89

BMR Journal ofAustralian Geology & Geophysics. 2 (1977) 89-95

Absolute electromagnetic scale modelling

and its use in interpretation of TEM response*
B. R. Spies
Scale model studies are often used to determine the distribution of secondary magnetic fields when the
nmnerlcal approach Is unreasonably difficult. Reduction in size (L) is compensated by an increase in con
ductlvity (0) or frequency (f), or both. In most cases the results are expressed in a way whereby they are
independent of the absolute dimensions of the model and the general modelling relation
o,..fL' = constant, Is valid.

However in some applications such as transient electromagnetic (TEM) modelling it Is necessary to

determine the power level ofthe fullscale system. An expression relating'the voltage levels of the model and
the fullscale system Is:
Vm/V = Lm TILT m

where the subscript m represents the model system. ThIs expression must he used in conjunction with the
general modelling relation in order to compensate for changes in the modelling parameters.
It is possible to scale time, as weD as dimensions and conductivity and, using the above relationships,
many geological situations may be simulated with a single model. By varying either one of the parameters
conductivity or dimensions while keeping the other constant, the same transient decay curve can be
obtained by transposing the individual curves (plotted on a log. log scale) along the time and response axes.
An example illustrates that by modelling field cases absolute quantities such as conductivity can be esti
mated. Modelling of the Woodlawn orebody gave bulk conductivities of 5 S/m and 20 S/m.
The analysis of TEM data using early and late time responses must be done with care. The response
typical of a late time may be similar to an early time response if the electrical or dimensional properties of
the conductor are varied. More correctly one should analyse responses in terms of large and smaU values of
the parameter o,..V/T.

Introduction
Scale model studies have been used for over 30 years to
determine the response of electromagnetic exploration
methods to simulated geological conditions and to assist in
the design of field systems_
With the advent of computer facilities, numerical
modelling methods have become popular; these include the
finite element, finite difference, and integral equation
methods (Praus, 1975). Numerical modelling methods are
too difficult or too expensive in some applications, and sC,ale
model studies must be used_ This is particularly true in the
case of the transient electromagnetic (TEM) method
because mathematical solutions are known only for simple
models.

electromagnetic scale modelling was given by Sinclair in

1948. A good overall description of scale modelling is given
in a paper by Frischknecht (1971). Briefly, scale modelling
involves simulating geological conductors by smallscale
models of a size that can be handled in a laboratory. The
reduction in size is accompanied by an increase in con
ductivity, or frequency, or both.
Sinclair distinguished between two types of models,
geometric and absolute. A geometric model is one in which
only the geometric relationships of the electric alJj
magnetic fields are simulated. An absolute model is one in
which all scalar and vector components are simulated.
In most electromagnetic systems commonly used in geo
physical exploration the parameters measured are

The Transient Electromagnetic Method

The transient electromagnetic method, developed in the
USSR in the early 1960's, is described by Velikin & Bulgakov
(1967). Briefly, the method involves pulsing a square wave of
current into an ungrounded loop (see Fig. 1)_ The changing
current in the loop generates a timedependent magnetic
field which causes eddy currents to flow in subsurface
conductors_ The secondary electromagnetic fields caused by
changes in these eddy currents are detected in the same
loop, and measured at various sample times ranging from
0.5 to tens of milliseconds. The results are displayed as
curves of the voltage decay as a function of time, or
transient decay curves, as shown in the bottom of Figure 1.
Examples of the use of the TEM method in Australia are
given by Spies (1976a).

PRI MARY WAVEFORM

TRANSIENT DECAY CURVE

Electromagnetic scale modelling

The first complete description of the basic theory of

Paper presented at 25th International Geological OIngress,

Sydney, August 1976.

TIME (ms)
Figure 1.

TEM waveform

G29-230A

90

B. R. SPIES

dimensionless (see Table 1), and model results can be

expressed so that they are independent of the absolute
dimensions of the model. Since only the geometric
configuration of the lines of flux needs to be simulated,
models of these systems are geometric models.
In the TEM meth9d (Table 1) the quantity measured is a
voltage induced in the loop; this voltage is normalized by
dividing by the primary current. The quantity measured is
not dimensionless and in fact will change if the con
ventional scale modelling relation (given in the next section)
is used. This would also apply to the magnetotelluric
method.

Electromagnetic induction response

Geometric scale modelling

where'L is the loop radius.

This expression can be divided into two parts. The
dimensionless quantity, o/AL'IT, is the same parameter as
used in the general modelling relation, and the expression
-2L/AVn i T is proportional to LIT. If we wish to construct a
model of a homogeneous half-space by varying linear dimen
sions and time, we can see that by keeping O/AL' I Tconstant
the magnitude of the response, e(t)/ I, will vary as we scale
LandT.

The response of induction methods is determined by

"induction numbers" or "conductivity parameters" which
have the same form as the dimensionless modelling
parameter o/Aft'. For example, the response of a sphere in a
uniform alternating field is a function of the parameter
o/Awa' , where a is the radius of the sphere (Wait, 1953).
Similarly, the theoretical TEM response of a homo
geneous half-space, given by Lee & Lewis (1974) is

The relations for geometric scale modelling are:

Frequency domain
o,/A, f, L',= Om/AmfmL' m'
Time domain

(1)

o,/A,L', / T, = Om/AmL'm I Tm

where

conductivity,
magnetic permeability, usually assumed to
be/A .
f
frequency,
L
linear dimension,
T
time.
In these expressions the subscript f refers to the full scale
or field system and the subscript m refers to the model.
Simply stated, these expressions state that an equivalent
geometric response will be obtained if the product of con
ductivity, permeability, frequency (or l / time), and linear
dimension squared is kept constant.
0

/A

Absolute scale modelling

The relationship for absolute scale modelling is
Tm
Lm T,

~ = ..!:-L
Vm

(2)

HORIZONTALLOOP
SLINGRAM

F (O~L')

... (3)

Exatnples of absolute scale modelling

The results in this section were obtained using the scale
model facility at Macquarie University, Sydney, which
consists of an interactive mini-computer system.
The mini-computer, shown in the centre of Figure 2, is a
32K word Interdata 70, and is used for controlling output
waveforms and timing. It is also used for data acquisition
and subsequent processing.
.
The required waveform and sampling details are typed on
the teletype shown in the figure, and are transferred by the
. computer to a small multi-turn loop placed on the model.
The resultant transient decay curve can be studied on a
eRO, and then sampled, amplified and digitized, and
printed on the teletype. A detailed description of this facility
is given by Spies (in prep.).

where V is the measured voltage and the other symbols are

as defined earlier (Spies, 1976b). This expression states that
the measured voltage will change in a manner proportional
to the dimension-scaling factor and inversely proportional
to the time-scaling factor.
.
When modelling systems such as TEM in which the
measured quantity is not dimensionless, it is necessary to
use the amplitude correction factor (2) in conjunction with
the scale modelling relation (1). Sinclair called this type of
model an absolute model.
SYSTEM

-2 LJA: n

e(t)

Half-space model
Figure 3 shows a model of a homogeneous half-space con
structed out of typemetal, which is an alloy of lead,
antimony and tin. The model is only 8 cm thick, but
because of its high conductivity is large enough to simulate
a homogeneous half-space over a fairl y wide time range. To
attempt this with a tank of salt water would require a tank
15 m deep, which would be prohibitively large.
Table 2 shows the modelling parameters for the type
metal model. The model dimensions are 9520 tilDes smaller
UNITS

QUANTITY MEASURED

Ratio of coupling in air to coupling on earth

Dimensionless

Amplitude ratio and phase difference offield in two loops

Dimensionless

Dip of major axis of ellipse of ~Iarization

Dimensionless

Azimuth and dip of major axis of ellipse of polarization

Dimensionless

VLF

Dip and ratio of axes of ellipse of polarization

Dimensionless

TEM

Rate of decay of secondary field normalized by primary current

TURAM
DIP ANGLE
VERTICAL LOOP
AFMAG

MAGNETOTELLURIC

Electric field intensity normalized by magnetic field intensity

Table 1.

Parameters measured in common electromagnetic systems.

(voltamp- ')
(volt metre-' tesla- ' )

ABSOLUTE ELECTROMAGNETIC SCALE MODELLING

Figure 2.

91

Scale model facility, Macquarie University, Sydney.

than the field dimensions, and the simulated half-space

conductivity varies as the time-scaling factor. If we wish to
simulate a conductivity of 0.1 Si m for instance, the general
modelling relation shows that we must scale the time by a
factor of 2.3. The condition relating to amplitude however
requires that a correction of 4.1 x 103 be made to the
amplitude. The same model can be used to simulate half
spaces of different conductivity simply by varying the time
scaling factor.
FIELD

MODEL

HALFSPACE
CONDUCTIVITY

0.1 Si m

3.9 x 10' Si m

Loop radius

100 m

1.05 cm

Time

1 ms

0.43 ms

I'L'/T

10'

10'

Amplitude V
correction Vm

=~ . Tm
Lm

=4.1xl0J

Table 2. Modelling parameters used to simulate homogeneous

[Link].

Figure 4 shows the readings obtained over the model ,

together with a curve of the theoretical response. The
modelled readings have had the appropriate amplitude
correction applied to them. The agreement between
modelled and theoretical response is a good check on the
modelling relations.

Cylinder model
We now go to a more complex example for which
numerical results are not available. This example consists of

Figure 3.

Model of a homogeneous [Link] constructed out of

typemetal.

cylinders of varying conductivity and size. The cylinders

were constructed of copper, aluminium, brass, and type
metal; which gave a conductivity range of 5 x 10 6 to 5.8 X 10 7
S/ m. The cylinders were either of 2 or 1.4 cm diameter, and
the ratios between loop size, cylinder size, and depth in the
two models were kept constant. The transient decay curves
are shown1n Figure 5.
We now show that by using the modelling relation all the
models can in fact simulate one case, as long as the para
meter ol-lL' / T is kept constant. If we use different conduc
tivities we scale time directly as conductivity. If we use dif
ferent values of L (linear dimensions), we scale time as the
square root of the dimension-scaling factor. We then obtain
the results shown in Figure 6. For the cases in which L is

92

B. R. SPIES

_1

10

LOOP '
GI

.....

~ 1b ~

1-

~
~

'.,

-,...

CYLINDER IN AIR

'1

---

MODELLED READINGS

~<

"

0.1-

UJ

V>

THEORETICAL CURVE
{f

Z
0
D..

= 3 . 9 I 10 6 S m
= 1. 05 em

V>

\ ....

00'

DISPLACEMENT OF CURVES
FROM CURVE

0.001

'0

0'

TIME Ims)

155 / B1 - 15I A

[Link]

0.1

Comparison of modelled and theoretical TEM results

from a homogeneous half-space.

MATERIAL
COPPER I A)
CO PPER IB)
ALUMINIUM IAi
TYPEMETAL
COPPER IB)
ALUM INIUM IB)
BRASS
TYPEMETAL

'0

..

0.0 0 1 + - - - - - - - - - - , - - - - - - - - - - - - ,

CURVE No.

;,

0.01-

UJ

0::

Figure 4.

,.LL/l

1---17cm~

0'

e ~L/4

CONDUCTIVITY

x 10 7
x 10 7
x 10 7
x la '
x 10 7
x 10 7
1.2 x 10 7
5.0 x la'

TIME (ms)

5
155/ 87 -1 55A

Figure 6.

Composite curve obtained by applying the modelling

relation.

Figure 7.

Scale model of Woodlawn orebody.

I Si m)

5.8
3.6
2.3
5.0
3.6
1.8

0.5

}~

}s

~
z

0'

e;
w

'"
0.01

TIME I ms)

Figure S.

155/B1- 150A

Transient decay curves over cylinders of different size

and conductivity.

kept constant while a is varied, only time is scaled and the

amplitude correction factor VI Vm will be proportional to
T mi T. Similarly if a is kept constant and L varied , T IT mwill
be proportional to the square of LlLm , and VI Vm will be
proportional to the square root of Tm IT. If the curves are
plotted on a log-log scale, the effect of scaling is a simple
transposition of the curves along the time and response
axes.
Thus we see that the transient decay curves were not
unique. The same curve can come from a whole suite of
models simply by varying the conductivity or geometry,
provided that the modelling relation is adhered to.

Woodlawn copper-lead-zinc-deposit model

This example is a model study of a field case, the base
metal orebody at Woodlawn, near Tarago, NSW. The ore
body is a concordant lens of bedded high grade copper
lead-zinc sulphides, 20 m thick, which dips at about 45.
The near-surface oxidized zone is about 12 m thick, consists
mainly offerruginous clay, and has a sharp contact with the
sulphides. The shape of the body is known accurately from
more than 100 drill holes. A detailed report of the geology
of the ore body is given by Malone et at. (1975). Figure 7
shows a scale model of the massive sulphide orebody con-

ABSOLUTE ELECTROMAGNETIC SCALE MODELLING

WOODLAWN

93

MODEL
SIMULATED
CONDUCTIVITY
(S/m)

DEPTH TO TOP
(m)

0.23
0.25

1000

0. 33

14

~""
.......
z

100

<>

...e;
a:

10

(Geology 01111' Malone lit 01. 1975)

50
,

o
I

50

100

150 METRES
10

0.1

MIDDLE TO UPPER SILURIAN

I~ ~ ~ J DOLERITE

SHALE, FINE-GRAINED
ACID VOLCANICS

~
~---

BLACK SHALE

~
~

COARSE -GRAINED
ACID VOLCANICS

Figure 8.

TIME (ms)
_

GOSSAN

1,l1
~

POSITION
OF LOOP 1

Figure 9.

155/ B7- 153A

Model results for various depths, Woodlawnorebody.

10000

45m lOOP
POSITION (I)

TSF TIME SCALING FACTOR

155/B7-156A

Surface geology and locality plan, Woodlawn orebody.

structed out of graphite. Dimensions are scaled down 670

times, so that the model is 31 cm long. Time can be scaled
to allow a good fit between field and model curves, and esti
mates of conductivity can then be made. Figure 8 shows a
plan of the Woodlawn deposit. Both BMR and CSIRO have
carried out TEM test surveys at Woodlawn. Field curves
shown in Figures 10 and 11 were obtained by CSIRO at
positions 1 and 2 of Figure 8, using a sophisticated TEM
system (Buselli, 1974).
Initially in modelling the Woodlawn deposit it was
assumed the near-surface oxidized zone was resistive, but it
was found that the field and model results were dissimilar
for the early parts of the decay curves, in which near-surface
effects are dominant.
In order to investigate the effect of a conductive near
surface zone, measurements were made with the model
buried at various depths. Figure 9 shows transient decay
curves at position 2 at depths of burial of2, 6 and 14 m. The
curves have been transposed on the time axis so that they
coincide at late times. Because of the scale modelling
relations, this also changes the simulated conductivity ofthe
body. The best curve fit is for curve 1 (depth of2 m and con
ductivity of 0.23 S/m) and so at Woodlawn the near-surfacl;
oxidized zone must be moderately conductive-and not
resistive as suggested by earlier work.
Figures 10 and 11 show the model study and field curves
from positions 1 and 2 respectively. Curve I, Figure 10, is
the original model study curve, without scaling time. In this
loop position, by using the modelling relations, we have a
conductivity of 0.23 S/m. However, we can see there is .no

O"F' MODELLED FIELD

CONDUCTIVITY

1000

.:r.
\

LM

\
\

100

MODEL

CURVES

,
,

...

FIELD

CURVE
\

,,
,,
,,

(J)

~
(J)

0::

\
\
\
\

'-""'

...

DIMENSION SCALING. 670

FACTOR

10

\
\

\
\
\
\

\
\
\
\
\
\
\

1 \\
TSF'1
O"F 023 S/m

TSF'IO
O"F' 2 3 S/m

\
\

\
\

,,
2\,

,,

,
\

TSF 203
O"F ' 47 S/m

3\

OI+-------.--------,r---~-

01

TIME (mo)

10

_ _,

100

155/e7-154A

Figure 10. Comparison of model and field results, Woodlawn ore

body, position 1.

B. R. SPIES

94

'"
@--i

10000

LOOP

6:=200m' 1,

10000

45m lOOP

\
1000

POSITION (2)

...

CURVES \
TSF
(fF

1000

MODEL \

~c

SPHERE

1\
_'00

.>

z
'"

...

C)

2\

...a:
'"

1
\
0.23 S/m I

10

TSF

100

\
\
\
\
\

= 10

100~------~-------+2--------~----~

TIME

ITf= 2.3S m

(ms)

155 / B7 - 14 9'

Figure 12. Transient decay curves over spheres of different con

ductivity.

TSF

= llO

ITF= 25.3 S m\

0.1

"1"------r-------.---------.!0.1

TIME

10
(ms)

100
15 5/ 81 -152A

Figure 11. Comparison of model and field resnIts, Woodlawn

orebody, position 2.

resemblance between the model and field curves. By using

the modelling relations we can scale time and conductivity
and, on a log-log graph, the curve can be transposed along
the time axis and, also, vertically downwards to account for
the amplitude correction. Curve 2 has a time-scaling factor
of 10, and the simulated conductivity is 2.3 S/m. Curve 3
has a time-scaling factor of 20.3, and the simulated conduc
tivity is 4.7 S/ m. There is a good fit between curve 3 and the
field curve, and we can say that for the area under investiga
tion under the loop in position 1 the bulk conductivity of the
orebody is about 5 S/ m.
Figure 11 shows the results applying the same process in
loop position 2. Curve 1 is the model study curve without
scaling time. This curve can be moved across as already des
cribed to curve 3 which has a reasonable fit to the field
curve at late times but not at early times. Since the depth of
investigation is related to sample time this means that the
model is correct for deeper parts of the orebody, but should
have a greater conductivity or be thicker for the more near
surface parts. The overall b~lk conductivity of the orebody
at loop position 2 must be very high, at least 25 S/ m.

Application to rules of
interpreting TEM results
For many interpretational procedures rules of thumb are
used, and we make use of what are called 'early' and 'late'
time asymptotes. For example the response of a sphere at late
times is exponential, with a time constant equal to n'l0folb',
where 0, /-l and b are the conductivity, permeability and
radius of the sphere, respectively (Nabighian, 1970). Thus
by measuring the slope of the transient decay curve at late

times we can estimate the conductivity or size of the sphere.

But what is a 'late time'?
Figure 12 shows two transient decay curves over identical
spheres of different conductivity. The curyes are plotted on
a log-linear scale so that at late times the asymptotes are
straight lines. For the 2 Si m sphere the late time asymptote
occurs near 1.5 ms, whereas for the higher conductivity
5 Sim sphere the asymptote does not occur until 3.7 ms .
The asymptotes occur at similar values of the parameter
o/-lL'/T, and the response typical of a late time for a body
may be similar to an earlier time response ifthe electrical or
dimensional properties are varied.
.
It is obviously important to measure over as wide a time
range as possible. When analysing responses from early and
late times we must remember that the actual time that we
can measure the asymptote will depend on many factors,
e.g. the size and conductivity of the body, its depth of
burial, and the loop size. More correctly we should analyse
responses in terms of large and small values of the para
meter o/-lL'/ T rather than in terms of early and late sample
times.

Discussion and conclusions

Several r~ults emerge from this study.
The examples with the cylinders and sphere showed that
transient de~ay curves will appear to change in shape as we
change either the dimensions of the body or loop, or con
ductivity of the body. An 'early time' in gne situation may
be equivalent to a 'late time' in another, so we must
measure over a very wide time range in org:er to get as much
information as possible on the siIbsutface conductivity
distribution:
.
.The ability to scale time opens up a new range of possi
bilities in TEM scale modelling. We have seen that we can
estimate bulk parameters such as conductivity or
dimensions as long as one is known, a procedure often used
with frequency domain methods.
Once the physical and electrical parameters of mineraliz
ation in a particular area are specified, it is then possible to

ABSOLUTE ELECTROMAGNETIC SCALE MODELLING

design an electrical or electromagnetic survey which will

provide optimum detectability of that style of mineraliz
ation. When applied to geophysical exploration, modelling
enables one to assign quantitative values to such items as
depth of investigation of a method in a particular geological
environment. Once a target has been adequately defined,
the probability of detecting it is greatly increased.
The Woodlawn results show that such orebodies have a
very high bulk electrical conductivity, and therefore electro
magnetic methods have a great potential in their detection.
The major problem to date is' in instrumentation, especially
in having a wide enough frequency range to obtain informa
tion on the various conductors present (Ward et al., 1974).
For similar reasons TEM instruments should also cover a
wide range of sample times; 3 or 4 decades of time would be
preferable. TEM instruments presently on the market
typically have a I.S decade time range.

References
BUSELLI , G., 1974-Multichannel tran sient electromagnetic
measurements near Cloncurry. CSIRO Min erals R esearch
Laboratory Investigations Report 103.
FRISCHKNECHT, F.e., 1971-Electromagnetic scale modelling. In
J. R WAIT (Editor), ELECTROMAGNETIC PROBING IN GEOPHYSICS
265-320. Boulder. Golem Press.
LEE, T., & LEWIS, R., 1974-Transient EM response of a large loop
on a layered ground. Geophysical Prospecting. 22,430-44.

95

MALONE, E. J., OLGERS, F., CUCCHI, F. G .. NI CHOLAS , T. , &

M c KAY, W. J., 1975--Woodlawn copper-lead-zinc deposit. In
C. L. KNIGHT, (Editor), ECONOMIC GEOLOGY OF AUSTRALIA AND
PAPUA NEW GUINEA. Australasian Institut e of Mining and Metal
lurgy. M elbourne_
NABIGHIAN , M. N., 1970--Quasi-static transient response of a con
ducting permeable sphere in a dipolar field. Geophysics. 35,303309.
PRAUS , 0. , 1975--Numerical and analogue modelling of induction
effects in laterally non-uniform conductors. Physics of the Earth
and Planetary Interiors. 10,262-70.
SINCLAIR, G., 1948-Theory of models of electromagnetic systems.
Proceedings Institute Radio Engineers. 36, 1364-70.
SPIES. B. R., 1976a-The transient electromagnetic method in Aus
tralia. BMR Journal of Australian Geology & Geophysics. 1, 2332.
SPIES , B. R., 1976b--The derivation of absolute units in electro
magnetic scale modelling. Geophysics. 41,1042-47.
SPIES, B. R-An application of a mini-computer in scale model
studies of a transient electromagnetic prospecting system (in
prep.).
VELIKIN , A. B., & BULGAKOV, Yu I. , 1967-Transient method of
electrical prospecting (one-loop version). UNO International
Seminar of geophysical m ethods of prospecting for ore minerals_
Ministry of Geology of the USSR . Moscow.
WAIT, J. R , 1953--A conducting permeable sphere in the presence
of a coil carrying an oscillating current. Canadian Journal of
Physics, 31,670-78.
WARD, S. H., PRIDMORE, D. F., RIJo , L., & GLENN, W. E .,
1974-Multispectral electromagnetic exploration for sulphides.
Geophysics, 39, 666-82.