During ore body modeling normally additional samples are required to improve resources evaluation and mining planning. At this stage one rises the question about where additional samples should be located. Regular sampling pattern seems to be a reasonable choice, since all the regions within the deposit receive the same number of samples without a clear clustering. Nevertheless, values for certain attributes (grades for instance) from a mineral deposits may behave more erratic in some regions than others. Consequently, by adding samples to these regions could bring more benefit for reducing the uncertainty related to a transfer function such as Net Present Value (NPV), instead of adding samples to location of low grade uncertainty. Thus, for a given number of additional data, the test of a regular pattern and a pattern which add samples on areas of high uncertainty of the attribute of interest is worth. This paper presents an algorithm that allows the construction of these two patterns, automatically, on a software commonly used in the mining industry. The algorithm leads to a fast analysis on the efficiency of the two patterns in reducing the uncertainty associated with a transfer function.
展开▼