Exploration Geophysics - Volume 29, Issue 1-2, 1998

Airborne electromagnetic geophysics is based on analysis of the interaction of an electromagnetic field with the geoelectric properties of the earth. Inversion, or inverse modelling, of airborne electromagnetic (AEM) data refers to a particular mathematical methodology for solving the AEM inverse problem, that is, deducing the earth’s geoelectric properties from observed electromagnetic interactions. This is a difficult problem for several reasons. First, like most geophysical inverse problems, the AEM inverse problem with a finite number of noisy data is ill-posed, and consequently, the geoelectric properties of the earth cannot be uniquely determined. To generate a unique solution a priori information must be added to the inverse problem: a procedure referred to as regularisation. Second, since the geoelectric properties of the earth and the observed AEM data are not linearly related the inverse problem is nonlinear and requires solution by an iterative method. Third, the forward problem of calculating the response from a given geoelectric earth model, which is an essential part of the inverse problem, is itself a difficult and time consuming problem for 2.5D or 3D models. Fourth, AEM geophysics is characterised by enormous quantities of data. These difficulties and how they can be addressed are the focus of this paper. Particular emphasis is placed on the nonuniqueness of the AEM inverse problem and how it can be resolved through regularisation using a priori information. The applicability of 1D inversion in multi-dimensional environments and the advantages of multi-dimensional inversion are demonstrated, as is the potential value of joint inversion of AEM data and other geophysical data.

Because of the non-uniqueness of the EM inversion problem, one must constrain the inversion procedure in some way to achieve a geologically valid model. All constraints are regarded as subjective except when they agree with the prejudices of the person using the inversion. Conventionally, constraints have included: inverting to models with a small number of parameters; forcing conformation with a priori geological information; or requiring smooth or minimal spatial derivatives with respect to conductivity (Occam inversion).

We prefer to use the constraint philosophy of the damped eigenparameter method (also called the Jupp-Vozoff algorithm). After incorporating a priori geological knowledge in the starting model, this process constrains change to individual model parameters on the basis of the strength of their effect on the data when grouped into eigenparameters. The initial model can thus be regarded as a soft constraint. In this paper, we extend this inversion philosophy to the case where the number of unknown parameters exceeds the number of data points; i.e., an under-determined problem. Inverting to both undetermined and over-determined layered earth models, we have tested this method on two sets of synthetic AEM data: one generated from a 1D model, the other from a 2.5D model.

Since 1993 we have been using a program developed by the BGR for the automatic inversion of frequency-domain helicopter electromagnetic (HEM) data into the parameters of a layered half-space. This Marquardt-type inversion procedure uses sounding curves for generating suitable starting models. To better understand the results, model calculations have been carried out using 1-D and 3-D resistivity models to generate synthetic data which were inverted using homogeneous or layered half-space models. Inversion results demonstrate that reliable models can be obtained if sufficient and accurate data are available. In particular, we have shown that the 1-D inversion program provides very useful results even in the presence of lateral changes in resistivity. Several examples of inversion results based on theoretical data and survey data are discussed.

A technique for minimising the calibration errors in helicopter electromagnetic data (HEM) is developed. The method is based on resistivity-depth information obtained from auxiliary geophysical data at locations scattered over the survey area. The technique compares measured HEM values with the calculated responses from resistivity-depth models at the auxiliary data sites. Correction factors which minimise the differences between the calculated and measured values are obtained by a least-squares method. The correction factor incorporates amplitude, phase and bias adjustments. The analysis shows that only the amplitude and phase corrections are stable enough to be of use. After the correction factors are applied to the data, resistivity-depth models for the entire survey area are determined by inversion. The correction procedure results in better agreement between measured and inverted bird elevation, as well as a smaller inversion misfit error.

A concentric, multiple transmitter-receiver EM system has many advantages for the detection and characterisation of small, confined conductors. The abundant geometrical information provided by such a configuration can greatly facilitate the estimation of the shape and location of a discrete conductor. This configuration also provides extra information, which can assist in stripping out onedimensional overburden effects and thereby simplify the interpretation of the response of a discrete conductor beneath such overburden. The combination of multiple transmitter and receiver orientations with high-moment time-domain transmitters yields a new ‘Tensor TEM’ type of dataset which is unique among current airborne electromagnetic survey technologies. Substantial research is still required to develop practical methodologies for processing and interpretation of such datasets.

Artificial Neural networks (ANN) are used in the interpretation of airborne electromagnetic (AEM) data with horizontally layered earth structures. The problem is separated into three different tasks. At first an ANN is trained to interpret the AEM data with a homogeneous half-space, then a two layer model. Since the error of the ANN for the 2-layer model increases when data represents a half-space response, a separate ANN is trained to classify data as representative of a homogeneous half-space (HHS) or a 2-layer model (2LHS). The use of ANN classification for a 2-D (dyke) model is also discussed in this paper.

Airborne electromagnetic (AEM) survey data are traditionally presented as images of EM response, apparent conductivities, or apparent conductances at different delay times. Advances in control of system geometry and waveform monitoring, as well as increases in computing power, now makes it feasible to implement more rigorous processing algorithms. This paper discusses the application of layered-earth inversion to model subsurface conductivity structure from AEM data acquired with the QUESTEM-l00sw system.

Statistical methods are used to decide which parameters and data to include in a layered-earth inversion. Results show that the optimum number of layers is three and that it is preferable to model relative decay values (with respect to the last channel) than absolute values. Inversion results are presented as a regolith thickness map, bedrock topography map, conductivity depth sections and conductivity-depth slices. These maps, inverted sections, depth slices and profile data can be imported into a geographic information system (GIS) allowing the interactive overlay on, and integration with, other relevant data. Modelling entire QUESTEM-100sw surveys with layered-earth inversions provides the interpreting geologist with conductivity variations in three dimensions.

When airborne electromagnetic (AEM) data is acquired as a streamed or time-series data set, the great redundancy in the data favours compression as a first step in processing. Traditional data compression schemes are time windowing and spatial averaging. An alternative, more efficient data compression scheme is to transform time or frequency domain data to time-constant tau space, which has the effect of removing the waveform dependence of the AEM response.

When there are many local anomalies and a variable background, the next stage of rapid processing is to transform the response to a conductivity-depth image (CDI) to facilitate geological interpretation of the background response. Use of the full time range of recorded data, particularly the inclusion of on-time data, improves the stability of the CDI process.

The final AEM data processing step for mineral exploration is to assess the likelihood that any local anomaly corresponds to a desired economic target. This step involves the extraction of target geometry and conductivity information from the AEM data. The only economically feasible route at the present time is to parameterise both the data (using inductive and resistive limits) and the model to allow inversion of the local anomalies. A fit to one or two platelike conductors can be achieved in seconds; fits to a blocklike body take minutes on a fast PC. A significant research challenge remains to speed up and stabilise this process.

Decomposition of transient electromagnetic (TEM) data to a set of exponential basis functions (the τ-domain) is an essential step for some fast interpretation methods. Traditionally, the parameters in the τ-domain, namely the time constants τ, their corresponding amplitudes, and the number of τ‘s have been determined via trial-and-error. In this paper, all these parameters are automatically extracted from TEM data by using the matrix pencil method (MPM) based upon the representation of a sum of complex exponentials.

The MPM has been shown to be very robust and efficient computationally. For typical TEM data with about 20 channels, the transient responses have been reproduced with half a dozen exponentials within an RMS error of less than 5%. A by-product of this study is that three other representations of a transient are formulated in order to cover wider dynamic ranges of both responses and sampled times, which are likely to fit the data within much less fitting error. However, determination of the geometric and geoelectric parameters of the target in the ground from the natural frequencies, another inherent parameter related to time constants and estimated from the decomposition of the complex exponentials, remains an interesting and challenging task.

An improved conductivity-depth transform method for interpreting time-domain electromagnetic sounding data is amenable to a wide range of transmitter waveforms, source/receiver configurations, and geoelectrical situations. The method is based on a deconvolution scheme that can take into account an arbitrary transmitted current pulse to estimate the step response of the earth from traditional off-time transient electromagnetic (TEM) measurements. Then a rapid imaging scheme that makes use of a current-filament image of the source is applied to transform step response decay curves into resistivity-versus-depth profiles. No assumptions are made regarding the early-time or late-time behaviour of the measured transients. Because of the absence of this or other constraints, the method allows considerable flexibility in its application to real data. This interpretation technique is also fast enough to permit real-time processing of airborne data.

Some new approximate velocity-conductivity functions associated with horizontal and vertical airborne loop sources are derived from a study of the behaviour of electromagnetic fields diffusing through various quasi-layered models. Furthermore, new software has been successfully applied to several ground and airborne data sets, including a GEOTEM data set from Nevada. Comparison of these results with DIGHEM data, ground TEM data, drill hole logs, and geological control indicates that this improved interpretation method produces results that are more accurate than those generated using a commercial conductivity-depth transform scheme that was applied as part of the original survey.

A major use of airborne electromagnetic (AEM) surveys is to identify two and three-dimensional scattering effects that may be indicative of mineralisation. Where the background is resistive, these effects can be interpreted from simple properties of the scattered field. In conductive areas however, this approach is not generally feasible and conductivity maps are usually generated instead. Such maps assume that the ground can be approximated with a one-dimensional model. However, representing two and three-dimensional scattering effects with a one-dimensional model can be misleading.

This paper presents a new approach for processing AEM data to image two and three-dimensional structure in a conductive background. These structures are assumed to be discrete, compact impedance contrasts that are replaced by distributions of unknown induced-sources. Since the scattered field is linear function of the induced sources, a set of linear equations can be solved to find the impedance contrasts. An approximate solver is used in the solution. This solver is rapid enough to be applied routinely to large AEM data sets, and is tunable so unwanted artefacts can be rejected.

Two examples are presented using synthetic data. The first example illustrates how the solver can be tuned, and is presented for a two-dimensional imaging problem with free-space scattering. The second example illustrates three-dimensional imaging in a conducting environment.

AEM data collected over quasi-layered regolith structures is readily and conveniently interpreted using stitched 1D models. As a check on the validity of this process, a 3D modelling program was used to generate a ‘blocky’ regolith conductivity profile to characterise the gross conductivity structure seen on a stitched 1D model of a 12-km AEM survey line. The differences between model data based on ‘stitched 1D conductivities’ and the original field data highlight ‘interpretation errors’ resulting from a simplistic 1D assumption. In the case tested, large conductive zones (1 km or so in lateral scale) directly interpreted from a stitched 1D conductivity section underestimated the required local conductivity and overestimated the true lateral dimensions. Interspersed resistive areas are thus spatially more extensive than they appear on the stitched 1D model.

Simple 3D modelling does not however automatically correct for survey deficiencies. For example, probable errors in bird location resulted in surficial conductors being modelled at a ‘false’ depth of 10 to 20 m using the nominal but imprecisely known AEM system geometry. Topography should also be modelled to account for all observed features, but this was not possible with the 3D-modelling program used.

We derive a closed-form solution for the transient magnetic field developed by a magnetic source located above a thin conductive layer. The solution extends the solution found by J.C. Maxwell for a conductive thin sheet. In addition to Maxwell’s expression, the new solution explicitly depends on the thickness of the conductive layer. It can be used in a rapid inversion algorithm. The new solution allows an interpreter to estimate the thickness and conductivity of the conductive layer in addition to its conductance.

Airborne Electromagnetic (AEM) methods are very good for detecting conductive ore bodies within a resistive host, but have not been as successful for detecting ore bodies located beneath conductive regolith. We believe that in the effort to improve the ability of AEM systems to function as “bump finders” in regolith dominated terrains, the potential of AEM to contribute to other spheres of exploration activity in these settings has been neglected.

The evidence available from Australian surveys suggests that AEM could be an effective alternative to aeromagnetics when mapping geology through cover – particularly where the rocks are magnetically quiet and the presence of transported materials precludes the use of multispectral remote sensing data or airborne radiometrics. However, the success of AEM as a mapping tool in regolith dominated terrains is dependent on an adequate contrast in the conductance or conductivity-thickness product of saprolite developed over different rock types. An example where this contrast is present is in the vicinity of the Ida Fault in Eastern Goldfields Province of the Archaean Yilgarn Craton, Western Australia, where AEM has successfully mapped variability in the thickness and conductivity of saprolite developed over felsic and mafic rocks.

Contrasts in the conductance of saprolite derived from different lithologies may be obscured by saline ground-water. Saline groundwater may also blur differences in the conductivity of in-situ and transported materials. This is the case in the Black Flag area, Eastern Goldfields Province, Yilgarn Craton, where AEM appears to map total regolith thickness but fails to differentiate between the saprolite and palaeochannel clays.

On the other hand, if AEM can provide information on groundwater conditions, then the technology may have a role in mapping the dispersion of solutes within the regolith and in identifying possible sites of supergene enrichment. The potential of AEM in this regard is illustrated with reference to data from the Lawlers area, also in the Eastern Goldfields Province.

The advances in airborne time domain EM over the last seven years have facilitated its evolution into an effective mapping system from what has been ostensibly considered primarily a massive sulphide targeting tool. During this period Noranda undertook a series of programs which provided insights into the technical issues and interpretation challenges that the use of airborne time domain electromagnetics (TDEM) technology presents to exploration geologists and geophysicists. Following a model study and test flying program over a sediment hosted copper prospect on Victoria Island in the Canadian Arctic, a series of covered terrain porphyry copper exploration mapping programs were undertaken in southwest Arizona and Sonora Mexico between 1993 and 1997 using both the QUESTEM and GEOTEM commercial airborne TEM systems. The results of this work were encouraging but underscored the need for ground validation programs and robust geologic models to properly interpret the data. A need for lower noise and better calibrated EM systems was also recognized.

The thrust of Airborne TDEM technology development has been to improve conductor (massive sulphide) detection to greater depth in ever increasing complex geologic environments. This application is often referred to as bump finding in reference to discrete anomaly amplitude signatures superimposed along ostensibly flat background, profiled channel data. However, as processing software improved (i.e., conductivity-depth inversion/transformation algorithms, CDI/CDT) and integrated exploration models were applied, it became clear that this technology could be applied as a broad geologic mapping tool. This required a shift in how the data was viewed; instead of “anomaly picking’ which normally utilised around 1% of the data, the previously ignored background data which contained useful geologic information became the focus of analysis. As airborne TDEM instrumentation and interpretation software improved, airborne TDEM has become a much more effective tool for geologic mapping.

GEOTEMDEEP surveying in 1997 over Noranda’s San Jose, Arizona, and adjoining Sol porphyry properties reliably identified known subsurface bedrock features ranging in depth of 150m to 200m. Diamond drilling results suggest that airborne TDEM mapping locally defined the bedrock interface to depths in excess of 300 m. Evidence also suggests that lithologic discrimination under cover is locally possible.

Survey results over the Poston Butte, Arizona, porphyry copper deposit show how airborne TDEM technology can be used as a direct targeting tool. The signature is unfortunately not unique, and similar to inliers of conductive cover. For this reason ground validation programs are essential components of any exploration program.

While airborne TDEM technology has evolved to the point where pseudo 3-D mapping (spatial analysis of 1-dimensional CDTs) capabilities are possible, more reliable and robust inversion software, improvements in system noise levels and stability, plus a better understanding of the geologic models are still needed to provide reliable interpretation.

A multi-parameter helicopter-borne magnetic, electromagnetic (HEM) and radiometric survey completed in 1995 covered the entire Bathurst Mining Camp in Eastern Canada. The five frequency multi-coil HEM system used for this survey covers a wide frequency range and allows detailed conductivity mapping of the ground. In particular the apparent conductivity map generated from the 4433 Hz co-planar EM data shows good correlation with the bedrock geology. Conductive units are accurately delineated and can be used to help supplement detailed geological mapping. Moreover, the conductivity map shows excellent correlation with the calculated first vertical derivative of the total magnetic field, providing a means to distinguish between overburden and bedrock conductivity responses. Many orebodies are associated with discrete, coincident magnetic and conductivity anomalies.

In south east Australia during the first half of 1996, Rio Tinto Exploration Pty. Limited commissioned QUESTEM airborne time domain electromagnetic (AEM) surveys at Black Range in western Victoria and Vernon Hill in northern South Australia. The aim of these surveys was direct detection of conductivity anomalies: base metals in Cambrian greenstones at Black Range and diamonds in the Precambrian Yoolperlunna Inlier at Vernon Hill. Neither was successful in their primary objective, and the data from each survey are used here to investigate the value of AEM in geological mapping.

At Black Range, lacustrine deposits along the drainage features which cut through many of the major geological units. This conductive surficial material dominates the AEM survey data. Away from the streams, it is still possible to discriminate the major rock types of the area but not successfully map the boundaries between them.

In contrast, conductive surficial material at Vernon Hill was less extensive and the QUESTEM data appears to be providing a considerable amount of geological information. Although very little geological mapping has been done in the area, the close relationship between magnetic and conductive domains within the survey area and the evidence for structural boundaries in the topography which coincide with boundaries of the same conductive domains is very compelling.

The data show that, in areas where there is limited obscuration by surficial conductors, it is possible to use airborne electromagnetics to map geology. In such circumstances the AEM information is complementary to that provided by magnetics so that information from one enhances the interpretation of the other and vice versa. In this way, the use of both data sets provides more than twice as much information.

Explorationists are increasingly concerned with finding more difficult targets such as those under cover. Significant developments in Airborne Electromagnetic (AEM) technology are now providing advanced tools for exploration in these scenarios. Recovering the magnetic field B rather than dB/dt from a towed sensor boosts the response from a high-conductance target, as is readily shown with model calculations. The advantage of B-field data in the presence of noise is illustrated with a GEOTEMDEEP™ field example over a known conductive body that is covered by a thick conductive sequence and is difficult to detect from dB/dt data alone.

A comprehensive evaluation of the application of airborne geophysics to the management of dryland salinity is currently underway in Australia. Known as the National Airborne Geophysics Project (NAGP), it is a joint initiative of the Commonwealth and State Governments designed to assess the value-added contribution that can be made by time domain EM, frequency domain EM, magnetics and radiometrics to those data (including topography, soils and satellite imagery) already in use for salinity management. The project will concentrate on five selected catchments in Queensland, New South Wales, Victoria and Western Australia.

The focus of the project will be to define the geological characteristics, to measure three-dimensional variability in conductivity in the regolith system, and to determine the spatial variability in soil characteristics. The analysis will allow the derivation of products such as maps showing salinity hazard, water resource target, soils, regolith conductivity and thickness, and geological interpretation. These derived products will be integrated with other datasets to specifically assist with the development of land management plans at appropriate scales and to contribute to the understanding of hydrogeological processes. The project will also produce cost/benefit analyses and confidence level estimates associated with the use of airborne geophysical and other data (Woodgate et al., 1998).