Smoothing effect, conditional bias and recoverable reserves

"Although many factors play roles in the estimation of recoverable reserves, smoothing effect and conditional bias are considered major geostatistical issues in the estimation. This paper introduces measures for smoothing effect and conditional bias, and provides insights into their relation to the estimation of recoverable reserves. Both smoothing effect and conditional bias are intrinsic to the information effect and kriging estimators. For a given level of drilling information, however, it is possible to identify the optimal search radius that minimizes conditional bias. Furthermore, it is also feasible to define variance correction factor for reduction of the smoothing effect from exploration block models. Both of these components are key to the accurate estimation of recoverable reserves. IntroductionThe accurate estimation of recoverable reserves is critical to decision making in different phases of feasibility study and mine planning. Much attention for improving the methodology of estimation has been focussed on the effect of the change-of-support (Parker, 1980; Dagdelen, 1996). For example, it is essential to make a volume- variance correction to the estimate of multiple indicator kriging (Hoerger, 1992). When the size of block, b, in the geostatistical model differs from that of selective mining unit (SMU), v, a volume-variance correction is warranted for the prediction of recoverable reserves.Most mining plans and decisions are made prior to the availability of high density blast-hole samples. In most mining operations, the true SMU grade Zv is unknown at the time of selection. Instead, the selection is based on the estimate Z v**derived from high density blast-hole samples within the deposit G. Nevertheless, the estimate Z v * of recoverable reserves in the stage of mine planning is derived from exploration data. At the end, the production result must be reconciled to the estimate Z v * . In other words, the selection is based on Z v** from blast-hole samples, while the final production figures are reconciled to Z v * from exploration drilling. For a given preselected cutoff grade z0, the actual mining panels will consist of the units having the estimated grade Z v** =z0 rather than the true grade Zv =z0. The difference between Z v** and Zv is usually of less concern."