Exploration Geophysics - Volume 35, Issue 2, 2004

Porous reservoirs with aligned fractures exhibit frequency-dependent seismic anisotropy because of wave-induced fluid flow between pores and fractures. To relate the elastic properties of porous rocks with aligned fractures at low frequency, we use the linear slip model of fractures and anisotropic Gassmann fluid substitution. We combine this low-frequency anisotropic Gassmann model with a dispersion relationship, based on a penny-shaped crack model of fractures, to account for frequency-dependent anisotropy.

The combined model is validated using experimental measurements of angle-dependent wave velocities of synthetic porous sandstone with aligned disc-shaped cracks. For the low-frequency anisotropic Gassmann model, the agreement between the measured and predicted velocities is reasonably good for both 5-wave velocities, but P-wave anisotropy is overestimated by approximately 25%. This quantitative difference can be explained by fluid diffusion effects occurring at the relatively high frequencies used in the experiment (100 kHz), which are not accounted for by the low-frequency assumption of anisotropic Gassmann theory. The predictions of the combined frequency-dependent model, which considers this effect, give very good agreement with measured velocities.

A comparison of first-arrival times obtained by both finite-difference and ray trace modelling demonstrates the applicability of ray tracing to shallow refraction interpretation. Most shallow seismic refraction interpretations are still achieved with conventional time-term methods and those, such as the GRM, that are derived from these methods. In complex conditions, these methods are inaccurate, as many first-arrivals originate from diffracted waves and waves that have laterally penetrated irregular refractors. These processes also affect first-arrival amplitudes, as is demonstrated with an irregular finite-difference model; nevertheless, first-arrival times are still primarily used in shallow refraction interpretation.

An interactive ray-trace modelling method, based on two-point ray tracing through a discrete, two-dimensional model with constant velocity compartments and incorporating buried diffraction sources, is used to compute accurate first-arrival-times. This method is designed for checking and refining interpretations in pre-selected areas, rather than as a production tool for routine processing of large refraction data sets. Implementation of the method on a PC using screen displays of first-arrival ray paths, and field and synthetic arrival time data, allows inappropriate arrivals caused by model deficiencies to be recognised and eliminated. Model adjustments are made directly on the screen and are guided by their effect on computed times and the agreement with field times. The improved algorithms and interactive steps of the visual interactive ray trace (VIRT) modelling eliminate instabilities that occur with many automated, ray-based inversion methods and inaccuracies arising from use of oversimplified or inaccurate initial seismic models. This interpretation method is a powerful teaching tool that readily highlights model and data deficiencies and can generate robust models for automated inversion.

The VIRT method is applied to re-interpret shallow refraction data from the London-Victoria shallow gold deposit occurring in a major fault zone. The final model obtained with VIRT agreed closely with known geology and showed that the gold mineralisation occurs in a higher-velocity region within the weathered rock above a lower-velocity region in the underlying bedrock.

A detailed gravity survey carried out by the Geological Survey of Western Australia (GSWA) used seismic grids over the Beharra Springs and Mondarra gas fields in the northern Perth Basin. In this part of the basin, seismic data are poorly imaged where the near-surface Tamala Limestone is present, whereas major structural elements are easily recognised using gravity data.

The new gravity data reveal a major transfer zone and show distinctive signatures for the Dongara Terrace, Beharra Springs Terrace, and Allanooka High. Gravity lineaments within these structural units correlate with major faults interpreted from detailed seismic data. Additional features interpreted from gravity, but not identified on seismic sections, may lead to a revision of previous seismic interpretations. Furthermore, the majority of positive residual gravity anomalies correlate with seismically mapped structural highs that coincide with known hydrocarbon fields. Other positive gravity anomalies in the area may correspond to unidentified fields.

A new type of “maximum noise fraction” transform, for multivariate data, facilitates noise reduction without requiring the samples to vary continuously, and without need to collect “noise” statistics. The algorithm specifically targets data with a hundred or more channels. Its main applications will be to hyper-spectral remote sensing data and to airborne gamma-ray spectrometry.

Ocean swells have a magnetic signal, caused by the motional induction of seawater moving in the steady main magnetic field of Earth. These signals may be sensed by a low-flying aircraft, carrying out aeromagnetic measurements over the ocean. The apparent spatial wavelength that such signals will have, when observed data are plotted out for geological purposes, can vary greatly. It will depend particularly on the relative speeds and directions of travel of the observing aircraft and the ocean swells. The apparent wavelength of the ocean swell magnetic signal cannot be less than the actual ocean swell wavelength. Generally, it is greater, and it can range up to infinity in value. For observations over continental shelves, the situation is complicated by the dependence of the swell phase-velocity on water depth, by which the swell speed generally slows as land is approached.

Constrained 3D gravity and magnetic inversion has been applied to an area of approximately 11 km × 4.5 km enclosing the Pillara Pb/Zn mine, Lennard Shelf, Western Australia. The main aim was to better define the depth to the top of the host limestone as it plunges ENE away from the mine. The starting point for inversion was a simplified geological model based on three generalized litho-stratigraphic units: shale/siltstone, limestone, and basement.

A staged inversion procedure was adopted. First, the effects of the large property contrasts were accounted for, most notably the density contrast between limestones and elastics. Subsequently, the residual gravity and magnetic data were inverted to define more subtle contrasts within the sediments.

Gravity inversion involved adjustment of the limestone contact geometry as well as limestone and shale/siltstone densities. The contact was fixed where pierced by drill holes, and a priori upper and lower bounds were imposed on the densities of the geological units. The inferred limestone contact is a strong determinant of prospectivity, both in terms of depth and in terms of fault displacements. Final stage inversion highlighted coherent intra-sedimentary density trends oriented NE and NNE; these features could be associated with mineralizing faults.

Aeromagnetic inversion defined a basement susceptibility distribution generally decreasing from the SW to the NE, reflecting the character of the TMI data. More subtle susceptibility trends attributed to the sediments may reflect the underlying structural fabric, though the most pervasive residual gravity features are not strongly developed in the residual magnetic data.

3D gravity inversion is effective on the Lennard Shelf as a means for defining the depth to limestone. The reliability of the inversion will be enhanced in areas where the gross geometry of the limestone contact and basement unconformity are constrained by sparse drilling or by seismic data, and where the densities are well known from drill core determinations or wireline logging. Magnetic inversion can play a supporting role, insofar as it defines the basement structure.

AEM data acquired with the TEMPEST system in the Boteti area, Botswana, were used to map lithologies and structures for potential fresh groundwater. The survey area is characterised by the Makgadikgadi paleolake system bordered by elevated terrain to the south. Lower-lying areas are characterised by a thick layer of Kalahari Beds with saline groundwater generally situated less than 20 m below surface. Elevated areas are generally characterised by Karoo sediments, consisting of sandstones and mudstones, at shallow depths.

Within the paleolacustrine terrain, target aquifers are freshwater zones associated with recharge pans above a saline water table. The correlation of AEM-derived conductivity-depth profiles with borehole records shows that the conductivity of Kalahari Beds is primarily a function of clay content, water saturation, and water salinity. The shallow conductivity structure from modelled AEM data outlines several resistive zones, some of them located within Boteti River alluvium beneath the present and past river channels.

In the elevated terrain, target aquifers are sandstones lying between dry alluvium or basalt, and mudstone. Basalts are characterised by very low conductivities and a strong magnetic response. The underlying sandstones and mudstones have low and high conductivities, respectively, and are offset by horst and graben structures, which are mapped in both the magnetic data and conductivity-depth sections. Resistivity iso-surfaces generated from the 3D conductivity data facilitate the spatial appreciation of the geometry of modelled structures. Favourable locations for freshwater exploitation interpreted from TEMPEST data include faults, dykes, a paleochannel incised into mudstone, and shallow fractured sandstone especially when overlain by basalt.

We derive a numerical solution of the steady-state magnetic field due to a DC current source in a layered earth model. Such a solution is critical for interpretation of magnetometric resistivity (MMR) data. Our solution is achieved by solving a boundary value problem in the spatial wavenumber domain and then transforming back to the polar spatial domain. The propagator matrix technique is used to interrelate solutions between layers. Two simplified cases are examined to promote understanding of the magnetic fields. In particular, the derived formula illustrates why surface MMR data are insensitive to ID conductivity variations and why borehole measurements are effective in finding conductivity contrasts.