Integrating Data Of Different Precision

In the mining industry, it is common to have data collected in various formats. In an open pit mining operation, at the early stages of exploration samples are obtained from diamond drillholes. During the production stage, samples are collected from blastholes. The last is more numerous but less precise. From a geostatistical view, this dissimilarity has to be considered to integrate the two sources of information. To combine this different data, three methodologies were investigated: simple kriging with local varying mean, cokriging and ordinary kriging. The exhaustive Walker Lake data set was used and it is considered the source to obtain the true grades. Initially, samples were obtained from the exhaustive data set at a regular spacing grid of 20 x 20 meters. Next, samples were obtained again from the exhaustive dataset at a regular spacing of 5 x 5 meters and a relative error of 25% was added. Then, both data were used to estimate blocks using the three methods mentioned. The grade tonnage curves were compared with the true block grade distribution. Moreover, the block misclassification was evaluated. The results showed that ordinary kriging produced estimates closer to the true block grade distribution and reduced the block misclassification.