Geophysical data acquisition, processing and interpretation - Geophysics for the Mineral Exploration Geoscientist - page 38

imperfectly addressed if information about geological/
anomaly trends is known or can be inferred. Some gridd-
ing algorithms include a user-de ned directional bias,
referred to as trend enhancement. This has the disadvan-
tage of assuming a consistent trend across the entire survey
area, but can be highly effective if this assumption is valid
( Fig. 2.17c ). An alternative approach is to use the data to
estimate between-line values and use these in the gridding
process. The most effective remedy is to measure gradients
of the field in the across-line direction, which act as con-
straints on the permissible interpolated values and enhance
across-line trends, (O

that the various datasets have adequate overlap, especially
if survey parameters are different.

2.7.4 Enhancement of data

Data enhancement techniques involve numerical algo-
rithms operating on the survey data. We recognise three
basic kinds of algorithms: the combining of repeat readings
in a process known as stacking; the comparison of two
measurements by computing their ratio; and more math-
ematically sophisticated manipulations of the data for spe-
cific purposes which we refer to as filtering. Note that the
term filtering is often loosely used to describe any process
that alters the data in some way.
Enhancements are applied to geophysical data for three
basic purposes:

Connell et al., 2005 ) .
There is an extensive literature on gridding methods, for
example Braile ( 1978 ) and Li and Göetze ( 1999 ). Different
methods perform better in different circumstances; but
both inverse-square distance with minimum curvature,
and spline algorithms are widely used for geophysical data.

'

To enhance the signal-to-noise ratio (SNR) of the data.
This involves identifying a characteristic, or characteris-
tics, which are not shared by both the signal and the
noise, and then using this as a basis for attenuating the
noise component of the data. Characteristics that can be
used include randomness,

￿

2.7.3 Merging of datasets

When several datasets are available from an area it can be
useful to merge or stitch these into a single dataset for
processing and analysis. The requirement is to form a
coherent and

spatial and/or temporal

coherence, trend, periodicity and wavelength.

composite dataset from individual
surveys that often have quite different survey parameters.
This may not be achievable in practice because the differ-
ent survey parameters will produce datasets with different
wavelength content (see Section 2.6.1 ), and similar geology
will not produce identical responses in the different
datasets.
Data are usually merged in their gridded form.
A description of one kind of grid stitching algorithm is
provided by Cheesman et al.( 1998 ) . Accounting for the
range of wavelengths in the datasets is a fundamental
aspect of the process. These may vary for each survey in
the overlap zone owing to differences in survey size, shape,
acquisition parameters and noise levels. Also important is
the extent of the overlap between the surveys, with the
amount of overlap and the nature of the geophysical eld
in the overlap zone and at the joins strongly in uencing
the result. The more similar the datasets in the overlap
zone, and the larger their overlap, the better the results are
likely to be (particularly in terms of longer wavelengths).
When there is no overlap it will be necessary to acquire
additional data to link the surveys being merged. For this
reason, it is important that any new survey extends far
enough into the area of the existing data coverage to ensure

' seamless ' ￿ Most types of geophysical data can also be processed to

enhance characteristics of the signal that are considered
particularly signi cant. For example, filtering may
enhance short-wavelength responses originating in the
near-surface, or longer wavelengths related to deeper
features, or may compute the various derivatives (gradi-
ents), which are greatest near physical-property contacts
etc. (see Fig. 2.3 ). Ratioing may allow regions with par-
ticular combinations of responses to be recognised, for
example anomalously high values in one dataset and low
values in another.

For completeness we mention a third form of enhance-
ment that is based on the principle that some types of
data can be transformed to produce a dataset that would
have been measured with another type of sensor, or with
different survey parameters. For example, magnetic and
gravity data can be transformed into the equivalent data
that would have been obtained if the survey were con-
ducted higher (or lower) above the ground. This not only
emphasises features of interest, it also enables data
acquired with different survey parameters to be inte-
grated and merged (see Section 3.7.3.2 ). These kinds of
enhancement are described in our descriptions of each
geophysical method.

￿