Page 78 - Geochemical Anomaly and Mineral Prospectivity Mapping in GIS

Exploratory Analysis of Geochemical Anomalies 77

TABLE 3-V

Correlations coefficients of the log e -transformed uni-element data exclusive of samples with
censored As values (n=95), Aroroy district (Philippines).

Cu Zn Ni Co Mn
Zn 0.342
Ni 0.560 0.537
Co 0.396 0.589 0.754
Mn 0.268 0.730 0.440 0.661
As 0.341 0.414 0.496 0.391 0.462
All coefficients are significant at the 0.01 probability level (2-tailed test).

scatterplots that are optimal for visual analysis of inter-element relationships. The
scatterplots in Fig. 3-18D indicate that all the elements under study have positive
relationships with each other and there is no obvious presence of more than one
population, except for a small cluster of high Mn values and low As values in the Mn-As
plot. This small cluster in the Mn-As plot pertains to 13 samples in areas underlain by
diorite with As concentrations above detection limit (see Table 3-II or 3-IV)).
Visual interpretation of a scatterplot can be aided by estimation of correlation
coefficients and of covariance values between two uni-element data sets to obtain
impressions about, respectively, inter-relation of data values and mutual variability of
values. Estimates of correlation coefficients and covariance values are affected by data
form and presence of censored values, outliers and multiple populations. The correlation
coefficients and covariance values of element pairs corresponding to the data scatterplots
in Fig. 3-18D are shown in Tables 3-V and 3-VI, respectively. Statistically significant
positive correlations exist between all pairs of the log e-transformed uni-element data sets
exclusive of samples with censored As values. The strongest correlation is between Ni
and Co, followed by inter-element correlations with Mn. These correlations suggest
controls by either lithology or scavenging by Mn-oxides. From the estimated correlation
coefficients (Table 3-V), there are no obvious inter-element relationships reflecting
presence of mineralisation. Estimates of covariance values for each pair of the uni-

TABLE 3-VI

Covariance matrix of the log e -transformed uni-element data exclusive of samples with censored As
values (n=95), Aroroy district (Philippines).

Cu Zn Ni Co Mn
Zn 0.091
Ni 0.188 0.135
Co 0.104 0.115 0.187
Mn 0.066 0.134 0.103 0.120
As 0.237 0.214 0.326 0.200 0.222