The Application Of Spatial Factor Analysis To Unconditional Simulations With Implications For Mineral Exploration

Spatial auto- and crosscorrelation assessment of variables can be carried out using quadratic function estimates. From the function estimates, multivariate relationships of the variables can be determined in the form of spatial factors. The calculation of the spatial factors requires computation of the corresponding amplitude vectors from the eigenvalue solution. This vector reflects the relative amplitudes by which the variables follow the spatial factors. Instability of some of the eigenvalue solutions requires that caution be used in interpreting the resulting factor pat- terns. A measure of the predictive power of the spatial factors can be determined from the auto- correlation coefficients and squared multiple correlation co-efficients which indicate which of the variables are significant in any given factor. The use of unconditional simulations provides a basis by which geological variables of varying spatial ranges can be modelled using the spatial factor technique. Simulations were generated for a variety of ranges and variables were created that form single or multiple combinations of the spatial patterns. The success of the spatial factor technique depends on the amount of associated noise, and the size of the neighbourhood relative to the spatial range of the variable. The predictive power of the components is dependent on these two factors. Problems with collinearity of variables can result in unrealistic eigenvalues and result in the failure of the method. Application of this method to spatial data in mineral exploration can assist in outlining spatially continuous zones of alteration associated with mineralization.