Exploration Geophysics - Volume 35, Issue 4, 2004

Knowledge of the contemporary and palaeo-orientation of maximum horizontal compressive stress (Sh^) in the Earth's crust is important for the exploration and recovery of hydrocarbons, and provides insights into the mechanisms driving plate motion. To date, most approaches for modelling intraplate stress orientations have been based on applying forces to homogeneous elastic plates. However, real tectonic plates consist of oceanic and continental lithosphere, including sedimentary basins, fold belts, and cratons with large differences in elastic parameters. We have used the finite-element method, as implemented in the software package ABAQUS along with the automatic optimisation software Nimrod/O, to model the orientations and magnitudes of SHmax over the Indo-Australian plate for the present and the Miocene. An elastic 2D plane-stress model incorporating realistic mechanical parameters for the Australian continent was used, consisting of 24 400 elements, providing a resolution of 0.2̊ in both latitude and longitude. In general, modelled SHmax directions correlate well with observed contemporary stress indicator data and reactivation histories over the Northwest Shelf and Bass Strait regions of the Australian continent, where Tertiary tectonic reactivation through time is best documented. Large perturbations in SHmax orientation over the Australian continent are shown to occur in and around regions of heterogeneous material parameters.

The configuration of earthquake belts around Australia provides a wealth of events at suitable distances to be used as probes into the seismic structure of the upper mantle. The few permanent seismic stations have been supplemented with extensive deployments of portable broadband stations for periods of a few months at each site. The broadband records have been used in a variety of studies of three-dimensional (3D) structure.

Surface wave tomography is based on matching seismic waveforms on individual paths and then mapping the path-specific constraints on shear structure into a 3D model. Higher-frequency body wave arrivals are refracted back from the variations in structure in the mantle and are particularly sensitive to discontinuities in structure. Observations out to 3000 km provide coverage of the structures down through the transition zone. For northern Australia, the combination of short-period and broadband observations provides detailed information on both P and S wavespeeds and attenuation structure.

There is a complex pattern of 3D structure beneath the Australian region. The cratonic region in the centre and west is underlain by a thick mantle lithosphere extending to around 200 km depth with fast wavespeeds (especially for S waves). However, the mobile belt in Central Australia has comparatively low wavespeeds to at least 75 km depth with fast lithospheric material beneath. In the asthenosphere, the S wavespeeds diminish and there is significant attenuation. Beneath the eastern zone of Phanerozoic outcrop the lithosphere is generally thinner (less than 140 km) and the asthenosphere beneath has a pronounced low velocity zone for S again with high attenuation.

Mohr circles have recently been used as an aid in the analysis of the magnetotelluric (MT) impedance tensor. Although well known as a representation of the stress tensor in elasticity theory, the potential application of Mohr circles to MT data was virtually unrecognised by the geoelectromagnetic induction community until the pioneer paper of Lilley (1976). An important difference between the stress tensor and the MT tensor is that the former is real while the latter is complex, which means that the MT tensor must be represented by two Mohr circles rather than one. Although early discussions on the behaviour of MT data concentrated mainly on the real part of the MT tensor, it became necessary in more detailed treatments to consider both real and imaginary Mohr circles together. In particular, identification of seven independent invariants of the complex MT tensor as geometric invariants on a Mohr circle diagram, and their physical interpretation, required the two Mohr circles to be plotted together. A significant advance has been made with the introduction of the (real) phase tensor by Caldwell, Bibby, and Brown (2004). Although the phase tensor has only three independent invariants, it has been shown by Weaver, Agarwal, and Lilley (2003) that they retain the important physical properties of the seven invariants of the MT tensor introduced earlier but with the distinct advantage that they can be displayed graphically in a single Mohr circle diagram. In particular, identification of the dimensionality of the regional conductivity structure becomes a straightforward matter whether or not the data are distorted by near-surface conductivity anomalies.

We discuss the feasibility of performing SWD (Seismic While Drilling) surveys utilizing a percussion drill as an energy source. This drilling technology is widely used in the drilling industry because it enables safe, economical, and rapid drilling. For SWD surveys, a broad source frequency spectrum is essential; however, the frequency content of seismic signals normally generated by percussion drilling is severely band limited because it is primarily controlled by the impact rate of the drill. In this paper, we report attempts at broadening the spectral energy output of a percussion drill. The frequency content of percussion drill signals can be controlled by manual variation of the hydraulic pressure input during drilling. Because hydraulic pressure change affects both the impact power and the rate of drilling, this results in a change in the frequency of drilling vibrations. We show that the response after processing is comparable with a conventional shot-record, but the percussion drill and cross-correlation method still requires improvements before it can be considered equal to conventional data.

Filtering methods based on the Fourier transform are routinely used in the processing of geophysical data. Because of the nature of the Fourier transform, the data must be prepared before the transform is calculated. This preparation usually takes the form of the removal of any trend from the data, combined with the padding of the data to 2^N points between the data edges. However, no data preparation procedure is perfect, and the result is that problems (in the form of edge effects) appear in the filtered data. When high-pass filters (such as derivatives or downward continuation) are subsequently used, then these edge effects become particularly apparent.

This paper suggests three methods for the stable downward continuation of geophysical data (two of which may be combined). The first method is applied to an integrated horizontal derivative of the data rather than to the data itself. Since the horizontal derivative can be calculated in the space domain where fast Fourier transform (FFT) edge effects are not present, this reduces the enhancement of the data at frequencies near the Nyquist, resulting in smaller edge effect problems. The second method measures the FFT-induced noise by comparing data that has been downward continued using both the space- and frequency-domain methods. The data is then compensated accordingly, and the compensated data may be downward continued to arbitrary distances that are not possible using space-domain operators. The final method treats downward continuation as an inverse problem, which allows the control of both FFT-induced noise and other noise that is intrinsic to the dataset. This method is computationally slow compared to the first two methods because of the inversion of large matrices that is required. The methods are demonstrated on synthetic models and on aeromagnetic data from the Bushveld igneous complex, South Africa.

Electrical resistivity imaging surveys widely used in many environmental and engineering studies have also been conducted in water-covered areas. Surveys in water-covered areas include conventional surveys using multi-electrode resistivity systems where part of the survey line crosses a river or stream, and surveys conducted entirely within a water-covered environment. Surveys that are located entirely within a water-covered environment utilise electrodes mounted on a streamer, towed behind a boat. The streamer can be dragged along the water bottom, or float on the water surface. In this paper, the smoothness-constrained least-squares inversion method commonly used to interpret electrical resistivity imaging data from land surveys is adapted for underwater surveys. To accommodate the water bottom topography, a distorted finite-element grid is used to calculate the apparent resistivity values for the inversion model. The first few rows of elements are used to model the water layer, while the lower part of the grid is used for the sub-bottom resistivity distribution. For robust inversion, the water column resistivity and geometry must be known accurately as a large proportion of the current flows through the water layer. The section of the Earth below the bottom surface is subdivided into a large number of rectangular cells. The water column resistivity and geometry in the earth model is fixed, and the inversion program attempts to determine the resistivity of the cells that would most accurately reproduce the observed resistivity measurements. Implementation of water column resistivity and geometric constraints is demonstrated using numerical simulations and field data. Examples of electrical resistivity imaging surveys conducted on and across water bodies including rivers and near-shore marine environments are shown.

A time-domain airborne electromagnetic (AEM) survey was flown with the HoistEM system over the Woodie Woodie manganese mine corridor in the east Pilbara of Western Australia. Conductivity Depth Image (CDI) processing helped to discriminate shallow, regolith-related responses from conductive ore. About half of the known manganese ore zones correlate with elevated conductivities on the CDIs, and recent drilling of several new EM targets in areas of regolith and bedrock cover has discovered over six large-tonnage manganese ore bodies to date. EMFlow software has made it practical to process data gathered by HoistEM to separate the ore body response from the responses of conductive palaeochannels and other conductive features related to bedrock geology or the regolith. Experimentation with EMFlow input parameters helped to produce CDI results that were more reliable. The CDI results were further refined by calibrating conductivity values from CDI processing to correlate with borehole conductivity results. Borehole-calibrated CDI results have lower noise, line levelling problems are suppressed, and depths of conductive horizons are more accurate. The HoistEM survey-design tests and processing results demonstrate that HoistEM is a cost-effective method for exploration of podiform, high-grade manganese ore bodies, as long as the host rocks are comparatively resistive, and a flight line spacing of 80 m or less is used.

A recent increase in the environmental usage of Airborne EM has shown the need to provide accurate values of depth and conductivity. Calibration problems in helicopter EM data produce imprecise conductivity depth inversions and images (CDIs), whether plotted in maps or sections. Accurate images are essential requirements in order to target smaller, near-surface objectives such as salinity outbreaks.

To ensure agreement between ground-truth (such as conductivity logs or ground EM data inversions), data recalibration has been applied before processing. A limitation of ground-based methods is that they tend to provide spatially restricted constraints. This paper presents an alternative statistical method developed to provide consistency with simple conductivity models.

Measured data and theoretical models are transformed, from amplitude in real and quadrature phase with the primary field, to two dimensionless quantities: /3, the ratio between a data prediction and a calculated inductive limit, and a, which is a standard EM response parameter. In conductivity-independent a/3 domain, the response of a variety of synthetic models based on expected geology is calculated and compared to the median of the larger amplitude field data. The data are then rescaled in the a/3 domain so that the recalibrated median response lies exactly on the theoretical curve.

The amplitude rescaling was applied to HEM datasets collected in the Riverland and Tintinara areas in South Australia. The results were compared using maps and CDI images of the raw and recalibrated data. The original data, as delivered, produced CDI images that were generally inconsistent with borehole conductivity data. However, amplitude rescaling to ensure 'thin-sheet' consistency has produced remarkable agreement in depth between ground-truth and the CDI sections.

Sea-ice thickness measurements using electromagnetic (EM) instruments require accurate data. Calibration of sea-ice thickness data acquired using a low induction number (LIN) EM sensor can be performed by conducting a geometric sounding at a range of heights over level sea ice of known thickness, and by comparing the observed data with the expected layered-earth response. Calibration corrections for scaling, phase-mixing, and zero-offset errors can be derived using least-squares inversion to minimise the misfit between the observed data and the theoretical response, and can be incorporated in modelling algorithms used to determine sea-ice thickness.

This paper presents a case history illustrating identification and correction of calibration errors in frequency-domain EM data for Antarctic sea-ice thickness measurements. Comparison of coincident EM measurements made using three EM31 instruments showed that measured apparent conductivities disagreed by up to around 100 mS/m, resulting in errors in the estimated sea-ice thickness of up to 60%. Separate calibration corrections were determined for each instrument by analysis of geometrical sounding data acquired over level sea ice. Sea-ice thickness at the calibration site was determined by making a large number of drilled thickness measurements over the footprint of the EM instrument, and seawater and sea-ice conductivities were determined using independent measurements. Best-fit scaling, phase-mixing, and zero-offset errors were determined via inversion of the geometrical sounding data, with the sea-ice thickness and seawater conductivity fixed at their known values. After application of the calibration corrections, sea-ice thicknesses derived from the three instruments agreed closely with each other and with drilling results. Calibration corrections derived in this manner have been shown to be valid over a period of at least several weeks.

Airborne electromagnetics (AEM) has been used on a test basis for determining water-depth (bathymetric sounding), with both fixed- and rotary-winged aircraft systems. Useful application is restricted to shallow water (up to a few tens of metres), and the systems can be used through sea-ice. Many processing methods to date, however, have involved extensive post-processing of data before bathymetric information can be extracted, with lag times of the order of months.

We have investigated a number of methods for rapid bathymetric depth estimation that could be used in real time with AEM data acquisition. The fastest methods are all based on an initial rapid transform of data to conductance-depth sections. Once in conductance-depth space, it is possible to estimate depth from the maximum conductance encountered at each reading if the conductivity of seawater is known, and if the sea floor can be assumed electrically resistive.

Alternatively, it is possible to attempt to solve for both seawater conductivity and depth independently using a thick-layer approximation, but this is a slower and potentially unstable process, but with the potential advantage that seafloor conductivity can be assessed. AEM and bathymetric sounding data have been used to appraise the validity of the approximate methods. These data included helicopter DIGHEM data collected over Sydney Harbour and fixed-wing GEOTEM data collected over and near Geographe Bay, Western Australia.

Interpretation of helicopter time-domain EM data from a 2002 HoistEM survey of Sydney Harbour is under way to define seawater depth. Forward modelling, conductivity-depth imaging (CDI), and ID inversion have been applied. The effectiveness of airborne EM bathymetry can be appraised by comparing interpreted seawater depths with accurate bathymetric soundings.

An extensive comparison of modelled and observed HoistEM responses over deep (~33 m) and shallow (~6 m) seawater showed that good agreement could be achieved for models with the correct depth of water if the seawater conductivity was assigned an unrealistically low value of about 3.5 S/m. The actual seawater conductivity in Sydney Harbour is 4.65 + 0.15 S/m.

CDI processing reinforced the conclusion that the measured HoistEM responses are incompatible with the known seawater conductivity. However, a qualitative agreement was observed between the CDI seafloor topography and the true bathymetry. In an effort to reconcile the data with the “ground truth”, transmitter moment and altitude were adjusted, for a given seawater conductivity, prior to CDI processing.

ID inversion of HoistEM data yielded sub-metre water-depth accuracy for depths shallower than 22 m, if an unrealistically low seawater conductivity was allowed. Thus forward modelling, CDI, and initial ID inversion results all suggest that the HoistEM system is viable for bathymetric surveys but that rescaling of the Sydney Harbour dataset will be required in order to reconcile the measured responses with the known seawater conductivity.

A novel rotating magnetic gradiometer system (GETMAG) has been designed, constructed, and demonstrated. The sensor is a high temperature superconducting quantum interference device (SQUID) operating in liquid nitrogen (-196°C). By making measurements about three separate axes, the full magnetic gradient tensor is determined.

The system has been demonstrated over a magnetite skarn deposit at Tallawang, near Gulgong, NSW, which is essentially two-dimensional (2D). The 2D structure was important because it allowed an unaliased total magnetic intensity (TMI) survey to be carried out in parallel, from which tensor components could be calculated to directly compare with the tensor components measured with the rotating gradiometer. The agreement was found to be excellent, confirming the accuracy and calibration of GETMAG and the Fourier filtering method of the calculation.

Methods developed to analyse 2D tensor data utilise eigenvalue and eigenvector systematics. Notwithstanding underlying non-uniqueness of solutions for 2D structures, all determinable parameters of location, geometry, and magnetization were found to accord with directly measured properties, i.e., information gained from either exploration drilling or laboratory measurements. Only a few tensor gradiometer stations were needed to extract the same information as a whole TMI survey.

A more general method to determine structure and location of sources is Tensor Euler Deconvolution. This method has been adapted for the magnetic gradient tensor in a least-squares fashion and applied to the tensors calculated from the TMI survey. Generally, an Euler index of n = 1 was found, as is expected for 2D sources. However, this approach allowed second-order features of the source, where n > 1, to be discerned. The skarn is interpreted to comprise a fresh, more highly magnetic, core at depth (~25-30 m) surrounded by a less magnetic mantle. The skarn extends to within 10 m of the surface where weathering has presumably contributed to diminished magnetization. Elsewhere, away from the skarn, Euler indices are low, approaching that of a magnetic annihilator = 0). This presumably reflects the uniformly magnetized alluvial soil cover.

Our next stage is to demonstrate an airborne capability of GETMAG, beginning with a helicopter platform before moving to fixed-wing. In addition to mineral and oil exploration, we envisage applications in environmental, military, and unexploded ordnance (UXO) surveys.

Delineating the edges of magnetized bodies is a fundamental application of magnetic data to geological mapping in areas of limited exposure. Especially in Pre-Cambrian shield-like regions, locating lateral changes in magnetization of the outcropping crystalline rocks provides spatial information that is crucial in extending mapped geology into sparsely exposed or completely covered areas. Although not all magnetic contacts correspond to lithological contacts, the former provide key information on structural regimes, deformation styles and trends, and magnetic texture.

Many techniques for contact mapping have been developed, some originally based on profile (2D) data and others designed specifically for grid-based (3D) data sets. Here, we evaluate five methods applied to gridded data. The first three are based on finding maxima of the horizontal gradient magnitude of the total field (TF-hgm), tilt (TI-hgm), and pseudogravity (PSG-hgm). The fourth and fifth methods rely on locating maxima of the analytic signal (AS) and the 3D local wavenumber (LW).

Method TF-hgm produces theoretically correct contact locations only when the data is reduced to the pole, and even then may produce false or secondary solutions mimicking contact trends. Method TI-hgm is less sensitive to field direction but also suffers from secondary maxima. Method PSG-hgm is perhaps the most established approach of those mentioned, and in the case of vertical contacts produces reliable maxima, however knowledge of magnetization direction is required. Method AS and LW theoretically produce maxima directly over contacts and are insensitive to magnetization direction and body dip but are more sensitive to noise than the other three methods, which limits their application to higher-quality data sets.

The effect of emanation radon is shown to be a significant problem for estimating uranium concentration from ground-based gamma-ray spectrometric surveys. Radon gas (a daughter product in the U²³⁸ decay series) escapes from rocks and soils near the Earth's surface into the lower atmosphere. Under early morning, still-air conditions, radon concentrates as a thin layer near the Earth's surface. If ground radiometric surveying is undertaken before this radon layer is mixed into the lower atmosphere, large errors in U concentration estimates result.

This paper shows the effect of early-morning radon accumulation on a gamma-ray spectrometric survey near Boorowa, NSW. Paddocks surveyed in the early morning show much higher apparent uranium concentrations than those surveyed later in the day. We demonstrate the radon diurnal effect by monitoring the equivalent U concentration at a fixed site over several weeks. Typically, there is a build-up of radon near the Earth's surface overnight. Radon concentration reaches a maximum at about one hour after sunrise, before slowly dispersing over a period of 2-3 hours. By reporting both the Boorowa survey and the radon monitoring results in the same units of equivalent uranium concentration, we demonstrate that the anomalous effects seen in the Boorowa survey are consistent with that of a radon diurnal.

The monitoring data also show the effect of rainfall on apparent U concentrations. Rain deposits radioactive daughter products of atmospheric radon onto the ground, resulting in a significant increase in apparent U concentration. These short-lived daughter products decay to insignificant concentrations within about 3 hours. Ground surveys should not be conducted during or within 3 hours of the cessation of rain, or under early-morning, still-air conditions. Also, because high moisture concentrations in soil reduces the gamma-ray fluence rate at the Earth's surface, surveys should not be conducted after prolonged or heavy rainfall where the soil profile is saturated.

A large portion of South Africa‘s gold reserve comes from the Ventersdorp Contact Reef (VCR). The VCR is a thin, tabular ore body, generally less than 1.2 m thick. The geometry of the VCR is mainly controlled by slopes, terraces, and faults. The gold is concentrated in palaeo-river channels situated on river terraces and not on the confining terrace risers or slopes, so there is a good correlation between topography and grade. The geophysical problem is to determine the reef geometry before mining begins, in order to plan optimal ore extraction.

Borehole radar has already been shown to be a suitable technique for mapping topography on the VCR because of the favourable electrical property contrast between the footwall rocks and the overlying lavas. Here, a forward model is used to resolve the directional ambiguity inherent in low-frequency borehole radar data and to accurately position the target reflection in space. The data from five boreholes within a 200 m square block of ground are combined in a single model to estimate the topography of the ore body within the block. The illumination lines from the radar forward model are positioned in 3D. They are incorporated with other ore-body elevation data from mapping and drilling to improve the geological model of the ore body within the block.

The spatial resolution of the geological model is substantially improved with borehole radar. Small faults and other changes in topography are accurately positioned, giving warning of oncoming changes in mining conditions. Terraces and slopes are mapped, facilitating planning of the optimal extraction of the gold within the block. In areas where exploration drilling is primarily undertaken to determine the structure of the ore body rather than its grade, borehole radar provides an effective alternative to conventional drilling.

When seismic data are presented as two-way reflection times, these times are not easily directly scalable to depths, because of spatial variations of seismic velocity. Apparent structures in the time domain can be misleading. Correct conversion of seismic time sections to depth sections removes this ambiguity. In general, seismic depth conversion is a complex process requiring careful use of NMO and migration velocities, the study of well data, and the generation of synthetic seismograms. The process is usually iterative, especially when the structures are complex. In this paper, we present a depth-conversion algorithm designed for coal seismic data. Our method exploits the relatively simple structure of coal seams and the availability of many exploration boreholes to constrain the process. Once converted to depth, the seismic data can be exported into mine-planning software and used to provide seam elevations for tasks such as in-seam drilling and other mine activities. Our method has been implemented into an MS Windows-based program, and allows new boreholes to be incorporated without the need to go back to a seismic processing centre.

A 3D seismic data set from Xstrata‘s Sandy Creek coal mine is used to demonstrate our method. The results show that the depth-conversion algorithm can accommodate differences in seismic processing. The depth-converted seismic data agrees with the geological model based on the borehole results and underground mine surveys. Given confidence in the depth conversion, it is possible to look more closely into the seismic data in order to make interpretations that are more detailed.