ASEG Extended Abstracts - 1st Australasian Exploration Geoscience Conference – Exploration Innovation Integration, 2018

Three-dimensional controlled-source electromagnetic (CSEM) surveys can be a useful technique for oil and gas hydrate detection in marine environments. Electromagnetic waves are emitted from sources, and the ensuing electric and/or magnetic fields are recorded at one, or more receivers. The number, frequency, and position of sources and the placement of receivers depends on the particular application. The solution of an inverse problem is required to recover the earth’s conductivity, which can be either isotropic or anisotropic in nature.

A major issue with either an isotropic or anisotropic CSEM inversion is the computational cost associated with the solution of many linear systems of equations. This is a result of a large spatial domain potentially containing complicated bathymetry, as well as the existence of thousands of source and frequency combinations. Overall, there could be thousands or even millions of systems of equations to solve on expansive meshes. To assist with these numerical issues, we use ideas developed for airborne electromagnetic inversions. First, we incorporate a locally refined mesh for the forward problem, specifically optimized for a source and set of receivers. Second, we use stochastic programming techniques to solve the CSEM problem with many sources and receivers. These methods dramatically reduce the numerical cost of each forward model as well as the total number of simulations. In this work we describe the methods used to overcome these computational difficulties.

The anisotropic behaviour of shales is commonly associated with the properties of a transversely isotropic medium, which are determined by five elastic constants such as five independent components of the compliance or stiffness matrix. In this study, we utilize the laboratory low-frequency technique based on stress-strain relationship to measure the dispersion of five independent stiffness tensor components and Thomsen’s anisotropy parameters of shale samples saturated with water at four different values of humidity in the range from 12% to 97.5% (12, 44, 72 and 97.5%). We have investigated three shale samples from the Wellington formation cored along the horizontal, vertical and 45°-inclination directions with respect to the bedding plane at seismic frequencies between 0.1 Hz and 100 Hz.

The obtained experimental data show an increased softening of the samples, which manifests itself in reduction of the transversely isotropic Young’s moduli and Thomsen’s parameters of elastic anisotropy e and y, no noticeable changes in parameter S were found. We also observed large reductions in normal and shear stiffness tensor components with saturation. When the samples were saturated at a relative humidity of 97.5 %, the softening at the higher frequencies was partly compensated by the modulus dispersion.

We presume that the weakening of the elastic moduli and components of the stiffness tensor is caused by the significant percentage of water-swellable smectite in the Wellington shale.

It is common to use ultrasonic techniques to measure elastic properties of the porous media. However, conventional methods are unable to measure local strain in ultrasonic wave. This is not clear how the velocity of wave depends on its amplitude. In this work we, 1) measured the particle displacement in the ultrasonic wave using a Laser Doppler Interferometry (LDI) and 2) measured changes of P-wave velocities with wave amplitude for elastic (Aluminium), viscoelastic (Polymethylmethacrylate), and granular media (dried Gosford sandstone). We checked this phenomena using a conventional ultrasonic receiver and linked this changes to a local strain in wave. The study indicated that for a sandstone sample by increasing of local strain produced by an ultrasonic wave from 7*10^-6 to 2*10^-5 the P wave velocity increase by 0.7%. We also analysed the accuracy of velocity measured using LDI as a receiver and compare the results with that using conventional transducers. Moreover, the effect of proper couplant of the sample and the transducer was investigated in details.

We use full wave forward and inversion modelling to estimate the elastic properties of rock samples from ultrasonic waveforms. The finite element algorithm (ABAQUS modelling software) is used to model a forward wave propagation within a homogeneous medium. For 19 mm diameter P-wave transducers, the result of the displacement waveform for a uniform source signal is obtained using both a linear and radial (about 2 mm) receiver arrays. Also, the use of a non-uniform source amplitude such as Gaussian distribution improves the displacement waveforms by few percent. The results accuracy is increased with increasing values of Gaussian standard deviation. However, for a nominal frequency of 1 MHz, the same error increases with the decreasing frequencies. Additionally, our inversion algorithm (written in Python) searches for the best Young modulus (E) and Poison ratio (v) of the medium iteratively. Finally, without prior knowledge of any threshold, the elastic parameters are estimated, and the results are consistent with the experimental measurements. These results provide a new modelling workflow to estimate the elastic parameters of the homogeneous and isotropic sample._

With the objective of improved understanding of frequency-dependent properties of cracked and fluid-saturated media, permeability and forced oscillation measurements are being undertaken on dry and fluid-saturated cracked glass media. In particular, we address some of the issues outstanding from a prior intensive study across a wide range of frequency. Our preliminary results demonstrate the potential of improved methods for the measurement of the very low permeabilities of such cracked glass media, the feasibility of torsional forced oscillation measurements at longer oscillation periods and lower differential pressures, and the benefits of improved alignment of our specimen assembly for complementary flexural oscillation tests. Such complementary flexural oscillation tests on glass-rod specimens are expected to provide further insight into the relevant fluid-flow regimes.

Current horizon interpretation techniques are based primarily on the use of seismic reflectivity. While there have been robust algorithms developed to work with seismic data, there are limitations and trade-offs with each of these approaches.

In this presentation we investigate the merits of interpretation based on the use of a colour blend comprised of phase, amplitude and fault datasets. The colour blend used in this workflow is a Hue-Saturation-Value (HSV) blend. In this blend the hue/colour is controlled by the instantaneous phase, the saturation is controlled by the amplitude, and the value/blackness is controlled by a fault detect volume.

Combining these three datasets provides a greater level of explicit information when interpreting an event. In standard cases this information can be inferred by the interpreter using secondary attribute volumes, or an autotracking algorithm performing extra calculations in the background. However, both of these approaches add extra overhead to the work being performed, reducing efficiency. Explicitly encoding phase, amplitude and fault information allows:

• Reduced incidence of cycle skipping
• The ability to pick on a particular phase angle
• Honouring of faults in autotracking
• Increased visual information in manual interpretation

These points will be reviewed through the interpretation of a number of 3D seismic datasets, with varying data quality and covering a range of geological settings.

This paper gives explicit expressions for the internal field and gradient components for a layered earth, a layer with gradational susceptibility, a dipping sheet, a sphere and a cylinder, that are exposed to an external field with a uniform gradient. However, the internal field and gradient components cannot be measured directly, as magnetic sensors must be placed within a cavity in the magnetic medium. This modifies the measured field and gradients. For low to moderate susceptibilities, the cavity effect can be calculated assuming that the magnetisation of the surrounding medium is essentially unperturbed by the presence of the cavity. This assumption is unacceptable when the surrounding medium has high susceptibility. I also give expressions that allow the true field and gradient components within a high susceptibility body to be calculated from measurements made in cylindrical cavities, such as boreholes, or in spherical or disc-like cavities.

A three component magnetometer developed to be drawn along a track beside drill core to log its magnetic susceptibility and remanent magnetization has been used to measure the external magnetic field of a palaeomagnetic sample over a survey area of side length 30 cm. The objective is to generate survey data from a source of known magnetization which can be used in testing magnetic field analysis algorithms (specifically those related to determination of source magnetization direction). The 3 component data directly emulates vector component survey data, is readily processed to derive gradient tensor data, and can be combined to generate TMI data. We have recovered measured magnetization directions to within 2° and 7° by inversion of TMI data from two initial test surveys.

The katangan sequence of the domes region in Northwest Zambia has delivered major copper-gold mines such as Kansanshi, Lumwana and Kalumbila, making it a prime area for exploring sediment hosted copper. Kansanshi with its 750Mt @ 0.74% Cu and 0.13g/t Au is among major copper-gold producer in Africa. SE dome a satellite deposit of Kansanshi was recently explored and drilled out with a significant input from both surface and borehole geophysics surveys including airborne magnetic and radiometric, airborne electromagnetic (AEM), natural source audio-magnetotelluric (NSAMT), Sub audio magnetics (SAM), induced polarization (IP) and 2D seismic profiles and downhole acoustics and radiometric . Systematic physical property data such as magnetic susceptibility, specific gravity, resistivity and IP have been measured on cores. This paper intends to integrate and visualize the data in order to fully characterize the deposit.

The depth to the top of a remanently magnetised source near Teetulpa in South Australia is estimated using multi-level (airborne & ground) magnetic data.

Remanently magnetised sources are considered challenging to interpret, mainly because the shape of the anomalism is a product of both the dip of the source and the magnetisation vector, with the two indistinguishable if both are unknown.

It is shown here that by interpreting the source geometry, in this case a sub-vertical pipe, the appropriate formula to calculate the magnetic response can be integrated with respect to z (vertical separation of source and sensor) irrespective of the dip or magnetisation vector, and when data exists at multiple z levels the z values can be estimated by solving a system of linear equations.

Depth estimation remains an important interpretation goal of the exploration geophysicist, especially when cover of uncertain thickness is present. There are many methods of estimating depth using magnetic data ranging from simple 1D (profile) analysis to complex 3D inversion, each with respective strengths and weaknesses. The method discussed here involves analysis of standard 2D magnetic grids.

The dipolar nature of magnetic anomalies is exploited, with the location of the nearest pole in relation to the magnetic sensor the parameter estimated. For sphere-like sources the result plots the centre of the body, where-as for bodies with extensive strike in 1D (pipe) or 2D (dyke) the result plots at the pole nearest the sensor (the depth to top).

Reduction to pole (RTP) filtering is performed on the standard 2D total magnetic intensity grid to make the anomalies symmetrical and centred over source bodies (assuming induced magnetism). The 1^st order total horizontal gradient (THG) and vertical gradient (1VD) of the RTP grid are then generated. The ratio of 1VD/THG is then calculated and gridded with contouring used to highlight the isograd with a value of 1 (THG=1VD). The depth estimate is calculated by dividing the shortest diameter of this isograd by 2 and subtracting the ground clearance. Dip information can also be interpreted.

Forward modelling of simulated data illustrates the method, proves the concept and discusses weaknesses. Several different scenarios identify the role of source topology. A real world example using open file magnetic data with known depths from drilling is then presented.

Ground-based, time-domain electromagnetic, magnetic and gravity datasets were obtained for the southern-section of the Gurubang VHMS deposit located approximately 15 km east of Cooma, NSW. The Gurubang deposit is hosted in a mid-late Silurian sequence of rock composed of shallow marine sediments and felsic volcanic rocks. The aim of this research was to ascertain the usefulness of high-resolution geophysical techniques in targeting and evaluating a small-scale polymetallic massive sulphide deposit, and to investigate how the detailed geophysics relates to the overall geological framework of the prospect area. The acquired data was analysed using a forward modelling approach. Due to the deposits high concentration of conductive minerals, a coincident loop time-domain electromagnetic 2D survey effectively delineated the sulphide mineralisation, and was useful in interpreting and adapting deposit parameters such as the azimuth, dip and strike length. Based on the physical nature of the target deposit, it was determined that high-resolution magnetic and gravity surveys would not be effective methods in directly delineating these smaller-scaled (10’s of m’s) mineral deposits. However, magnetics and gravity did prove effective in depicting the surrounding geology, including potential volcanic intrusions and basement lithologies and structures.

Borehole data is critical information to constrain geophysical inversions. Borehole information is usually used as a petrophysical constraint of the inversion scheme. Therefore, borehole information is only valuable if the borehole features are the same as the physical parameters of geophysical models. In fact, we may have many drilled holes, but the physical parameter of the geophysical model is only available in a few holes. Thus, the question is how we can exploit other borehole features to assist the inversion. We present an application of fuzzy clustering to incorporate multiple borehole features such as lithological, assay and wireline logs in the geophysical inversion. The integration of this extra information assist inversion process to build a model that fit surface geophysical data and simultaneously honour the prior information of borehole data. We apply this approach to the case study over the Kevitsa deposit within the Kevitsa ultramafic intrusion in northern Finland. The inversion of seismic reflection data with assistance from borehole information produces a more geologically interpretable image than seismic reflection data. The integration of both petrophysical and spatial attribute of the borehole enable the inversion to build a better model than only petrophysical constraint.

In situ dewatering of iron ore deposits is essential for safe and efficient mining operations, as well as reducing requirements for subsequent moisture removal for processing and transportation. Evaluating porosity, residual moisture content, and hydraulic conductivity is key to designing effective dewatering schemes.

Modern borehole magnetic resonance has been used in the oil and gas industry for over twenty years to provide continuous evaluation of porosity, bound and free fluid volumes, and permeability. As such, it is uniquely suited to provide subsurface characterisation data for dewatering scheme design. However, applying these methods in iron ore settings introduces complications that are not observed in typical oil and gas environments due to the high concentrations of paramagnetic and ferromagnetic iron-containing compounds making up the ores. This requires explicitly accounting for the impact of these compounds on surface and diffusional relaxation when estimating fluid volumes and permeability from magnetic resonance measurements.

Development of robust methods for accommodating these effects would allow for practical application of borehole magnetic resonance measurements in iron ore settings, providing continuous and cost effective hydrogeological characterisation.

Dry bulk density is a key parameter in resource estimation and mine and process planning. Ore bodies are mapped as volumes, whereas mineralisation grade is reported as mass fractions, requiring rock density to complete the reserves calculation. Similarly, although a volume of rock is to be excavated, planning for the transport and processing of this material takes place in terms of the mass of ore to be handled, again requiring rock density information to convert between the two.

Although many different densities can be defined based on the underlying mass and volume definitions, the one of most interest to the mining industry is dry bulk density, or the dry mass per unit volume of in-situ rock. This contrasts with the in-situ bulk density, which includes the mass of any fluids in the pore space of the rock. In-situ bulk density can be accurately measured using borehole geophysical techniques, but no direct downhole measurement of dry bulk density is possible. Therefore, common practice is to determine mass, after drying, and volume of core samples for calculation of dry bulk density. However, this process can be time consuming and problematic with porous or unconsolidated samples.

Another approach to estimate dry bulk density, amenable to downhole application and therefore avoiding many of the complications related to core measurements, utilises in-situ bulk density and magnetic resonance porosity measurements.Combining these two measurements allows for continuous dry bulk density evaluation without the need for coring.

Many opencast mines inhabit thousands of square km area, which are productive and commercial Australia wide. Hundreds of volumes and varieties of data dimensions and facts exist in the opencast mining areas. The data sources linked with various opencast mines are often heterogeneous and multidimensional. Data modelling is challenging in a Big Data scale, at times precluding the data integration process. The mineralization connected to opencast mines occurs in shafts, pit slopes, ramps and benches with varying geometries and configurations in large-scale geographic and periodic dimensions. The limits of the mineralization at places are either unknown and or ambiguously interpreted. The Big Data, in the context of the Australian mining industry, are due to the explosive growth of data sources and their uncontrolled management in many national and multinational companies. New knowledge is required for interpreting new opencast mining areas and their mineralization. For sustainable production, the knowledge of the connectivity between mineralization and its associated opencast mines is constrained. We propose an empirical modelling, analysing hundreds of attribute dimensions and fact instances of geological and geophysical vintages in the mining areas. Different data constructs and models are built for logical metadata, accommodating it in a multidimensional warehouse repository, as a DOME solution. It is an innovative solution to the mining industry’s Big Data problem including the opencast mine planning and design, adding values to the existing domain knowledge with new interpretations. Various geological events attributed to the interpretation and distribution of mineralization are useful for the opencast mine managers.

Fine layering is known to cause both seismic attenuation and VTI-anisotropy. In typical geological environments, the contrast in elastic properties of adjacent layers rarely exceeds 30% so that the layer-induced effects are negligible. However, it’s not true for the overburden of Northern Carnarvon basin (Northwest shelf of Western Australia) that is characterized by very stiff carbonates alternated with more porous and softer rocks.

In this paper, we present a workflow for preliminary analysis of seismic wavefield and, in particular, effects of layer-induced scattering attenuation and anisotropy in the target area. The workflow is based on the walk-away VSP full-wave modelling (5-100 Hz) for the flat-layered elastic models that are constructed using logs of four wells, namely, Dampier 1, Parker 1, Wilcox 2 and Withnell 1.

We show that particular sequences with carbonates produce significant amplitude loss and degradation of spectrum of a transient seismic pulse. Maximum attenuation is observed for Withnell 1 borehole and is characterized by the drop of 30% in the centroid frequency in the 200 m interval. Anellipticity parameter n is estimated by fitting of the moveout curves and varies from 0.1 to 1.5. In addition, the modelling reveals a very complex wavetrain with energetic reflected and converted waves at large offsets. All these effects should be taken into account in seismic processing and imaging.

On the basis of conventional methods for edge detection on potential field data, various source edge enhancement techniques have been studied to improve signal-to-noise ratio and localization accuracy. However, problems such as low resolution, noise interference and false edge information still exist. In this paper, three image processing methods are introduced, which use the Canny, LoG and Sobel operators. We describe briefly the principle of the methods and apply them to edge detection of geological bodies. As well, three typical numerical calculation methods of edge detection are selected and compared with image processing methods on edge detection effect. The results show that image processing methods can effectively identify the edge of geological bodies, especially for the Canny operator, which can prevent the introduction of errors and is insensitive to noise. To verify the practical application effect of these image processing methods, magnetic anomaly data from the Zhurihe area in China are processed in this paper. The results indicate that the Canny operator is capable of detecting the edge position of geological bodies in the study area more clearly, and that the edges corresponds to known information. Therefore, image processing methods can be used in edge detection and that satisfactory practical application can be achieved.

Low calculation efficiency and insufficient resolution in depth direction exist in inversion of underground 3D density distribution. In this paper, we proposed a fast inversion algorithm. It decomposed the gravity anomaly on multi-scale with wavelet, represented the original data sparsely with wavelet coefficients on each scale and carried on the inversion in wavelet domain. An appropriate threshold is set to process the coefficients in order to enhance the sparsity of coefficient matrix. This would furthermore improve the compressed ratio of data and save calculation time. Gravity anomalies on each scale represent response of sources at different depth as inverse relation exists between scale and frequency. So the inversion results would mainly reflect density distribution at the corresponding depth and the final inversion could be achieved by summing up results at all scales. It is not necessary to set the range of depth related to anomalies at each scale and besides, the inversion scheme is applied without depth weighting. The method could increase the resolution in depth direction of inversion results effectively and provide more detailed deep density distribution. As an iterative algorithm which can take advantage of sparse matrix to improve calculation efficiency, conjugate gradient was used for inversion. The proposed method will be applied into inversion of synthetic model data and real gravity anomaly to demonstrate its effect.