the Waihi- The use of detailed 3D AMT to define the geometry of the Waitekauri region of the North Island of New Zealand (see diamondiferous Regis kimberlite pipe, Minas Gerais, Fig. 4.20). It is a low-sulphide epithermal deposit with Brazil, is illustrated by this example. The results are mineralisation occurring in quartz veins within a hydro- described by La Terra and Menezes (2012) and are sum- thermal alteration system hosted by Miocene–Pliocene marised in Fig. A4.8. The data comprised 111 stations. volcanic rocks (Hay, 1989; Simmons et al., 2000; Simpson Typical acquisition time for each station was 40 minutes, et al., 2001). The deposit was discovered as a result of with frequencies recorded in three bands spanning the geological investigation of a zone of low magnetic relief range 10 Hz to 100 kHz. To counter the deadband in the (see Fig. 4.20). Figure A4.9 shows a geological map of the natural EM spectrum at about 1–10 kHz (see Fig. 5.26) area and a cross-section in the vicinity of the deposit. the natural source was supplemented by an EM transmitter Zonge and Hughes (1991) describe CSAMT data from the positioned so that the survey area was in the far-field zone area, and Collins (1989) provides a more general descrip- (see Section A4.3.2). tion of the geophysical characteristics of the deposit. Drilling-constrained 3D inverse modelling of the data Mineralisation occurs in a silicified zone surrounded by shows the pipe, as is usual, to be more conductive than the argillic and propylitic alteration zones. Unconformably country rocks. The conductive zone extends to around overlying the mineralised sequence is an unaltered andesite 550 m and is broadly conical with a greater diameter near unit. The silicified zone has a high resistivity of about the surface. The modelled geometry is consistent with the 1000 Ω m. The argillic zone has a low resistivity of about ideal form of a kimberlite pipe comprising sediments 1 Ω m and the propylitic zone has a slightly higher 12 Magnetotelluric electromagnetic methods

a) Easting (m) a) 8500 9000 9500 1000

2500

2800 3000 3400 3600 3800 2000 4850N

Kimberlite 1500 Northing (m)

Stations 1000 0 200 Metres

0 500 500 Whakamoehau Andesite Silicic alteration Metres 9300E Waiharakeke Dacite Argillic alteration b) S N Waipupu Formation Propylitic alteration Northing (m) Quartz veins 500 1000 1500 2000 2500 0 9300E W E Location (m) 100 30003200 3400 1000 200 299 500 83 b) Surface 300 24

Depth (m) 7 400 2 400 Resistivity u/c 500 (W m) Elevation (m) 300 c) Depth (m) 0 200

100 Drillholes 100 Faults 200 Pit outline 4850N Base of oxidation 0 368 300

135 Figure A4.9 Geology of the Golden Cross epithermal gold deposit. 400 49 (a) Geological map and (b) cross-section along traverse 4850N.

18 Redrawn, with additions, and with permission, from Simpson 500 et al. (2001). Crater 7

600 2 Ω Resistivity resistivity of about 10 m. The unaltered host sequence 700 ( W m) and the younger andesite have resistivities of about Diatreme 100 Ω m. Resistivity/IP surveys detected the conductive 10000 alteration zone, but did not delineate the high-resistivity 2500 9500 2000 mineralised area. 1500 9000 Easting (m) 1000 8500 Northing (m) 500 CSAMT data were collected with electric receiver dipoles spaced at 50 m intervals. Cagniard resistivity Figure A4.8 Regis diamondiferous kimberlite pipe. (a) Surface outline and phase pseudosections are shown in Figs. A4.10a of the pipe and the location of the AMT stations. (b) Resistivity model derived from 3D inversion along section 9300E, showing the and b. The data show strong contact responses at conductivity distribution within the pipe, and (c) the full 3D stations 3050 E and 3350 E bounding a distinct zone of resistivity model. Based on diagrams in La Terra and Menezes (2012). disruption in the electrical properties of the whole A4.7 Examples of magnetotelluric data 13 section. This zone is coincident with the mineralised mineralised zone. Although the argillic and propylitic zone. The area east of station 3350 E shows very little zones are comparatively conductive, the response lateral change in electrical properties. Inversion of these appears to be controlled by the resistive siliciﬁed zone. data produced the resistivity model cross-section shown Known faults correspond with lateral changes in resist- in Fig. A4.10c. It shows a conductive feature dipping ivity and phase, but the 1D nature of the pseudosections shallowly to the east which appears to be associated with creates the appearance of vertical contrasts in electrical the unconformity at the base of the younger andesite properties. To some extent this is addressed in the unit. A resistive zone in this feature coincides with the inverted data, but by no means entirely.

W E a) Location (m) 3000 E 3400 E 3800 E 4200 E 4096

40 100 2048 100

1 10 25

1024 6 25 40

100

512 6

256 63

128 Low

Frequency (Hz) 64 Onset of transition zone responses 25 16 32

398 Transition zone ‘notch’ 16 Low 6 10 6 8 b) Location (m) 3000 E 3400 E 3800 E 4200 E 4096 1000 2048 1100

1024 700 900 1150 512

1000 850 Low

256 800

0 128 Low 105

60 Frequency (Hz) 64 0 1050

Onset of transition zone responses 10 32 00

0 16

8 c) Location (m) 3000 E 3400 E 3800 E 4200 E

Surface 400 25 300 Low 6 200 Figure A4.10 Pseudosections of CSAMT data from

100 10 16 traverse 4850N, Golden Cross epithermal gold deposit. (a) 100 Apparent (Cagniard) resistivity (Ω m), (b) phase

25 10 63 40

1 Elevation (m) 0 100 0 difference (mrad) and (c) resistivity cross-section

158

-100 158 obtained by inverse modelling the data in (a) and (b). 251 251 8 Selected components of the local geology are shown. See -200 39 text for details. CSAMT data redrawn, with permission, Argillic & propylitic alterationSilicic alteration Quartz vein Fault from Zonge and Hughes (1991). 14 Magnetotelluric electromagnetic methods

Figure A4.10 also illustrates how transition zone A4.8.1.1 ZTEM responses appear in pseudosection displays. The rapid The ZTEM system (an acronym for Z-axis tipper electro- variations of the transition zone notch produce sub- magnetic) (Legault et al., 2012) is an airborne AFMAG horizontal contours in the phase-difference data coincident system comprising a towed-bird receiver coil of 7.4 m in with a sub-horizontal zone of low apparent resistivities. To diameter measuring the vertical component of the mag- the east, transition zone responses begin at 32 Hz, but to netic field (H ) at six frequencies in the range 25–600 Hz. the west they appear at 64 Hz. This is due to the overall Z It is towed about 80 m below a helicopter or fixed-wing difference in resistivities between the two regions. aircraft at a nominal terrain clearance of 50–100 m. Orien- tation of the coil is monitored and corrections applied to A4.8 Natural source airborne EM systems the data to compensate for off-vertical orientation errors. Two perpendicular receiver coils of 3.2 m in size are Airborne methods using natural source fields are confined located on the ground near the survey area as a base station to measuring only the magnetic field (H-field). The ampli- to monitor variations in the perpendicular horizontal com- tude, phase and directional relationships between the vari- ponents, H and H , of the field. For each frequency, the ous components of the H-field depend on the resistivity X Y vertical magnetic field is related to the horizontal compon- distribution of the subsurface and produce an anomalous fi ents as follows: vertical eld (HZ). The measurements are made in the audio-frequency range so the method is known as audio- HZ ¼ TZXHX+TZYHY ðA4:6Þ frequency magnetics (AFMAG).

TZX is the portion of HX contributing to HZ, and is the

along-line component; and TZY is that due to HY, and is the A4.8.1 AFMAG fi across-line component. TZX and TZY form a set of coef - AFMAG is a passive frequency domain method that offers cients known as the ‘tipper’ (T) because, they describe the several advantages over artificial source AEM methods; it has amount that the horizontal field is ‘tipped’ into the vertical

great depth of penetration, and the intensity and direction of to form HZ. The tipper T is resolved using the base station the planar uniform source field are constant throughout the measurements based on the assumption that the horizontal area and energise all conductors uniformly. Within the fre- field is relatively homogenous throughout the survey area. quency range of the measurements, the depth of penetration Each coefficient is phase-shifted with respect to its hori- depends only on skin depth (see Section 5.2.3.1), i.e. only on zontal component of the field. From these the magnitude the resistivity of the rocks and not the geometry of the and phase of T is obtained, for each frequency. transmitter–receiver configuration as with all other electrical For a 1D subsurface (see One-dimensional model in

and EM methods. The method offers greater sensitivity to Section 2.11.1.3) HZ is zero. It is non-zero where the deep large conductors in highly resistive terrains and pro- horizontal ﬁeld is distorted by variations in the subsurface vides greater resolution of resistive targets than conventional conductivity. Phase relationships of T are interpreted in AEM. It is also useful for general geological mapping. terms of subsurface conductivity. These are plotted for Surveys can be conducted higher than is normal for AEM each frequency, and the data can be inversion modelled surveys, at 200 m above the terrain, and variations in survey to produce resistivity models of the subsurface (Holtham height have little effect on the measured response. and Oldenburg, 2010).

...... Summary • Natural source methods use electromagnetic signals originating from the magnetosphere and distant thunderstorm activity as the signal source, and these induce electric currents in the subsurface. • As this is a passive geophysical method, only receivers are required to make measurements. These measure the strength and direction of the electric and magnetic ﬁelds associated with the telluric currents, from which the subsurface resistivity is obtained. References 15

• CSAMT is similar in principle to natural source methods but uses the signal transmitted from a very large electric dipole located at distance from the survey area. • Interpretation of CSAMT data is strongly dependent on signal frequency and proximity of the measurement to the transmitter dipole. Measurements may be in the far-field or the near-field of the transmitter dipole, the latter complicating the interpretation. • Depth of investigation in MT/AMT and CSAMT depends only on frequency and the resistivity of the subsurface. The very low frequency of the natural fields means that the depth of investigation in MT/AMT can extend many kilometres. • Natural source EM methods and CSAMT find application for low-cost exploration for large and deep electrical targets.

...... Review questions 1. Compare and contrast natural source methods, CSAMT, and conventional resistivity and EM methods. 2. Describe how to recognise near-field and far-field CSAMT measurements and their significance in terms of depth of investigation and resolution. How do they influence survey design?