Geoscience Reference
In-Depth Information
5
Data Collection and Handling
Abstract
Estimates of mineral resources are dependent on the available data. This Chapter reviews
the main challenges related to data collection and handling. Special attention is given to
data representativity, extreme high grades, and compositing the data to practical and con-
sistent lengths. A summary of sampling theory is also presented.
5.1
Data
The samples should also be representative in a spatial
sense, which means that the spatial coverage of the deposit
is adequate. For example, the samples may have been taken
in an approximately regular or quasi-regular sampling grid,
such that each sample represents a similar volume or area
within the orebody of interest. In practice, this is rarely the
case and some declustering may be requried.
To ensure sample representativity, strict quality assurance
and quality control programs should be put in place. If the
samples are not representative, then there may be sample
bias that will directly affect the final resource estimate. A
number of issues need to be considered in relation to sample
collection, handling, preparation, and analysis.
The mining industry collects more data than other natural-
resource industries. This provides an opportunity to better
understand local variations and obtain robust local estimates.
The abundance of data play a major role in defining the mod-
eling techniques used and their implementation, and has his-
torically influenced the development of geostatistical tech-
niques. This is in contrast with, for example, some petroleum
and environmental modeling applications, where the amount
of data collected is limited, and the final results are more
model-dependent.
The quality of the mineral resource estimate is dependent
on the quality of the data collection and handling procedures
used (Erickson and Padgett
2011
; Magri
1987
). A number of
technical issues affect the overall quality of the data, but only
the most important ones are discussed here. The concept of
data quality is used in a pragmatic way, that is, with a view
to how the data affect the tonnage and grade estimates in the
resource model.
Sample data will be used to predict tonnages and grades.
Statistical analyses will be used with geological and other
technical information to make inference decisions. The sam-
ple database has to provide for sound and robust decision-
making. Although there may be many data, only a small por-
tion of the deposit is actually sampled; often less than one
billionth of the mass of a deposit is drilled.
The samples should be representative of the material in-
tended for sampling which means that the sample obtained
should result in a value that is similar to any other sample that
could have been obtained for the same volume or material.
5.1.1
Location of Drill Holes, Trenches, and Pits
The geostatistical tools used to predict the tonnages and
grade of ore material are based on knowledge of the loca-
tion of the samples. An exception is Random Kriging that is
designed for those cases where only imprecise sample loca-
tions within a defined domain are available, see Journel and
Huijbregts (
1978
, pp. 352-355) and Rossi and Posa (
1990
)
for a case study. The location of each sample is expressed as
a two or three dimensional coordinates (X, Y, and Z) and is
obtained by surveying its position in space. There are several
surveying methods that can be used. The location of the drill
hole collar as well as the deviations down the hole are sur-
veyed. The location information can be handled using differ-
ent coordinate systems, see Chap. 3, but one system should
be used for the project to avoid errors.
