ASEG Extended Abstracts - 2nd Australasian Exploration Geoscience Conference: Data to Discovery, 2019

A high-resolution shallow seismic investigation involving a 3D-3C, 3D-1C, 19 zero-offset ZVSP and a 3D SWD was undertaken in an effort to image a subsurface iron ore deposit. The same project area was used in all of the surveys and was located in north Western Australia. The surveys were acquired over period of 18 months between June 2017 and November 2018. The aim of the investigation was to determine if high-resolution seismic imaging could provide a cost effective means to explore and define shallow high-grade iron ore resource.

In this presentation we analyse active source seismic results achieved by two successive 3D seismic surveys. Both sets were calibrated with borehole measurements. The first prominent reflection in the seismic data is at a depth of 15 – 30 m related to the contact at the base of the tertiary detrital sequence. The major iron ore unit is located directly beneath the unconformity and dips gently to the south-west. The grade is variable throughout the survey area and its seismic signature appears to identify some of these changes. Velocity models produced using 3D refraction tomography and surface techniques shows anomalous interval velocities related to the iron rich zone. An elevated amplitude response was also detected in the same zones.

An in-hole electrode array was installed and cemented behind a fibreglass casing in a vertical monitoring well as part of an in-situ laboratory test for CO2 migration in a shallow fault zone in Harvey, Western Australia. The array consisted of 32 electrodes spaced at 3 m intervals, located between 277 m and 370 m below ground level. The intention of the experiments was to assess the value of permanent in-hole electrical resistivity systems, during a shallow-release of CO2. The distance between the monitoring and injection wells was close to 7 m. Food grade CO2 gas was injected through a perforated interval located between 336 m and 342 m below ground level. In the order of 100,000 time-lapse dipole-dipole electrode quadrupole measurements were acquired during the experiment. There was a clear change in measurements at the depth of the injection interval during the release of CO2 gas. Results are being compared with seismic, pressure and temperature data from the monitoring and injections wells from periods before, during and after injection.

Trial passive seismic HVSR survey lines were carried out in the north-western part of the Wilga sedimentary basin in Western Australia, which is a small and shallow isolated sub-basin to the Perth Basin, sitting above rifted Archaean granite-gneiss basement rocks. The HVSR survey method was trialled to determine if it could resolve subsurface basin features quickly and cheaply. This trial HVSR survey was carried out as a series of transects along open walking tracks. It was very successful for mapping the general basin architecture by resolving bedrock depth to a maximum depth of close to 200 m along the survey transects, and defining a half-graben structure that forms the faulted eastern boundary of the basin, with the basin depth gently deepening toward the west. The HVSR data are also interpreted to have detected subtle layering within the basin sedimentary sequence to possibly map Permian black shale beds, coal measures and splits sitting below an unconformity of Cretaceous sedimentary cover. The passive seismic HVSR method has potential to be a robust survey method for defining basin architecture in shallow coal basins with poorly indurated basin fill deposits, especially in remote parts of the world with difficult access and limited exploration funds to carry out active seismic surveys.

Petrophysics prove powerful as a complement to geological and mineral exploration geophysical studies. Statistical relationships between magnetic properties and thin section analysis can be used in a methodology to characterise iron oxide ore deposits. In this study drill core samples are studied and reveal noticeable changes in susceptibility across the Verwey transition temperature (153°C), Curie temperature (580°C) and Néel temperature (680°C). High temperature measurements allow for the contribution of magnetic susceptibility to be separated out, and directly related to ore content in drill core samples. This study shows that a mere 20 samples were enough to delineate a linear relationship susceptibility and magnetite content at Blötberget, Sweden with R²=0.73. In 6 out of the 20 measurements a similar relationship with R²=0.81 was indicated for hematite. Low temperature measurements also found that samples within this lower-amphibolite facies setting held a Morin transition suppressed to -60 °C, or missing. It has been suggested that the temperature discrepancy is likely an effect of elevated traces of Vanadium and Titanium. It is also possible that suppression is caused during the formation of martite and/or Fe(III) replacement by Fe(II).

Ambient noise tomography is a new method that does not require an active source to image the subsurface. As a result the method is a low-cost, environmentally friendly technique that can be used to explore mineral deposits. Recent developments in compact autonomous seismic recording stations allow for continuous recording of seismic data for months at a time at low cost, further increasing the attractiveness of the method. In this presentation we show results from a recent experiment where 200 seismic stations were places adjacent to an active mine in order to image the subsurface. By using ambient noise tomography, body-wave travel time and reflection tomography, we were able to create high-resolution images of the subsurface . This in turn could significantly reduce the amount of drilling that is needed in the area to probe potential ore localising structures. The results of the experiment indicate that passive seismic methods are well suited to mineral exploration.

Portions of the VTEM ET dataset from the Upper East River survey in Colorado were strongly affected by airborne induced polarization (AIP) effects. Ignoring AIP contributions to the data leads to difficulties in geological interpretation and loss of information. Undertaking full IP modelling in the inversions produces a 3D description of the Cole-Cole parameters that yields more robust and complete result. Preliminary estimates of apparent chargeability display significant differences compared with the full inverted chargeability.

Predicting the mineralogical composition of shales is crucial for drilling operations related to hydrocarbon exploration/production as well as for the assessment of their sealing capacity as hydrocarbon or CO2 barriers. Despite the importance of inferring the mineralogical composition, few methods have been developed.

A regional smectite-rich seal with a thickness of more than 1 km is hindering hydrocarbon exploration in the Northern Carnarvon Basin, North West Shelf of Australia. The complex structures of the channelised oil and gas fields in the area make it necessary to drill deviated wells through that seal. The maximum deviation angle at which successful drilling is possible strongly depends on the fraction of swelling clay minerals in shale composition, in particular, on the smectite content. Here we introduce a workflow to infer shale composition that combines seismic data, well logs, and laboratory measurements. It is applied to the Duyfken 3D seismic survey in the central part of the Northern Carnarvon Basin. Results of the interpretation are verified against laboratory XRD measurements from a test well that was not used for interpretation. The results match the test data well within the determined uncertainty bounds. Previously unpublished knowledge of shale mineralogical composition allows for a further analysis of the rock physics properties such as hydraulic permeability which is crucial for reservoir engineering and fluid flow simulations for CO2 sequestration and nuclear waste disposal.

The Bremer Sub-basin occupies a unique position on Australia’s south-west margin. It is influenced by two rift events associated with the breakup of Gondwana: rifting along Australia’s western margin in the Early Cretaceous as well as the southern margin rifting in the Late Cretaceous. The basin is underlain by Proterozoic granites, gneisses and sedimentary rocks of the Albany- Fraser Orogen and the structural fabric of the basin is strongly influenced by the basement architecture. Weaknesses along shear zones localised deformation in the early rift phases, resulting in complex structures including ramp-flat faults and associated extensional folds.

The western part of the basin contains reactivated NWtrending shear zones in the basement, which are visible onshore in magnetic data and are shown to extend offshore using satellite gravity data. Strike-slip reactivation of these shear zones resulted in isolated deep depocentres and basement highs. In the central and eastern part of the basin, salt diapirs and associated salt withdrawal and expulsion structures were active during rifting and breakup. The salt is likely Proterozoic in age and is tentatively correlated with salt in the Polda Trough, 1500 km east of the Bremer Sub-basin, which has implications for our understanding of the pre-rift architecture of Australia’s southern margin.

Badlands is a basin and landscape evolution forward model for simulating the evolution of surface topography, sediment transport and sedimentation at a large range of spatial and time scales. Here we use the Bayesian paradigm to find the best-fit parameters driving basin evolution models using Badlands. Inference in a Bayesian framework is obtained via the modelled distribution of the unknown parameters. We implement parallel tempering Markov chain Monte Carlo (PT-MCMC) using high performance computing to accelerate parameter space exploration of the computationally expensive Badlands model. Our results show that traditional implementations of single chain MCMCs rarely converge and lead to misleading inference. In contrast, PT-MCMC not only reduces the computation time, but also provides a means to improve the sampling for multi-modal posterior distributions. This motivates its usage in regional basin and landscape evolution models, allowing us to determine the relative importance of different parameters driving basin stratigraphic evolution. Parameters that can be explored include time-dependent tectonic and dynamic topography, precipitation, rock erodibility, flexural rigidity of the lithosphere and relative sea level fluctuations.

We created a funnel-shaped magnetic body from magnetite powder dispersed in plaster and used a travelling 3-component fluxgate magnetometer to map the magnetic field at a low elevation above it. This provided a dataset with signal and noise characteristics similar to those of a field survey, but for a source much better known than any buried geological body. We then used this survey data and the known source details to evaluate recovery of that information from inversions with different degrees of freedom and constraint. This provides guidance in evaluation of inversion results from field data for which the source characteristics are unknown.

We found that because of small imperfections in the data and model, the inversion result closest to the truth, although fitting the data quite acceptably, is not the model with the smallest data misfit. Chasing further reduction in data misfit in some cases leads to inversion results which better fit the data but which diverge from the known magnetization. Furthermore, inversions to fit the noise-free field forward computed from a digital version of the model do not recover that exact model, with increasing deviation (but smaller data misfits) as increasing complexity is added to the inversion models.

Noise is a fundamentally limiting factor for what aspects of the subsurface can be imaged using an AEM system. This paper compares the use of two inversion techniques and two different methods of estimating data noise, to determine how the choice of noise handling impacts inversion results. The use of processing to remove noisy data from an AEM dataset is shown to provide an improved result over using more traditional survey wide system noise level estimates. This result highlights the importance of critically assessing noise present within AEM data and provides a foundation for further development of techniques which can identify and remove noisy data.

Broadband magnetotelluric data were recorded at 104 sites between 2015-2018 in northern Sweden to image the geoelectrical upper and lower crustal structures. Data processing was performed using a robust multi-remote reference technique. The dimensionality analysis of the phase tensors indicate complex 3D structures in the area. A 3D crustal model of the electrical conductivity structure was derived based on 3D inversion of the data. Processing of regional potential field data was performed and structural information derived from these data were compared with the 3D conductivity model.

Locations of known mineralizations are compared to the regional geophysical data in order to investigate how the regional geophysical data can be used for better informed mineral exploration. The analyses indicate that regional geophysical can provide very useful information with respect to the prospectivity of different areas.

Carbonate cemented zones were identified within the lower Paaratte Formation of the eastern Otway Basin, Victoria, in southeastern Australia. A wireline log model trained on these zones can predict carbonate cementation within Late Cretaceous to Paleocene reservoir sandstones of the Latrobe Group, Gippsland Basin, 100s of km to the east. Predictions are supported by published evidence and match cementation facies interpreted for other data acquired there. These sandstones are thought to have once been heavily cemented prior to development of secondary porosity that produced the world-class petroleum reservoirs we see today. Cemented zones that remain must act as obstructions to reservoir fluid migration. They may also react with the mild carbonic acid that would be introduced by CO2 storage (sequestration) operations of the future. Model predictions show that cemented zones are sparse, spatially sporadic and fall well below seismic survey resolution at modern-day reservoir depths. This study emphasises the challenge in mapping their distribution but provides predictions against which to measure future mapping attempts.

This paper present results of Monte Carlo simulations of shielding design against neutron and gamma-rays from a D-D 2.5MeV neutron generator. The generator will be located in a restricted access laboratory at the Department Of Exploration Geophysics at Curtin University. To protect staff and students from radiation we need to calculate shielding characteristics needed to reduce the effective dose, from the generator, to safe limits.

Since operation facility is of limited dimensions, shielding needs to be optimised in terms of it thickness and the cost as well. Shielding calculations were made using the MCNP6.1 Monte Carlo code. We were required by Radiological Council of Western Australia to put sufficient shielding to achieve a conservative dose constraint for non-radiation workers of 0.5 mSv per year or 9.6 μSv in a week. The shielding was modelled as a hollow sphere of varying shielding thickness of borated polyethylene (BPE), concrete and lead (Pb). Our goal was to determine thickness of concrete needed to decrease the effective dose below prescribed limits. We already purchased 15cm thick BPE and 2.2mm Pb slabs. As a result, we concluded that 15cm thick concrete shielding will be enough to safely operate neutron generator.

Our neutron generator will be one of the main components of our proposed prompt gamma neutron activation (PGNA) logging-while-drilling (LWD) tool. This tool should be able to reliably identify the major elements of rock units, including the presence of metallic ores. The availability of such real-time information should improve almost every stage of mining and mineral processing.

We investigate the applicability of different types of deep neural networks for the estimation of subsurface properties from seismic data. The pre-trained networks can predict velocity models from new data in a few milliseconds, which makes this data-driven approach especially important for multidimensional inversion, where conventional methods inversion methods suffer from large computational cost. At the same time, realistic one-dimensional models such as the 160-layer velocity model used as an example in this study require large synthetic datasets for training, which are not always possible to obtain. Hence, we also study the impact of extending the training data by adding random noise to the modelled examples. We observe that enlarging training datasets by adding synthetic noise to existing samples improves the quality of inversion without a significant increase in computational complexity.

Within the South West Hub Project we conducted a comprehensive integrated study to map the fault and fracture network surrounding the Harvey 3 well to obtain an understanding of their propensity to act as conduits for the injected CO2. Here we report only on the seismic investigations that had two objectives: The first one was to provide precise structural model for subsequent geomechanical studies. The second objective was design more environmentally friendly acquisition and overcome the land access issues. This is of a crucial importance at Harvey where farmers are opposed to any land disturbance that can affect their activities. We thus designed a comprehensive survey that addressed both objectives. The survey included a conventional component consisting of simultaneous recorded 3D surface and 3D VSP data sets and an alternative component that recorded multi-offset VSP survey along the public roads. 3D imaging results using diverse VSP geometries were compared to 3D surface data. Both products, borehole and surface 3D images, were used for the structural analysis. Multi-offset VSP survey on the other had was reanalysed using different decimation strategies for the purpose of optimising CO2 sequestration monitoring strategy. The main outcome of this analysis was the imaging concept that is limited only to public roads which bypasses access restriction and improves chance for better public acceptance in the future.

The standard methods used for measuring gas relative permeability during brine imbibition have been found to be inadequate as the resulting curves are incomplete and may include invalid points. CSIRO, together with an industry partner, developed new methods to improve on existing special core analysis, and generate substantially more complete gas relative permeability curves for samples from a NW Shelf gas field and also a Berea sandstone sample considered as a laboratory standard.

The true relative permeability curve is typically ‘S’ shaped or has a rolling over, convex-up shape that is completely different from the concave-up shape of the Corey relperm curve usually fitted to SCAL test data and also assumed in reservoir simulations models as the default.

Using the complete and more accurate relative permeability curves can give reservoir engineers more confidence in field development planning, and along the life of the field should result in better history matching. There may also be an economic upside if the reservoir produces gas at a high rate for longer than was originally predicted based on the old relperm curves.

Modern machine learning approaches to multivariate geochemical tectonic discrimination overcome limitations of classical graphical methods, and leverage large public geochemical databases to achieve high classification accuracy. Here we document practical approaches to developing classification models using free and open source tools, and investigate methods for visualising classification uncertainties and high-dimensional spatial relationships. In applying these classification methods into deep time, issues of biases and secular change become apparent. We discuss some opportunities to build upon these models and provide richer insight for future classification models.

Large geophysical data has traditionally been difficult to manage in a consistent, open, and efficient manner. The demands of modern, large-scale computing techniques, coupled with the need for sound data and metadata management, mean that established data formats and access methods are no longer adequate.

Geoscience Australia (GA) has been working with its partners to leverage and extend existing data standards to represent various geophysical data in modern scientific container formats including netCDF & HDF. The new data encodings support rapid and efficient data subsetting, either directly from a file or remotely via web services. These will underpin GA’s future data delivery pipelines for Australian government-funded geophysical data.

NetCDF efficiently handles multi-variate raster, line, and point data, as well as n-dimensional data structures supporting more demanding applications such as AEM and airborne gravity data. Structural and metadata standards deliver interoperability, and existing and emerging data types are supported without loss of precision or other information.

This extended abstract will cover:

1. The rationale for Modernising GA’s geophysical data holdings into modern open standard container formats
2. An outline of the netCDF4 file format and associated tools, and some of the benefits they provide
3. The open-source tools and methodology used to translate grid, line, point and other data into netCDF4, and to perform metadata synchronisation
4. A brief description of a live use case exploiting web services