Geoscience Reference
In-Depth Information
information, and in a format that non-specialists and industry
professionals are less familiar with. Visualization tools are
useful for qualitative demonstrations of the models, but typi-
cally the reporting of the simulation models depends on their
objective, and the risk assessment they provide.
The description of joint uncertainty is also a key element
that should be considered. Examples are shown of what is
meant by joint uncertainty, and its possible applications and
consequences on downstream work.
Another aspect that requires some detailed discussion is
the issue of ranking the individual simulations. The discus-
sion includes criteria for deciding if there is a need for rank-
ing, the methods that can be used to rank individual simula-
tions¸ and the tradeoffs that typically are made.
Minimum practice includes simulating specific domains,
typically the same estimation (simulation) domains defined
through the combination of geology and statistical analyses.
A clear statement of objectives, the justification of the simu-
lation method chosen, the related risk assessment of inter-
est, and the implementation-specific parameters should be
clearly stated and demonstrated when appropriate. Some of
the simulations should be carefully validated to the extent
described in detail in Chap. 11, while the overall resulting
uncertainty model should be checked against other models,
including estimated grade models and known production
values, if available. The reporting and visualization of the
simulation models are critical, and they should be related al-
ways to the original objective of the conditional simulation
study. Commonly, only a subset of the simulations obtained
is used in the risk assessment study, and so all decisions re-
lated to individual simulation ranking should be presented
and justified. The simulation work should be documented
with some detail, particularly with respect to validation, ap-
plications, and perceived model limitations.
Good practice requires, in addition to the above, a more
detailed justification of the simulation model. Calibrations
on a smaller subset of the overall volume, significantly more
iterations and comparisons before an individual simulation is
considered acceptable, and much more validation and check-
ing of the resulting uncertainty model than before. Also, the
assessment on the geologic model should be done (simula-
tion of categorical variables) and introduced into the over-
all resource simulation. Comparison with past production
should be done in detail whenever possible. The full report-
ing of all relevant aspects of the models and the correspond-
ing risk assessment is required, as well as full and detailed
documentation. It is important to emphasize the correct use
of the simulation models, their limitations, and possible fu-
ture improvements to the work.
Best practice consists of full implementation of the
techniques available, and the use of alternative models to
verify their relevancy. All potential sources of uncertainty
in an ore resource model should be investigated, including
the uncertainty of the original drill hole or blast hole data
used, the uncertainty of the geologic model and the estima-
tion (simulation) domains defined, etc. The consequences
of using alternative implementation parameters should be
fully explored and documented. All relevant production and
calibration data should be used to indicate whether the simu-
lation model is performing as expected. All possible uncer-
tainty measures in relation to an ore resource model should
be quantitatively described and discussed, whether global or
local, including internal and geologic contact dilution, the
impact of the information effect, etc. Checking and valida-
tion should be exhaustive, as well as the model presentation,
reporting, and visualization. The full set of individual simu-
lations should be used in the risk assessment study, and it in
turn fully described, validated, and documented.
10.9
Exercises
The objective of this exercise is to review a variety of simula-
tion techniques and post processing methods. Some specific
(geo)statistical software may be required. The functionality
may be available in different public domain or commercial
software. Please acquire the required software before begin-
ning the exercise. The data files are available for download
from the author's website—a search engine will reveal the
location.
10.9.1
Part One: Sequential Indicator
Simulation
The objective of this exercise is to construct SIS realizations
of a categorical variable. It is common in mining applica-
tions to have a deterministic rock type model; however, the
boundaries between the rock types are not hard boundaries.
We often want to consider fuzzy or soft boundaries. Some
topographic data is used for this exercise, but it is adequate
to communicate the principles and methodology.
Consider the 2-D
SIC.dat
data that has been provided
for your analysis. You will have to perform a cursory data
analysis to determine the X/Y limits and the variables
present. Choose a reasonable grid size relative to the data
spacing. This data was used for an international spatial inter-
polation contest in 1997. The categorical variable in the data
file is of interest here.
Question 1:
Plot a 2-D map of the indicator and note the
direction of continuity. Calculate, plot and it
indicator variograms in the principal direc-
tions.
Question 2:
Print the SIS parameter ile and comment on
the appropriate settings for this exercise. Cre-
ate two SIS realizations and plot the results.
Perform reasonable sensitivity studies on the
parameters that you are uncertain of.
