Recoverable resource estimation mixing different quality of data

Working with different databases in the process of mineral resource estimation is a common challenge to be addressed by the industry. These are produced from different sampling methods, having therefore different quality, being obtained in different times of the run of the mine and even measuring different key variables.
Comparative exploratory data analysis, in a global and local scale, are used to verify if different databases are sampling the same distribution. Frequently, the results show differences in the statistics, for instance: in the distribution and in the experimental variography. This demonstrates that different databases cannot be just merged and used in estimation and simulation processes if these are not previously treated.
One way of integrating the different databases with different qualities into the in-situ resource estimation process is to attribute a variance of measurement error to the inexact dataset (low quality samples) However, the estimation of recoverable resources and risk analyzes remains to be verified. The methodology proposed, enables an estimation and risk analysis of the resources of interest by considering the variance of measurement error calculated in-situ of the estimation process:
1. Transforming raw variables od different databases into their Gaussian equivalent through Gaussian Anamorphosis and the calculation of experimental covariances and cross-covariances on Gaussian transforms;
2. Point-wise co-Kriging from Gaussian exact assay (high quality samples) and inexact assays;
3. Block-wise kriging from the Gaussian exact assay and the Gaussian pseudo exact values at inexact location of the main variables with a variance of the measurement error; and
4. Turning-bands conditional simulations with variance of the measurement error from Gaussian variables kriged (1) and with a local mean (2) are applied.
5. Back transformation of the Gaussian variables into raw variables and re-blocking of these to calculate the grade tonnage variables.
This paper illustrates an innovative methodology with two applications in a polymetallic massive suphide deposit and bauxite deposit.