6th SAGA Biennial Conference and Exhibition

Anglogold has used 3D seismic surveys to assist in planning operations over several gold
mines in the Archaean Witwatersrand seidmentaiy basin. Mine planning benefits from the
ability of 3D seismic data to accurately image the subsurface, preventing mis-placed
development from incorrect geological models. The WUDL survey is the largest mineral 3D
survey in southern Africa at nearly 300km2. It imaged much of the Carletonville Goldfield~
located 80km southwest of Johannesburg, from Deelkraal G.M. in the west, to Driefontein
G.M. in the east. Results included the geological structure of economic targets such as the
Ventersdorp Contact Reef (VCR), a coherent model for the development of the region, and
some indication of the prospectivity of unmined ground.

The Nkomati MSB Mi, a joint venture between Anglovaal Mining and Anglo American,
exploits a magmatic Ni-Cu-Co-PGE massive sulphide deposit associated with the Uitkomst
Complex, in Mpumalanga, South Africa. The Uitkomst Complex is a linear, northwest-southeast
trending, layered mafic-ultramafic body, intruded into the Transvaal Sediments. It hosts
disseminated and massive sulphide mineralisation. An exploration program was undertaken to
search for further zones of massive sulphide mineralisation associated with the complex.

The use of long reach, sub-horizontal wells in the development of structurally complex
oil and gas fields has become an increasingly common practice. This paper describes
some ways in which the reservoir related risks associated with drilling such wells can
be reduced using a variety of geological and geophysical techniques. The E-M gas
field, offshore South Africa is presented as a case study.

The National Water Master Plan study, NWMP, (DWA, 1991 and 1992) proposed that
the future water supply to the south-east and central Botswana should use sulface water
from dams along the Motloutse and Shashe rivers. To convey the water to the demand
areas, the North South Carrier (NSC) was proposed. The NSC runs through the Palla
Road and Khurutshe areas which have major groundwater potential which are also being
developed to be used as a strategic backup supply to these dams. The groundwater
resources are encountered in the Ntane Sandstone in Upper Karoo formations.

Separation filtering of magnetic data can be extended from simple regional/residual
separation to the general multilayer case it is theoretically possible to design filters to
deconvolve the effects of causative sources occurring around a particular level. The problem
posed is non-linear since the field equations contain terms relating to source width and
thickness as well as source depth.
Separation filtering becomes very difficult when there is considerable overlap in the spectra
of individual depth ensembles. The degree of separation achieved depends on the depth
difference and the spectral b/B ratio, the ratio of the amplitudes of the shallow and deep
ensembles. A high b/B ratio is needed if the fîltering objective is to deconvolve the effects of
shallow sources with minimum contamination by deeper sources.

Preliminary analyses of seismic records of mining-induced earthquakes from the
Kaapvaal Craton experiment indicate that these data are of sufficiently high quality
and large quantity to enable significant improvements of the crustal velocity structure
of southern Africa through the routine identification of seismic phases. Principal
event parameters, such as origin time, location, depth and local magnitude, are
collated from the bulletins published by the Council for Geoscience (Graham et al.,
1997) and from individual mine databases from gold mining companies throughout
South Africa. The waveform data for events with a magnitude greater than 2.0 are
extracted from the Kaapvaal database. This database was acquired through the
deployment of a 54-station, SO-site network of broadband seismic stations distributed
across the southern African region. as part of the multinational project “The anatomy
of an Archean craton” (Carlson et al., 1996). The restricted time of this deployment
is from April 1997 until June 1999. Careful identification of phases across and
around Kaapvaal craton are fundamental in deriving more representative velocity
models of the crust and upper mantle of the southern African region. Results from
other current seismic research, including receiver function and two-station surface
wave dispersion studies, will also be used to assist refinement of the S wave velocity
models for the crust and upper mantle (James et al, 1998). The event location data
provided by the mines are accurate to a few hundred meters, providing a unique
opportunity of combining very accurate locations with waveform data from the
regional Kaapvaal network. This will enable an improved velocity model of the crust
and upper mantle to be calculated and used in the future for locating events.

Analysis of the refracted arrivals on a seismic reflection profile recorded along the wall of a
tunnel al an iron mine near Thabazimbi, South Africa, shows significant variation in P-wave
velocity in dolomite in and away from the de-stressed zone. Very low seismic velocities away
from the tunnel wall are associated with alcoves or ‘cubbies' involving offsets in the wall of up
to 10 m. Large variations in seismic velocity resolved over distances less than 15 m with
signals of wavelength around 6-9 m are attributed to variations in the fracture systems in the
de-stressed zone. The results show that refraction static processing of ‘in-mine’ seismic
reflection profiles is as important as in surface surveys, since the frequency content of deep
mine seismic energy is higher than on surface, and can assist in analysis of stress regimes
within in-mine developments.

A new airborne electromagnetic system has been developed by the CRC for Australian
Mineral Exploration Technologies and is operated by World Geoscience Corporation.
TEMPEST operates in the “INPUT” configuration with a towed bird and a transmitter
located on the aircraft. It utilizes a 25 Hz square-wave transmitter with a 50% duty cycle
and a variable switching ramp. The system has been designed to measure a wide
bandwidth (25Hz - 3ZSkRz) EM response for both geological mapping and recognition
of conductive targets.

In time-domain systems, the traditional sensor used in transient electromagnetic (EM) systems
is an induction coil. This sensor measures a voltage response proportional to the time rate of
change of the magnetic field in the EM bandwidth. By simply integrating the digitized output
voltage from the induction coil, it is possible to obtain an indirect measurement of the magnetic
field in the same bandwidth.
The simple integration methodology is validated by showing that there is good agreement
between: a) synthetic voltage data integrated to a magnetic field, and b) synthetic magnetic-field
data calculated directly. Further experimental work shows that airborne EM magnetic-field data
collected with an induction coil correspond quite closely to that measured using a SQUID
magnetometer.
Comparison of measured voltage and magnetic-field data show that the two sets of profiles have
quite different characteristics. The magnetic-field data is better for identifying, discriminating
and interpreting good conductors, while suppressing the less conductive targets.

Contamination of primary reflections by remnant multiple reflections continues to
degrade the results of rock property estimation from Amplitude Versus Offset (A VO)
analysis of seismic data. Conventional stacked seismic sections benefit from the
application of near and far trace tapering prior to stacking, but this would not seem
applicable in AVO analysis. Presented here is a means whereby offset weighting
schemes can be incorporated into A VO analyses, resulting in improved multiple
suppression. The weighting is inserted into the Aki and Richards approximation to the
Zoeppritz equations, and thus can be used with any method which attempts to
minimise the error between the observed and predicted seismic amplitudes. The
incorporation of this weighting into the GEOSTACK method of Smith and Gidlow is
demonstrated, together with a real data example showing the multiple suppression
achieved by this approach. The combination of a suitable multiple suppressing
weighting scheme and a least squares minimisation approach produces a more robust
method of AVO processing in the presence of multiples or any other noise.