Through the SATS system, the resource model can extrapolate outside the boundaries of the input dataset. Therefore, together with our traditional conditional simulation-like testing methodologies and first principles uncertainty analysis of each block grade prediction, the SATS system can produce potential mineralization areas with high confidence. Through our drillhole simulator tool, infrastructure and economic conditions can be used as constraints, thereby producing drill targets (both origins and orientations) that will most optimally intersect the target mineralization clusters.
For marginal, high volume assets, the SATS system can more accurately predict the exact grade of any given block and can correctly sort the extracted blocks into the correct grade buckets to ensure optimal stockpiling. The SATS models' added benefit is the highly precise ore/waste classification under dynamic economic conditions while also ensuring that the average grade of high grade stockpiles is increased to ensure maximum value captured when sent to the mill.
The SATS system can model all types of structured data (i.e. x,y,z and a measurement); therefore, modelling metallurgical parameters like ore type and hardness can be estimated much like a resource estimation. Using multiple types of inputs and finding non-linear correlations between data types, less than ideal metallurgical datasets (e.g. <50% of drillhole samples have hardness measurements) can be estimated on a block-by-block basis. Due to the inherent quality control within the SATS system, relevant data sources can be identified easily to improve the parameter's prediction.
Polymetallic assets benefit from the SATS system due to the multiple element and data types that can be used as inputs. Besides modelling just the primary commodity, the SATS system can model secondary commodities and leverage the non-linear correlations between the different elements to improve the grade prediction accuracy of all the elements being modelled. For elements that aren't assayed consistently, using other correlative elements can be used to "fill in the blanks" where data is lacking.
Via more traditional conditional simulation-like techniques and first principles identification and propagation of error sources through the prediction, every block has a confidence interval that shows the best, worst, and most likely case grade prediction. Therefore, an activation map showing which data points help or hinder model performance is useful for resolving a resource (i.e. converting to Measured/Indicated) and can address risk by indicating areas of the asset that would benefit from additional short-range drilling.