Geovariances consultant in resource evaluation is to present a paper in partnership with Comilog
"Sensitivity Study of the Estimation Variance Approximation of a Quotient – Comparison with Conditional Simulations in the Mn Deposit of Bangombé (Gabon)"
It is common geostatistical knowledge that grade measured on uneven sample support sizes is not additive. This is the case for example in thin orebodies, where grades of intercepts with varying lengths cannot be estimated directly. The necessity to work with additive variables (i.e. variables that maintain their physical meaning through linear averaging) imposes the estimation to be performed on thickness (T) and the grade accumulation (A). The estimation of these variables can be performed by independent kriging or co-kriging, the grade (G) estimate is simply obtained as the ratio of the two input estimates (A/T). As both estimates for A and T are geostatistically based their precision can be characterized by their kriging variances. The grade estimate however does not come directly with its own kriging variance from the solved kriging system.