Geoscience Reference
In-Depth Information
different than the bench height. For example, for a drill hole
with a 45° inclination, a 10 m bench composite will incorpo-
rate over 14 m of sample. Bench compositing should be re-
stricted to cases where all drill holes have no more than a 70°
inclination, although this choice is subjective. This method
should not be used when there is a significant mixture of
vertical, subvertical, and inclined drill holes.
Down-the-hole compositing usually begins at the top of
the hole, although most mining software packages will allow
for other options such as truncating at important geological
contacts. The length of material composited is always the
same, with the exceptions noted below. The inclination of
the drill hole is no longer a factor. The centroid of the com-
posite corresponds to the exact location in space of the com-
bined samples.
An important decision is whether to truncate the compos-
ite at geologic boundaries or not. The decision amounts to
incorporate some contact dilution (mixtures of grades from
either side of the contact) or to avoid such mixing. If the
composites are truncated at the boundaries, the estimation
domains will be more sharply defined and contact dilution
better controlled. The cost is a larger number of composites
of a shorter length. The most appropriate decision will de-
pend on the characteristics of the deposit and the amount of
information available.
Missing sample intervals within a composite may be
problematic if the voids are significant because there will
be a significant difference between the nominal length of the
composite versus the actual length of sample intervals com-
posited. Simple statistics can be length-weighted, using the
actual length of material composited, which is often record-
ed by most mining software packages. A maximum length
of voids within any composite is tolerated before discarding
the composite altogether. The weighting of mass fractions
should be by mass, that is, specific gravity should be consid-
ered as well as length.
The decision related to the acceptable minimum length
of a composite depends on the representativity and support
of the actual composite. This issue is more relevant if com-
posites are truncated at geologic boundaries, because there
can be a large proportion of the total number of composites
that are shorter than the nominal length. If no boundary trun-
cation is applied, then the only composites shorter than the
nominal length will be at the end of the drill holes.
A common choice in industry is to use 50 % of the nomi-
nal length as the cutoff for acceptable composite lengths.
Discarding all composites less than 50 % of the nominal
length is arbitrary. Since the main concern is representativ-
ity, a correlation study between composite length and grades
can be used to support the choice of minimum acceptable
composite length. A much shorter minimum composite
length may be acceptable where there is no detectable corre-
lation between composite length and grade. The relationship
between grade and composite length will likely be small in
base metal and massive deposits, with good core recovery.
The opposite is generally true for massive sulfide deposits,
some precious metals deposits, and in cases where core re-
coveries are poor.
5.6.3
Outliers
The term outliers is used to describe extreme high values
since many grade distributions are positively skewed. Some
distributions have low grade outliers, but this situation is less
common. These grades deviate from the general tendency of
most other grades and can be spatially and statistically isolat-
ed. In the discussion that follows, outliers are valid assayed
samples, not a consequence of spurious or erroneous data
collection. It is assumed that all potential database or ana-
lytical errors have been checked for and all possible errors
discarded or corrected in the database. Outliers are defined
in terms of geological and statistical populations.
Extreme grades are consequential in precious metal de-
posits, but not as much in base metal deposits. A statistical
analysis is always warranted to quantify how much impact
the outliers have on the final resource estimate.
The importance of outliers is often described in terms of
their contribution to the overall metal content of the deposit.
This is because not handling them appropriately can lead
to overestimation of the recoverable resource. Two aspects
must be resolved: (a) what assays should be considered out-
lier values, and (b) how to deal with them at the time of es-
timating the resources. In all cases, the analysis should be
done on the original, assayed samples. If performed on the
composited data, the outlier values may already be smoothed
depending on the type and length of the composites.
The presence of extreme grades is particularly problem-
atic if they have little spatial connectivity, that is, they are
located within a small spatially restricted volume. The more
skewed the grade distribution, the larger the potential impact
of outliers on the resource estimation process.
The determination of what values are considered outli-
ers is subjective. Outlier values are commonly examined on
a log-normal cumulative frequency plot. Breaks at the high
end of the distribution may represent outlier populations. For
example, Fig.
5.18
shows the log-normal probability plot of
Au grade in a copper-gold porphyry deposit. Note how, for
grades higher than 5.0 g/t, the distribution appears to break
up and exhibits a sudden slope change, represented by less
than 0.1 % of the total samples. Outliers are also sometimes
defined as those values that are outside a specific interval,
such as plus or minus 2 or 3 standard deviations ( ± 2
σ
or
± 3
σ
) with respect to the mean or median of the distribution.
There will always be outliers according to this definition;
professional judgement is required.
