Mining Goes Digital – Mueller et al. (Eds)
© 2019 Taylor & Francis Group, London, ISBN 978-0-367-33604-2
235
Recoverable resource estimation mixing different quality of data
C.R.O. Mariz
Geovariances, Avon, France
A. Prior
Helmholtz-Zentrum Dresden-Rossendorf, Helmholtz Institute Freiberg for Resource Technology,
Freiberg, Germany
Modelling and Valuation Department, Faculty of Geoscience, Geotechnology and Mining, TU
Bergakademie, Freiberg, Germany
J. Benndorf
Faculty of Geoscience, Geotechnology and Mining, TU Bergakademie, Freiberg, Germany
ABSTRACT: Working with different data sets in the mineral resource estimation is
a common challenge to be addressed by the industry. Sampling methods, sensor devices,
measurement times along the ROM and key variables measured might differ between data sets.
These variations are reflected in the quality of each data set. Comparative exploratory data
analysis is used to verify if different data sets are sampling the same distribution. Frequently,
they show differences in the statistics and variography. This demonstrates that different sets
cannot just be merged and used in processes if these are not previously treated. One way of
integrating is to attribute a measurement error to the unreliable data set. The methodology
proposed enables a resources estimation and risk analysis by considering the variance of
measurement error calculated from the cokriging between reliable and unreliable data sets.
This paper illustrates an innovative methodology with an application to a sulphide deposit.
Keywords: Geostatistics, Mineral Resource Estimation, Variance of Measurement Error
1 INTRODUCTION
Information about the deposit is essential and scarce. Every sample is important in the process to determine the tonnage, the metal content and the mean grade in the orebody. Often,
the global data set contains samples obtained from different sampling methods, with diverse
levels of confidence, as qualities, and to estimate using these data sets is a big challenge.
Nowadays, the industry is using sensor technology in order to characterize the mine face.
The characterization of this ore presents a metrological problem of instrument calibration.
Some of these kinds of instruments compare the signal obtained from in-situ chip samples
with reference-signal databases (Van Den Boogaart, 2018). The measurement value will
depend on the subjective selection of the reference data set. This represents a source of uncertainty in the modelling. However, both data sets are still important in the estimation process.
When this kind of problem is present, a step is added to the estimation process: comparative
global and local exploratory data analysis. The goal is to compare the two data sets both at
global and local scale, verifying if they are sampling the same distribution in both scales.
The global exploratory analysis involves the calculation of statistics, histograms, variograms, and analysis of statistical profiles by slices (swath plots) for each data set for the key
variables. The local exploratory analysis involves comparison of the key variables’ behavior at
the short-scale for each data set; in other words, to compare the samples localized in the same
coordinates. However, normally the information is not known at the same location for the
© 2019 Taylor & Francis Group, London, ISBN 978-0-367-33604-2
235
Recoverable resource estimation mixing different quality of data
C.R.O. Mariz
Geovariances, Avon, France
A. Prior
Helmholtz-Zentrum Dresden-Rossendorf, Helmholtz Institute Freiberg for Resource Technology,
Freiberg, Germany
Modelling and Valuation Department, Faculty of Geoscience, Geotechnology and Mining, TU
Bergakademie, Freiberg, Germany
J. Benndorf
Faculty of Geoscience, Geotechnology and Mining, TU Bergakademie, Freiberg, Germany
ABSTRACT: Working with different data sets in the mineral resource estimation is
a common challenge to be addressed by the industry. Sampling methods, sensor devices,
measurement times along the ROM and key variables measured might differ between data sets.
These variations are reflected in the quality of each data set. Comparative exploratory data
analysis is used to verify if different data sets are sampling the same distribution. Frequently,
they show differences in the statistics and variography. This demonstrates that different sets
cannot just be merged and used in processes if these are not previously treated. One way of
integrating is to attribute a measurement error to the unreliable data set. The methodology
proposed enables a resources estimation and risk analysis by considering the variance of
measurement error calculated from the cokriging between reliable and unreliable data sets.
This paper illustrates an innovative methodology with an application to a sulphide deposit.
Keywords: Geostatistics, Mineral Resource Estimation, Variance of Measurement Error
1 INTRODUCTION
Information about the deposit is essential and scarce. Every sample is important in the process to determine the tonnage, the metal content and the mean grade in the orebody. Often,
the global data set contains samples obtained from different sampling methods, with diverse
levels of confidence, as qualities, and to estimate using these data sets is a big challenge.
Nowadays, the industry is using sensor technology in order to characterize the mine face.
The characterization of this ore presents a metrological problem of instrument calibration.
Some of these kinds of instruments compare the signal obtained from in-situ chip samples
with reference-signal databases (Van Den Boogaart, 2018). The measurement value will
depend on the subjective selection of the reference data set. This represents a source of uncertainty in the modelling. However, both data sets are still important in the estimation process.
When this kind of problem is present, a step is added to the estimation process: comparative
global and local exploratory data analysis. The goal is to compare the two data sets both at
global and local scale, verifying if they are sampling the same distribution in both scales.
The global exploratory analysis involves the calculation of statistics, histograms, variograms, and analysis of statistical profiles by slices (swath plots) for each data set for the key
variables. The local exploratory analysis involves comparison of the key variables’ behavior at
the short-scale for each data set; in other words, to compare the samples localized in the same
coordinates. However, normally the information is not known at the same location for the