Date of Award


Document Type



Edith Cowan University

Degree Name

Master of Science


School of Computer, lnformation and Mathematical Sciences


Faculty of Science, Technology and Engineering

First Supervisor

Geoff Comber

Second Supervisor

Dr LynBloom


The nature of spatial distributions of geological variables such as ore grades is of primary concern when modelling ore bodies and mineral resources. The aim of any mineral resource evaluation process is to determine the location, extent, volume and average grade of that resource by a trade off between maximum confidence in the results and minimum sampling effort. The principal aim of almost every geostatistical modelling process is to predict the spatial variation of one or more geological variables in order to estimate values of those variables at locations that have not been sampled. From the spatial analysis of these variables, in conjunction with the physical geology of the region of interest, the location, extent and volume, or series of discrete volumes, whose average ore grade exceeds a specific ore grade cut off value determined' by economic parameters can be determined, Of interest are not only the volume and average grade of the material but also the degree of uncertainty associated with each of these. Geostatistics currently provides many methods of assessing spatial variability. Fractal dimensions also give us a measure of spatial variability and have been found to model many natural phenomenon successfully (Mandelbrot 1983, Burrough 1981), but until now fractal modelling techniques have not been able to match the versatility and accuracy of geostatistical methods. Fractal ideas and use of the fractal dimension may in certain cases provide a better understanding of the way in which spatial variability manifests itself in geostatistical situations. This research will propose and investigate a new application of fractal simulation methods to spatial variability and spatial interpolation techniques as they relate to ore body modelling. The results show some advantages over existing techniques of geostatistical simulation.