Selected Topics for Further Research - Geomathematics: Theoretical Foundations, Applications and Future Developments - page 473

Phases

Star Kimberlite - Log-centerd Values

mJF IJf

eJF Pense Cantuar

2.5 3.0 3.5 -2.4 -2.0 -1.6

Yb

P

La

Zr

-8.4 -8.0 -7.6 -3.0 -2.6 -2.2 -1.8

Fig. 12.5 Scatter plot matrix of selected elements that represent lithospheric mantle contamina-
tion from the five eruptive phases of the Star kimberlites. The data have been log-centred
transformed. Early Joli Fou observations are shown as solid circles, mid Joli Fou as triangles,
late Joli Fou as vertical crosses, Pense as diagonal crosses and Cantuar as diamonds (Source:
Grunsky and Kjarsgaard 2008 , Fig. 3)

Figure 12.6 shows pair plots of four elements (Yb, P, La and Zr) selected to
represent kimberlite magma composition variability. The distinctive trends of
relative enrichment of these elements indicate progressive fractionation of olivine
plus spinel and decreased lithospheric mantle contamination. The evaluation of
lithogeochemical data requires transformation from the constant-sum composi-
tional data space to real number space through the application of a log-centered
transformation. Principal component analysis was applied to the log-centred data.
Figure 12.7 shows the resulting screeplot of ordered eigenvalues and their
corresponding significance for the Star dataset. The first two eigenvalues contribute
34.6 and 32.0 % of total variance of the data. Scores of observations and variables