TSA
TSA in TSG uses training libraries of
pure spectra
either to match the unknown project spectrum against a single
mineral or to identify the linear mixture of two minerals in the
relevant training library that best matches
the project spectrum. It can also advise whether a single mineral or
mixture of two minerals gives a better fit. TSA runs automatically in the background when
creating a new TSG dataset, and the results are written into the TSG
file as Protected System Scalars (i.e. they can be viewed and
used in TSG but cannot be directly modified).
There are two system TSA libraries in
TSG Core, one for the visible- near infrared (VNIR) and one for the
short wave infrared (SWIR), compared with only a SWIR database in
TSG Pro.
A dataset can have up
to two sets of TSA results and what is output depends on spectral
coverage of the data set. If the dataset includes the
1304-2496nm interval it provides SWIR results, if it includes the
452-1100nm interval it provides VNIR results, and if it includes the full range then
it provides both sets of results.
The TSA
SWIR training library consists of over 500 samples,
representing 46 pure minerals and 10 different non-mineral
artefact spectra. The samples have been collected from many sites
around the world, in an attempt to represent the diversity of
samples of the same mineral. The non-mineral artefacts include
“junk” classes such as the plastic, wood and galvanised iron
commonly found in drill-core and chip trays, as well as dry
vegetation.
The smaller TSA
VNIR training library consists of only 90 samples, representing 14
pure mineral and 3 different non-mineral artefact spectra, including
green vegetation and galvanised iron. Currently only some of these
are turned on by default.
In the TSG Core
version of TSA, the user can re-run the TSA Algorithm after turning
off selected minerals if it is thought that these may not occur in
the current data set. This has the benefit of limiting possible
errors of commission, but should be carried out with care so as not
to miss finding the unexpected!
Reference:
M.Berman, L.Bischof and J.Huntington (1999):
Algorithms and software for the automated identification of minerals
using field spectra or hyperspectral imagery. In Proceedings of the
13th International Conference on Applied Geologic Remote Sensing,
Volume 1, Vancouver, Canada (pp 222-232).
Aux matching
is a shape matching process which matches the signature of the
project spectrum with those in a series of spectra in a Custom
or Aux library. The Custom Library is simply another
TSG file of interpreted and/or characterised spectra either from the
same project area or from a similar geological setting.
The
idea behind custom libraries is that the Aux function allows users
to use their own characterised datasets that relate to specific
projects, alteration styles or geologic environments to match and
assist in interpreting further data sets from the same or similar
project areas and geological settings. You can also use other data
files as custom libraries simply to compare the mineralogical
similarities between two data sets, these might be two drill holes
from the same project or they could be from new targets suspected to
be similar to known mineralisation.
The big advantage of
Custom Libraries is that they can comprise spectra of mixtures
of spectra and actual examples of the spectra from a project area
rather than pure mineral spectra from museum or private collections.
A Custom Library can be:
-
A
project specific TSG file built as a result of a detailed pilot
study of the spectral characteristics of a project area. This
type of Custom Library often comprises spectra that have been
interpreted in detail and are representative of the full
spectral variation in a project area.
-
A
deposit specific TSG file comprising spectra that are
characteristic of the deposit style (e.g. such as Ausspec's
"Geological Environment Specific Spectral Libraries" or GESSL
series of TSG files).
-
A
commercially available spectral library of characterised spectra
(e.g. AusSpec's GMEX spectra).
-
Any TSG file from any area that the user wishes to compare and
overlay with
spectra in their current TSG file.
