ASEG Extended Abstracts - ASEG2012 - 22nd Geophysical Conference, 2012

As the number of near surface deposits decreases, it becomes increasingly important to develop geophysical techniques to image at depth. Because of the penetration advantage of plane wave natural sources, these techniques are ideal to answer questions about the deep subsurface to the earth. A ZTEM survey is an airborne electromagnetic survey which records the vertical magnetic field that result from natural sources. The data are transfer functions that relate the local vertical field to orthogonal horizontal fields measured at a reference station on the ground. While the airborne nature of the survey means that large survey areas can be surveyed quickly and economically, the high number of cells required to discretize the entire survey area at reasonable resolution can make the computational costs of inverting the entire data set all at once prohibitively expensive. Here we present a workflow methodology that can be used to invert large natural source surveys by decomposing the large inverse problem into smaller more manageable problems before combining the tiles into a final inversion result. We use the procedure to invert synthetic ZTEM data for the Noranda mining camp as well as a field data example. Both of these data sets were far too large to solve on a single grid even with multiple processors at our disposal.

Cosmic rays producing muons shower the Earth daily. These natural, high-energy particles decay as they pass through matter and are directly affected by density. Recently, sensors have been placed in existing tunnels and mine shafts that observe muon flux in a brown-field mining scenario. We have developed an algorithm to invert these data individually, or jointly with gravity data, to recover a 3D distribution of density. Muon and gravity data are both linear functionals of density but the associated sensitivity functions are substantially different. These differences in physics between muon ray paths and gravity data provide a unique insight into the subsurface. This is illustrated through synthetic examples. Inversion of a set of field data, obtained at a mine site in south-west British Columbia, Canada, illustrates the potential benefits and challenges for the technique to be used in field surveys.

In this paper we show that 3D inversion of large airborne time domain EM data, which is traditionally considered impractical, can be rapidly carried out by using a thoughtful workflow. In our 3D inversion algorithm, the number of cells in the mesh and the number of soundings are two factors that slow down the inversion. Therefore, we develop a strategy of adaptive mesh and sounding refinement to minimize the number of cells and the number of soundings required by the inversion. At the beginning, a coarse mesh and a few soundings are used to quickly build up a large-scale model. Then the mesh is refined and more soundings are added based upon their data misfit. At each iteration of the inversion, a certain number of soundings are randomly selected, and we change the data selection from iteration to iteration. This allows us to down-sample the field data without much loss of information. Once the large-scale model is obtained, we carry out some tile inversions that focus on smaller areas with a locally refined mesh to better resolve the small-scale features. The workflow is demonstrated by a synthetic example with 2121 transmitters that takes about 10 hours to be solved compared to about 150 hours if we had started the inversion on a fine mesh and used all of the transmitters. The methodology of speeding up the inversion by adaptive mesh and data refinement can also be applied to other EM surveys.

Magnetic field inversions are non-unique but a realistic goal is to find a causative earth structure that is compatible with the geophysical data, the petrophysical constraints, and with geology. Invariably the inversion results are improved as the number and diversity of constraints are increased. In this paper we concentrate upon the inclusion of geologic structural information. Geologic structural modelling programs can import faults, boundaries, and strike and dips of geologic units and interpolate this sparse information in space. When provided with a 3D voxel mesh, they can compute a strike, dip, and plunge for each cell. Following previous work, structural geologic information is incorporated into the inversion as a weak constraint by encapsulating it into the model objective function. The model objective function is formed such that each prism has its own set of rotated vectors to enforce smoothness along the direction of the geology. User-controlled parameters specify the degree of smoothness throughout the 3D volume and thus allow additional geologic insight to be directly incorporated. In addition to structural geology, the inversion algorithm utilizes reference models and bound constraints that help us realize our goal of incorporating all available information. The efficacy of the inversion is demonstrated through a synthetic and a field example.

The distribution of groundwater salinity is a key input for management of water resources. Estimates of the three dimensional distribution of groundwater salinity below areas spanning thousands of square kilometres may be required. Airborne transient electromagnetic methods provide the possibility of recovering first pass large scale solute concentration distributions provided lithological influences on electrical conductivity distribution are not dominant. The Allanooka airborne TEM survey is located in the northern most portion of the Perth Basin in Western Australia. We use data from this Airborne TEM survey combined with data recovered from monitoring wells to highlight the steps used to construct a first pass large scale solute distribution model. We provide a method for converting airborne TEM datasets to estimates of solute concentration distribution for sandstone dominated sediments at a basin scale. For the Allanooka monitoring well network, base line empirical relationships are developed between laboratory derived total dissolved solids and formation conductivity derived from wire line logs. This relationship is then extended to include airborne TEM derived formation conductivities. Appropriate layer discretisation of input seed models for inversion of the airborne TEM data set are based on analysis of resistivities derived from wire-line logs. The interpretation of the inverted airborne TEM was assisted by geological constraints and high resolution seismic reflection transects. Selected inversion statistics were also mapped throughout the 3D volume to provide a quick method for assessing the “importance” of particular layers to the outcome of the inversion.

An approximate volume of low solute concentration sandstone dominated formation below the regional water table was extracted from the airborne TEM data. Our first pass basin-scale Airborne TEM derived 3D solute concentration provides a starting point for more detailed interpretation to commence.

Elang is a large porphyry Cu-Au deposit situated about 70 km east of Batu Hijau on Sumbawa Island, Indonesia. The deposit is associated with a series of tonalite porphyry intrusions that are hosted by andesitic volcanics. Mineralisation is associated with potassic alteration (chlorite-magnetite±biotite) that produces a discrete magnetic high of about 700 nT. Forward and inverse modelling have been used to define this zone and the results have been used to direct drilling.

Very strong chargeabilities are associated with the porphyry alteration and resistivity from pole-dipole IP and airborne EM shows the extent of the alteration system. The Elang alteration system clearly shows up as a conductive zone in a relatively resistive background. There is an advanced argillic lithocap, up to 200 m thick, covering much of the deposit and is highly resistive and well defined with resistivity.

Elang is typical of a number of Cu-Au porphyry systems in that magnetite is associated with mineralisation and produces a strong discrete magnetic anomaly. It has a larger potassic zone than most systems, which may be due to more than one porphyry centre. The Elang system is more pyrite rich than many, leading to very strong chargeabilities, both in the ore zone and the pyrite halo. There is a broad resistive low due to clay alteration and sulphide veining similar to some other porphyry systems

Multiple geophysical data collected over the same area but based on fundamentally different physics usually contain complementary information about the subsurface. Joint inversion combines the complementary information by integrating all the geophysical data into a single inversion scheme. Thus, models resulting from joint inversion are more likely to represent the subsurface better than models derived from a single type of data. In this study, we consider joint inversion of seismic traveltimes and gravity data, and present a new joint inversion algorithm that uses petrophysical information as constraints. Using a synthetic example, we show that this new method can effectively build the available petrophysical information into inversion and improve the definition of both structure and physical properties. We also show that this method can deal with the situation where only partial petrophysical information about the subsurface is available. An important component of our method is applying fuzzy c-means (FCM) clustering algorithm to the recovered physical property distribution to generate a lithology map that is consistent with both the observed geophysical data and the a priori petrophysical information.

Traditional Bouguer corrections assume that the Earth behaves as a flat slab (Bouguer reduction), an assumption that is incorrect. A more accurate method of removing the effects of an Earth model is to take into account the terrain information, and assume a curved Earth. Terrain corrections are becoming more commonplace in individual surveys as computer power increases. We have produced a regional (state-wide) scale terraincorrected gravity grid, using a 1-second DEM to calculate the correction. The correction adds a subtle change to the state wide gravity, removing some regional lows. It can be used as a simple visual tool to determine where terrain-corrections are more appropriately undertaken.

Simultaneous (blended) sources have attracted a great deal of attention recently because of their potential to increase significantly the rate at which seismic data can be acquired. The viability of the method was previously demonstrated through the use of small-scale tests on synthetic and field data. In this paper, we present a case history from Australia of the first field-development-scale use of this technology in the world.

Concept studies involving simulations of simultaneoussource data from conventional data indicated that the proposed survey design would yield data that were separable into components for each source. The resultant data set contains twice as many traces as its conventional equivalent, and provides improved sampling for important processing steps such as coherent noise attenuation.

Simultaneous-source acquisition requires quality control methods that are specific to the technique to ensure that the data are acquired as planned. New QC methods were developed specifically for this project, and showed that no problems related to the simultaneous-source technique were encountered.

Data processing involved source separation at an early stage, after which a conventional processing sequence could be used on the resultant, densely-sampled data set. Separation was performed using a sparse inversion technique, which proved very effective. Very little signal leakage was observed, and the interference was almost completely suppressed.

Through this case history, we demonstrate the viability of simultaneous sources as an effective marine seismic acquisition method.

Geoscientific data interpretation is a highly subjective task as human intuition and biases play a significant role. Based on these interpretations, however, mining and petroleum industries make decisions with paramount financial implications. The aim of this study is to better understand variability in geophysical data interpretation between and within individuals. We examine the data observation pattern during interpretation using an eye tracker that captures the interpreter’s eye gaze motion. Two preliminary experiments were conducted to analyse how individuals approached the task of identifying prescribed ‘targets’ in magnetic and seismic data respectively. Each experiment used five participants who have varying degrees of experience in these tasks.

The first experiment involved identifying responses from porphyry-style intrusive systems in magnetic data of an area from Reko Diq, Pakistan. The target responses are sub-circular positive magnetic anomalies surrounded by annular lows. The second experiment was to spot unconformities and faults in a seismic image using data from the Mentelle Basin in Western Australia.

The results show a significant variation in data observation patterns between interpreters. Some key findings include: a direct correlation between a higher target spotting success rate and a more systematic data search pattern; significant inconsistency in target spotting results when viewing a data in a different orientation; and a significant variation in the amount of time spent on noise dominated region.

Automated image analysis techniques can be effectively used to detect discontinuities (e.g. faults, pinchouts, channels, etc.) within seismic data in a non-subjective manner. Conventional image processing techniques, such as the coherency cube, typically locate discontinuities by finding regions of sharp intensity shifts and are thereby sensitive to contrast variations and noise. Here, we present a phase-based technique that offers contrast-invariant and noise-robust feature characterisation through local phase and orientation information.

Phase congruency is an edge-detection algorithm that differs from traditional approaches by defining edges as points where the Fourier components of a signal are maximally in phase. Applying 2D phase congruency to horizontal time slices extracted from a 3D seismic volume is problematic, though, because horizons are rarely parallel to horizontal time slices, causing horizon boundaries to appear artificially discontinuous. To better detect 3D seismic discontinuities, we extend phase congruency to a 3D algorithm using conic spread filters that provides a localised, multi-scale and dip-independent feature detector.

Preliminary results show that 3D phase congruency is capable of detecting velocity anomalies, but has some limitations in identifying fault boundaries in seismic data. However, it can provide an increased level of feature detail over conventional coherency cube processing. More importantly, these results indicate the potential for using multidimensional phase-based algorithms in 3D/4D seismic processing and imaging workflows, with particular applications in image denoising, image registration, feature detection, and velocity model verification.

Demand for water in the Perth Metropolitan Area, Western Australia, is increasing and new water supply options need to be considered. Aquifer replenishment by injection through wells is seen as a part of the solution however before any large scale implementation of an injection well field is considered several trials are being completed. Time lapse induction and temperature logging have been completed as part of two aquifer replenishment trials in the Perth Metropolitan area. The intention of the time lapse logging is to detail the movement of water away from the injector well into the Leederville formation. A hydrothermal computer model constrained by time lapse wireline logging induction and temperature results has been created to understand the movement of water and heat during injection into the Wanneroo sandstone formation. As with most practical numerical modelling, a level of non-uniqueness in the model parameters selected will exist. It is demonstrated that the calibration to time lapse logging results provides an important constraint on the range of flow, solute transport and heat parameters that can be used to build a reasonable hydrothermal computer model. First, the flow and solute transport model is constrained with time lapse electrical conductivity distributions at the monitoring wells. Next, the model is expanded to include heat transport. Results of our modelling provide the first field scale estimates of heat parameters in the Leederville Aquifer in Perth.

We present a new survey methodology to map the distribution of chargeable material in the subsurface using inductive electromagnetic sources and observations of the magnetic fields in the frequency domain. An accompanying inversion algorithm is developed, and the technique is tested on synthetic data.

Statistical de-noising and compressive inversion methods based on Principal Component Analysis can reduce random noise, separate desired signals from correlated noise, and improve the efficiency and results of airborne EM inversions. However, inversion of PCA-processed data with standard kernels produces inaccurate results due to the improper forward mapping operators used. These inversions must incorporate the PCA rotation in the inversion process for accurate results. In order to appropriately apply these operators to the inversion kernels, the statistical distribution of the noise before and after processing and its effect on the data misfit must be understood. We can then develop compressive inversion techniques utilising PCA.

In this presentation, we demonstrate the need for incorporation of rotation into the inversion kernels through linear examples and show the utility of principal component analysis in compressive inversion. We then examine the statistical distribution of TEM data and noise and show that the noise follows a multivariate tdistribution both before and after processing with PCA. We conclude by introducing a compressive inversion technique formulated in the principal component domain.

We demonstrate a robust workflow for time-lapse gravity modeling in reservoir sequestration/production monitoring applications. This systematic approach outlines a reliable methodology to understanding the value and limitations of 4D gravity at a particular site, for both pre-acquisition decision making, and as a guide for post-data acquisition interpretation. To demonstrate, we present a multi-faceted feasibility study for monitoring CO2 injection into a reservoir at various injection times using 4D micro-gravity method. The simulations are performed for a currently active CO2-EOR site, the Louisiana Delhi Field in the United State. We construct an accurate representation of the field directly from current seismic data, followed by application of binary inversion technology adapted to the time-lapse gravity problem and tailored to the specific site. Finally, we illustrate a method of resolution analysis to demonstrate the decreased recoverability of fluid movement at the site in the presence of varying data noise.

Two helicopter TEM systems (HoistEM and RepTEM) were flown over waters in Backstairs Passage, South Australia, in 2003 and 2010 respectively to test the bathymetric accuracy and hence the ability to resolve seafloor structure in shallow and deeper waters (extending to ~40m depth) that contain interesting seafloor topography. The topography that forms a rock peak (South Page) in the form of a mini-mountain that barely rises above the water surface was accurately delineated along its ridge from the start of its base (where the seafloor is relatively flat) in ~ 30 m water depth to its quasi-submerged peak. A much smaller submerged peak (Threshold Bank) of ~ 9 m peak height located in waters of 35 to 40 m depth was also accurately delineated. These observations when checked against known water depths showed that the two airborne TEM systems were operating correctly. The third component of the survey was flown over a series of quasi-parallel seafloor ridges (resembling large sand waves rising up to ~ 20 m from the seafloor) that branch out and gradually decrease in height as the ridges spread out across the seafloor. These features provide an interesting topography because the interpreted water depths obtained from 1D inversion of TEM data highlight the effect of the EM footprint in resolving both the separation between the ridges and the height of individual ridges, and possibly also the limitations of assuming a 1D model in areas where the topography is quasi-2D.

A deep seismic reflection survey was carried out over the Meso-Neoarchean Youanmi Terrane, Yilgarn Craton in Western Australia, including 3 seismic lines across the Windimurra Igneous Complex. This gabbroic intrusion is the largest exposed single mafic-ultramafic intrusion in Australia, currently mined for magnetite-hosted vanadium in its upper zone. Acquisition was carried out with 3 Hemi 60 vibrators, 80 or 40 m VP interval, 40 m group interval, and a 300 channel symmetric split spread, resulting in 75 to 150 fold high quality data. Key processing steps included refraction and residual automatic statics, spectral equalisation, detailed stacking velocity analysis, dip moveout (DMO) correction and post stack Kirchhoff migration. The Windimurra Igneous Complex exhibits high seismic velocity (approximately 6.5 km/s), both in interval velocity from stacking velocity and bedrock velocity from refraction statics analysis. Seismic attenuation is low, resulting in high frequency and good resolution to the base of the complex, seen at a maximum two way time of around 2.5 seconds. Gross structure on the seismic sections is also consistent with the gravity and magnetic data. An inward dipping conical structure is evident and consistent with surface geological observations. The base of the complex is marked by distinctive high amplitude multi-layered reflectors, which may represent significant thicknesses of ultramafic zone material unexposed at the surface. This zone has been a long sought after target for Cr-PGE exploration. The complex internal structure of the intrusion, possibly including discordant igneous layering features, has been well imaged up to the base of regolith.

A new numerical approach, called the “sub-domain Chebyshev spectral method”, has been developed to calculate differentiations in a curved coordinate system, which may be employed for 2D/3D geophysical forward modelling. The new method utilises non-linear transformations defined by the free-surface topography and subsurface interfaces and incorporates cubic-spline interpolations to convert the global domain into subdomains, and applies Chebyshev points in the model discretisation and computation of the spatial derivatives. Such effort makes the numerical differentiations have “spectral accuracy” inside the subdomains whose boundaries match the free-surface topography and subsurface interfaces.

2D and 3D synthetic experiments have been performed with two geological models, both having different free-surface topographies and sub-surface interfaces. The computational errors of the new approach were compared with traditional finite-difference schemes, and the results show that the sub-domain Chebyshev spectral method is superior to traditional finite-difference method in its accuracy and applicable for all of the geophysical forward modelling problems.

Most seismoelectric surveys to date have been acquired on a small scale in temperate regions. Our objective was to establish if seismoelectric data could be acquired on a large scale in an arid environment.

In April 2011, we acquired over 21,000 traces of 2D seismoelectric data at an arid site in Abu Dhabi, United Arab Emirates. The test also included seismic measurements made using WesternGeco’s UniQ singlesensor acquisition system. The source used for the acquisition was an 80,000 lb tracked Desert Explorer vibrator, the largest hydraulic Vibroseis source ever used for seismoelectric acquisition. This large source was used to attempt to overcome the low signal to noise issues inherent in seismoelectric acquisition that have been exacerbated in the past by the use of low energy sources.

We successfully acquired high quality data with coseismic signal present to the limits of our acquisition (420 m offset and 2 s record length). Our current equipment is, however, ill-suited to rapid deployment, having far too many components.

The acquisition of large seismoelectric datasets, such as that described here, enables the data to be viewed in the common receiver domain enhancing data processing and bad trace identification.