Geoscience Reference
In-Depth Information
would be tuned to match production data. Care should be
taken when considering the amount and location of the pro-
duction data, ensuring that it is relevant to the future mining
predicted by the resource model.
If no production data is available, then validations and
some subjective criteria need to be used to define the kriging
plans. The process is still iterative, but issues like reproduc-
tion of global declustered means; smoothing and dilution;
behavior of alternative models; past resource models; grade
profiles across contacts; and so on. These are all part of the
resource model validation process discussed in Chap. 11.
The rationale for the kriging plan adopted should be clearly
stated, as well as the “history” of the different estimation
runs and iterations possibly performed. The model should be
thoroughly checked for all variables involved, and according
to the details described in Chap. 11. Comparisons with prior
models and alternative models should also be made, and all
differences explained, whether due to incremental data, or
methodological differences. Potential risk areas and global
uncertainty measures should be included, in addition to the
standard resource classification scheme. The documentation
trail should be complete and thorough, with assumptions and
perceived limitations of the model clearly stated. Recom-
mendations for future work, presented as part of a suggested
risk mitigation plan, should be included.
Best practice includes the use of alternate models to
check the results of the intended final resource model. All
issues relating to the choice of estimation method, the pa-
rameters used, and the data selection adopted should be
clearly stated and justified. All possible production and cal-
ibration data should be used to indicate whether the model
is performing as expected, possibly including simulation
models to calibrate the recoverable resource model. The
validation of the estimates should include a validation of
the calibration data. The model should be fully diluted, and
should quantitatively describe the amount of the different
types of dilution included. Checking and validation, as de-
tailed in Chap. 11, should be fully implemented and docu-
mented. Model reporting should be complete, including all
zones, domains, variables, and aspects of the model that
are considered relevant. The documentation of the model
should also be complete, and should include the best avail-
able visualization tools for the benefit of the uninitiated. All
risk issues should be dealt with in detail, and if possible,
quantified. This will generally require performing one or
more simulation studies.
8.6
Summary of Minimum, Good and Best
Practices
In addition to some general comments on presentation and
block model reporting, this section presents the details of
what is considered minimum, good, and best practices in ob-
taining and handling ore resource models.
The estimated values obtained should be thoroughly
checked and validated, as detailed in Chap. 11, where a sum-
mary of minimum, good, and best practices for model check-
ing and validation is also presented. These checks should re-
sult in, among other comments, a statement about whether
the model can be considered “recoverable” or fully diluted.
Assuming that all checks have shown that the model is ac-
ceptable, other aspects need to be considered.
One important consideration is the reporting and trans-
mittal of the resource model to mine planning. It is important
to communicate the technical characteristics of the model,
the sources and amount of dilution included, and the model-
ing methodology for all variables.
The minimum practice consists of documented param-
eters within each estimation domain, including specific krig-
ing plans and using specific variogram models, according
to the each domain's geologic and statistical characteristics.
Clear documentation and justification of the kriging method
chosen should be presented, as well as basic checks, see cor-
responding section in Chap. 11. The reporting of the model
should clearly state the ore resource estimates, globally, by
mining phases, benches, or whichever other volume may
be appropriate. Grade-tonnage curves should be clear, and
should include all relevant economic cutoffs and estimated
variables. Subjective expressions of uncertainty are appro-
priate, as part of the limitations of the model. A report doc-
umenting in detail all relevant aspects should be prepared,
including recommendations for improvements.
In addition to the above, good practice requires a more
detailed justification of the estimates. Typically a degree of
calibration is also required, if possible according to past pro-
duction. At a minimum, the resource model should be able to
reproduce past production to within a reasonable tolerance.
8.7 Exercises
The objective of this exercise is to review the theory and
practice of kriging. Some specific (geo)statistical software
may be required. The functionality may be available in
different public domain or commercial software. Please ac-
quire the required software before beginning the exercise.
The data files are available for download from the author's
website—a search engine will reveal the location.
8.7.1
Part One: Kriging Theory
Question 1:
Derive the estimation variance in terms of the
covariance. Explain where the assumption of
stationarity comes into the derivation.
