NORTHERN TERRITORY GEOLOGICAL SURVEY Maximising the value of NVCL HyLogger data: Understanding automated mineralogical interpretations Belinda Smith, Mark Berman, Ralph Bottrill, Tania Dhu, Suraj Gopalakrishnan, Georgina Gordon, David Green, Jon Huntington, Alan Mauger
30
Embed
Maximising the value of NVCL HyLogger data: Understanding … · 2016-04-14 · NORTHERN TERRITORY GEOLOGICAL SURVEY Maximising the value of NVCL HyLogger data: Understanding automated
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
NORTHERN TERRITORY GEOLOGICAL SURVEY
Maximising the value of NVCL HyLogger
data: Understanding automated
mineralogical interpretations
Belinda Smith, Mark Berman, Ralph Bottrill, Tania Dhu, Suraj Gopalakrishnan,
Georgina Gordon, David Green, Jon Huntington, Alan Mauger
NORTHERN TERRITORY GEOLOGICAL SURVEY
Introduction
• Aims– Understanding what the HyLogging NVCL is
– New to HyLogging – understanding what it is and what it is
telling you
– HyLogger User – understanding the different levels of
processing
– Increase understanding of the methods and processes
• Voluminous data with 100,000s spectra in a single drillhole
• Hundreds of drillholes; thousands of metres nationwide
• Automate the mineral matching with;– TSA (The Spectral Assistant). Refine with;
– Restricting minerals in TSA (uTSA)
– Creating domains in TSA (dTSA)
– Using CLS with domains
Automated Mineralogical Interpretation
NORTHERN TERRITORY GEOLOGICAL SURVEY
Mineralogical Interpretation of HyLogger Data
The Spectral Assistant (TSA) – System MatchesAlgorithm matching to whole spectral library
TSA – User Matches (uTSA)
Manual restriction of unlikely mineralsMinerals match across whole hole
sTSA
ALL DATASETS
Domaining (dTSA)
CLS
Constrained Least
Squares
External
Validation
Creating and using scalars
(XRD, petrography, SEM)
uTSA +/- scalars
Good for basement / seds drillholes
Recommended for TIR spectra
processing (non-unique spectra) Steps are not sequential
NORTHERN TERRITORY GEOLOGICAL SURVEY
TSA mineral matching• ‘Unmixes’ spectral response into mineral
mixes of 2 (SWIR) or 3 (TIR) using an
algorithm
• Only matches to minerals in the library
• If a mineral isn’t active in that wavelength
range, you won’t see it
• If there’s more than 2 - 3 (SWIR) or 3 -4
(TIR) minerals per spectrometer you won’t
see them
• It is NOT an assay! You get a indication of
minerals present, NOT an absolute
quantifiable modal value.
• Nonetheless valuable as nothing else is as
fast.
NORTHERN TERRITORY GEOLOGICAL SURVEY
• Automated process which can handle (simple) mineral
mixtures very quickly
• Detailed: mineral match output for each spectrum
(8mm*18mm)
• Results returned as mineral names (rather than wavelengths)
so outputs easily understood by geoscientists
• Summary overviews can show bulk mineral changes in the
entire drilled length
Strengths of TSA Mineral Outputs
NORTHERN TERRITORY GEOLOGICAL SURVEY
Detailed TSA outputs
NORTHERN TERRITORY GEOLOGICAL SURVEY
TSA results…. sTSA Summary
Black shales, dolomitic quartz sandstones metamorphosed basementdolostones
epidote
talc
SYSTEM TSA
has
‘unsupervised/
unedited’
watermark.
Mathematical
fit; no geology
used.
NORTHERN TERRITORY GEOLOGICAL SURVEY
TSA results…. uTSA Summary
quartz
calcite
calcitedolomite
chlorite
amphibole
USER (uTSA) –
has been
refined to
exclude unlikely
minerals when
spectral
matching.
BUT sediments
show epidote,
phlogopite
(can’t exclude
because they’re
in basement).
talc
gone
epidote still present
phlogopite
NORTHERN TERRITORY GEOLOGICAL SURVEY
TSA results…. uTSA Summary DOMAINED
quartz
calcite
calcite dolomite chlorite
amphibole
Results have
been
DOMAINED; no
phlogopite,
epidote in
sediments but
is in basement. epidote gone
talc, phlogopite removed
epidote still here
NORTHERN TERRITORY GEOLOGICAL SURVEY
TSG SOFTWARE – mineral matching (TSA)
Well-crystalline kaolinite 100%SRSS = 146
Petermann Sandstone; from LA05DD01Quartz 100%Does not see kaolinite feature Appears to be a well-sorted sandstone
TSA SWIR response of wx kaolinite only
TSA TIR response of quartz only
SWIR can’t ‘see’ quartz
CLS has quartz AND kaolinite
SWIR & TIR results
93% quartz, 6% kaolinite
black = spectrumblue = modelled spectrum
black = spectrumblue = modelled spectrum
black = spectrumgreen = modelled spectrum
SWIR TSA says..
• FeMg Chlorite
• Phengitic mica
TIR Spectra Mixing
non-uniquenessTIR TSA
Result
Slide courtesy Andy Green, OTBC Pty Ltd
NORTHERN TERRITORY GEOLOGICAL SURVEY
• TSA can’t effectively look at as many mixtures in the TIR
(n=3). CLS can return 6 – 7 mixtures
• TSA has difficulty with non-unique spectra
• TSA concentrates on getting the best fit over the whole
spectrum and may miss small features (think kaolinite in
Petermann Sandstone)
• CLS has to use a manually selected Restricted Mineral
Set (optional in TSA)
CLS vs. TSA in the TIR
NORTHERN TERRITORY GEOLOGICAL SURVEY
• HyLogger returns voluminous whole-of-drillhole measurements vs. individual point sampling done with hand-held devices
• HyLog FIRST, destructive sampling SECOND (when wanting to compare mineral results from two techniques)
• HyLogger weakness is quantifiable mineralogy
• Mineralogy validation ongoing; many Surveys have in-house XRD facilities
HyLogging vs. other mineralogy outputs: XRD / EDAX / Petrography
NORTHERN TERRITORY GEOLOGICAL SURVEY
• Relatively semi-quantitative– May depend on absorption co-efficient
– Linear surface measurement (not whole core)
– Does not deal well with complex mixtures
– Proportions are RELATIVE…. If a dominant mineral is inactive in that wavelength range, then it won’t be returned as a mineral match from that spectrometer
– If a mineral is < 15% of the mineral mix, it won’t be listed (default)
TSA Mineral Mix Proportions
NORTHERN TERRITORY GEOLOGICAL SURVEY
• Core quality (look at imagery)
• Complexity of mineral
assemblage
• Degree of processing
• Incorporating scalars into the
processing workflow
• Domaining
• External validation
Factors influencing confidence levels of automated mineral modelling
The Spectral Assistant (TSA) – System Matches
Algorithm matching to whole spectral library
TSA – User Matches (uTSA)
Manual restriction of unlikely minerals
Domaining (dTSA)
CLS
Combined Least
Squares
External
Validation
Creating and using scalars
NORTHERN TERRITORY GEOLOGICAL SURVEY
• Good for voluminous whole-of-drillhole
data
• Good for mineral matches of 2 – 3
minerals
• Good for stratigraphic
boundaries/correlations; alteration haloes
• Increased level of processing (restricted
mineral matching of algorithm using
context and common sense) gives
increased confidence in the results
• Need to learn to recognise what is likely,
what is real; what needs validation with
external techniques
HyLogger Data – ‘fit for purpose’?
The Spectral Assistant (TSA) – System Matches
Algorithm matching to whole spectral library
TSA – User Matches (uTSA)
Manual restriction of unlikely minerals
Minerals match across whole hole
Domaining (dTSA)
CLS
Combined Least
Squares
External
Validation
Creating and using scalars
NORTHERN TERRITORY GEOLOGICAL SURVEY
• Use of scalars• Look at the imagery (core quality)• Look at the spectral fit to the algorithm• Validate using complementary
techniques (XRD, petrography, etc)• Use your geological knowledge:
– ‘am I likely to see that mineral in that geological environment?’
– is that mineral responsive in that wavelength range? (think quartz/ feldspars not showing in SWIR)
– is that mineral in the TSA library?
Validate the mineral response
NORTHERN TERRITORY GEOLOGICAL SURVEY
• HyLogging is a valuable technique
• Automated mineralogical interpretation methods are essential to deal with the volume of data
• Understanding the methods and processes behind the mineral outputs will assist geoscientists in applying NVCL HyLogger data to their specific problems with confidence
• Next speakers can show examples of applying HyLoggerdata
Conclusions
NORTHERN TERRITORY GEOLOGICAL SURVEY
• Andy Green (OTBC Pty Ltd) has contributed to much of this work, particularly CLS and TSA modelling work.