Why, and How, Should Geologists Use Compositional Data Analysis/The Model

Figure 1 shows a very simplistic version of a copper mineralization associated to a zone of fractures within a granodiorite intrusive.

'''Figure 1. Ore deposit model of a copper mineralization associated to a tectonic zone within a granodiorite intrusive.'''

Table 1 shows the results of a random sampling using existing access to the top of the intrusive.

'''Table 1. Chemical composition of the sampling of the granodiorite intrusive.'''

The initial data respond to the following pre-established conditions:


 * 1) Is a close system, meaning that the sum of all the values is equal or very close to 100%
 * 2) There are no zero values present (I will deal with zero values in a separate example to avoid complicating the initial model).
 * 3) There are no statistical outliers (hurricane values) present.
 * 4) There is a big difference between the concentRations of the major oxides with respect to the trace elements.
 * 5) As shown in Figs. 2-4, there are embedded positive and significant correlation between Cu and As, Ni and Co, and MgO and K2O.
 * 6) Finally, I introduced significant negative correlations between K2O and CaO (Fig. 5), SiO2 and Al2O3 (Fig 6), K2O and Cu (Fig. 7), Na2O and K2O (Fig. 8), and Co with Cu (Fig 9).



According to these embedded conditions, any correlation analysis must give us two coefficients (Range Correlation Coefficient or RCC):


 * Equation 1: Range Correlation Coefficient type A for the initial data, according to the embedded correlations.


 * Equation 2: Range Correlation Coefficient type B for the initial data, according to the embedded correlations.