M2RC third year development program

Given the success of M2RC two first years, partners were willing to pursue the consortium over 2008. The development axes for this third year are the following:

• Development of a plugin to integrate a new histogram modeling with a dispersion of the empirical distribution.

• Development of geostatistical tools for domaining within Isatis. For the first 6 months, we will begin with the two following points:
   Discussing with the 3rd year partners the actual practice of domaining in their companies in order to write a document resuming the different approaches and to analyse them with the geostatistician point of view. This will be used to select the estimation and simulation methods that will be developed later.
   Developping characterization tools in order to describe the nature and continuity of the boundaries of domains, and to analyze contacts and border effects.

This international project is open to all companies. If you wish to joint it or get more information about it, please contact Julien Tan.

Why Domaining?

Domaining is an important issue, and a major control for the volume of resources. It aims at partitioning the orebody into domains that have different characteristics (in term of lithology, mineralogy, alteration, concentration...) and that will be consequently treated distinctly during the exploitation.

It is important that domaining takes into account all geological information available, but one big problem is that delineating domains (at drillholes, on sections, between sections...) and updating these is often a subjective, time-consuming operation. The interest of geostatistics in domaining is:
 the possibility of a statistical control of the underlying hypotheses (characterization of domains);
 the use of geostatistical methods for estimation or simulation, if appropriate, and the resulting increased automatization;
 the possibility to provide a measure of uncertainty (variances, simulations).

The characterization of the domains begins by the criteria used to define these domains (lithology, grades...) or more directly their boundaries (eg faults). However are also important the properties of the frontiers, or related to the frontiers, in particular:
 the limit of a domain, or the contact between two domains, can be “hard”, that is, sharp, easily identified eg along holes (change of lithofacies, contact with a major fault), or on the contrary “soft”, fuzzy (transition between rich and poor, native and alterated...);
 the continuity in space, eg between holes, can be well established, or on the contrary subject to high uncertainty;
 the mineralization within a domain may be affected or not by the proximity of the frontiers (border effects, with grades that are either lower, or higher);
 the mineralization across the contact between two domains may exhibit some continuity or none.

Geostatistical estimation and simulation methods that can be useful for domaining are various:
 direct indicator based method (Indicator Kriging, Sequential Indicator Simulation), describing the volume of a domain by its indicator;
 estimation and simulation of the top or bottom limit surface represented by a regionalized variable z(x,y) (this supposes that each vertical line crosses the limit once, and can be generalized with another direction than vertical);
 isoline of a continuous variable to be estimated or simulated, eg cutoff grade limit, with the indicator of the domain being 1 above the cutoff;
 isoline of a hidden continuous variable to be estimated or simulated (truncated gaussian, iso-potential or implicit function).

Such methods may be used while considering additional elements e.g:
 cross-boundaries neighborhood;
 considering several domain indicators when related;
 considering categorical and continuous variables together (e.g domain indicator + grade).