Exploration Geophysics - Volume 19, Issue 3, 1988

A magnetotelluric (MT) system with sampling frequencies in decade steps from 1 kHz to .01 Hz was operated at a site in the Moreton Basin, near Peaks Crossing, Queensland. The data were analysed by both the conventional rotated impedance and the Spitz eigenvalue methods, then resistivity-depth profiles were calculated using a Molochnov/Bostick inversion technique. The magnitude of the resistivity information derived from the two methods was found to differ substantially, presumably due to the three-dimensional nature of some of the underlying geological structures. The finer structural features were similar between methods, though the depth of these features varied due to the differing apparent resistivities. The variation of the principle directions in the Spitz method was much smoother. Comparison between the resulting profiles and geological and geophysical information in the area showed reasonable agreement with expected geological structures.

The effective use of transient electromagnetic fields for mineral exploration depends on a proper understanding of how these fields propagate in and are affected by different geological materials and structures. The analytical methods for investigating these effects form the subject of the review. Some of the basic phenomena which can be explained without need for extensive numerical modelling are discussed. The following examples are among them: The rate of decay of the measured field, which is diagnostic of the general geometry of the conductivity distribution in the ground, can be forecast with simple analytical models. Transient decay curves of a significantly different character can be caused by geological materials whose conductivity or magnetic permeability varies with frequency — for an earth in which the conductivity is frequency-dependent, changes in sign of the transient may even be observed. A magnetic permeability that varies with frequency is sufficient to explain all the anomalous observations that have been made in surveys carried out over areas where superparamagnetic minerals are present on the Earth’s surface. Analytical methods can be used to show that the response of a more-conducting target in a less-conducting host can be represented as the sum of independent host and target responses, plus an interaction term; this result helps to justify the use of anomaly-separation methods to extract target responses. Inversion studies using analytical models show where supplementary data is necessary to define adequately the numerical values of the parameters of the model.

Electromagnetic transients measured with coincident transmitter and receiver loops are usually positive at all delay times. The response occasionally switches to negative at late times; a possible cause of which is an induced electrical polarization. Unfortunately, it is not possible to explain the negative transients with a half-space or a single bedrock conductor unless the polarizability of the model is unrealistically large. A negative response can be obtained with a realistic polarizability if the coincident-loop response of a body is large and positive at early time and decreases very rapidly. The current associated with the positive response polarizes the body: the larger the positive response, the greater the polarization. However, the positive response must also decay away rapidly, so that the smaller response associated with the polarization will be observed as a negative transient at late time. A body with such a response is a conductive overburden. Because the large early-time currents induced in an overburden flow close to the transmitter loop, the most significant polarization will occur here. When this polarization is large, the sign of the vertical-field response will be reversed when the receiver is inside, or coincident with, the transmitter loop. If the polarization is small, the distortion of the response near the wire is similar to what is termed ‘the loop effect’. These abnormal TEM responses can be modelled with a weakly polarizable overburden, suggesting that polarization effects are a plausible explanation.

Geomagnetic reference fields and corrections to aeromagnetic data were the subjects of a discussion workshop held in Canberra on 1st December 1987. The proceedings of the workshop that are most relevant to aeromagnetic exploration in Australia are summarized in this article and supplemented with our additional comments. An accompanying paper (Barton, this issue) provides a more detailed exposition on the derivation and applications of global and regional geomagnetic reference fields.

Initial discussions centred on recent developments in global and regional (Australian) modelling of the geomagnetic field, and applications and limitations on the use of available models. Attention then turned to the aeromagnetic ‘diurnal’ correction problem, the use of base-station and tie-line crossovers for making ‘diurnal’ corrections, continental and oceanic induction effects, the relative merits of magnetic gradiometer surveys, and ramifications of the modern trend towards the use of rapid sampling (≤ 10 second) and high sensitivity (subnanotesla) survey techniques. Investigations into the spatial variability of magnetic field fluctuations using magnetometer arrays are needed both on a regional scale and on a local scale. The proposed Australia-wide array of geomagnetic stations (AWAGS) experiment is an example of the former. New methods of presenting the results of such investigations are also needed, for example in the form of a set of ‘base-station reliability’ maps of the region.

A microcomputer program is described that uses the Hilbert transform to transform real data to imaginary and vice-versa in the frequency domain. This is achieved via specific integrals embodied in the Kramers-Kronig relationships. Cubic splines are applied to the measured data to interpolate 100 points per logarithmic decade. The Hilbert transform is computed at each frequency by applying the finite difference form of the algorithm over each logarithmic decade and then summing results for all decades. Two estimates of error are provided. One is a measure of the percentage error for each generated point relative to the measured data; the second is the chi-squared parameter.

An interactive computer program to model frequency domain complex resistivity data has been developed using the Cole-Cole model. This simple, well established model, based on the mathematical response of an electrical circuit, is effective in modelling the electrical impedance of a mineralized rock, particularly sulphide mineralization. Model parameters: Resistivity R; chargeability, m; frequency dependence, c; and time constant τ, are estimated by forward modelling. An inversion technique, ridge regression, is used to obtain the parameters corresponding to the “best fit” curve. The program, written in Turbo Basic, is designed to run on any IBM PC/XT/AT or close compatible. It incorporates a well structured approach with ease of operation, employing simple to use menus and descriptive error messages. Input is via the keyboard or through a file system. Output is tabular and graphical, showing the measured and predicted data, and a least squares error estimate.