Page 15 - Geochemical Anomaly and Mineral Prospectivity Mapping in GIS

10 Chapter 1

The traditional methods applied in modeling of uni-element geochemical anomalies
include (Levinson, 1974; Govett et al., 1975; Rose et al., 1979; Sinclair, 1983; Howarth
and Sinding-Larsen, 1983): (a) comparison of data from the literature; (b) comparison of
data with results of an orientation geochemical survey; (c) graphical discrimination from
a data histogram; (d) calculation of threshold as the sum of the mean and some multiples
of the standard deviation of data; (e) plotting cumulative frequency distributions of the
data and then partitioning the data into background and anomalous populations; (f) (e)
constructing normal probability plots of the data and then partitioning the data into
background and anomalous populations and (f) recognition of clusters of anomalous
samples when data are plotted on a map.
The traditional methods applied in modeling of multi-element geochemical
anomalies include (Rose et al., 1979; Howarth and Sinding-Larsen, 1983): (a) stacking
uni-element anomaly maps of the same scale on top of each other on a light table and
then outlining areas with multiple intersections of uni-element anomalies; (b)
determining inter-element correlations, usually by calculating Pearson correlation
coefficient; (c) recognising and quantifying multi-element associations based on their
correlations; and (d) mapping of quantified multi-element association scores.
Applications of certain methods to model multi-element geochemical anomalies
invariably require computer processing because such techniques are prohibitively
difficult and time-consuming to perform manually. In addition, conventional methods for
recognising and mapping multi-element associations vary depending on whether a-priori
knowledge is available or not about certain controls on geochemical variations.
Principal components analysis and cluster analysis are conventional methods in
identifying and mapping multi-element associations without a-priori information about
controls on geochemical variations or which samples were collected at or near
mineralised zones. In more recent times, fuzzy cluster analysis based on the fuzzy set
theory (Zadeh, 1965) has been demonstrated to be more advantageous than the
conventional cluster analysis in identifying boundaries of anomalous multi-element
clusters and quantifying degrees of membership of samples to every fuzzy cluster (Yu
and Xie, 1985; Vriend et al., 1988; Kramar, 1995; Rantitsch, 2000).
In situations where some controls on geochemical variations are known a-priori, a
simple conventional approach to recognise and map significant geochemical anomalies
is to apply element ratios (Plimer and Elliot, 1979; Brand, 1999; Garrett and Lalor,
2005). When scavenging effects of Fe-Mn oxides on variations of certain elements are
known a-priori or recognised during the analysis of inter-element correlations, a usual
approach is to apply regression analysis to estimate concentrations of certain elements as
a function of Fe and Mn contents in order to interpret if geochemical residuals depict
anomalous patterns or not. When it is known a-priori that some samples were collected
at or near mineralised zones, such samples can be used as a training set in discriminant
analysis to determine if the other samples are associated with mineralisation or not.
Alternatively, logistic regression analysis can be performed by using binary data [0,1] of
mineral deposit occurrence as target variable and the multi-element data as predictor
variables to derive predicted mineral deposit occurrence scores for each sample.