Australian Centre for Astrobiology

1
Australian Centre for Astrobiology
2
Hyperspectral and Field mapping of an Archean
Serpentinized Komatiite unit in Western
AustraliaApplications for CRISM
Adrian Brown Malcolm Walter Thomas
Cudahy Australian Centre for Astrobiology
3
Presentation Overview
Why?
3.5 Ga rocks Ancient Microfossils Signs of
water - Stromatolites - Pillow basalts -
Serpentine
How?

• Airborne Hyperspectral VisNIR Mapping
• Ground truthing through field work

4
Hyperspectral mapping project
HyMap instrument flown at 2km AGL 126 bands from
400-2400 nm 600 sq. km, 14 swathes, ea. approx.
2km wide 5m horizontal pixel size Data were
calibrated for instrument response, and then
atmosphere was removed using HyCorr (ATREM).
Before analysis, continuum was removed using IDL
continuum removal
5
Serpentine?

• Hydrously metamorphosed peridotites contain
  serpentine, these may be found near mid ocean
  ridges, or subduction zones, where olivine rich
  rocks react with sea water to form serpentine
• Archaean greenstone belts occurs as altered
  komatiite flows
• Usually associated with overlying High Mg
  komatiite basalts

6
Serpentinisation reaction

• 2Mg2SiO4 3H2O ? Mg3Si2O5(OH)4 Mg(OH)2
• forsterite water
  serpentine brucite
• Magnetite (Fe3O4) can form to take excess iron in
  olivine no Fe in serpentine, however chlorite
  does take Fe
• In some cases, further reaction of serpentine
  forms talc and carbonate
• 2Mg3Si2O5(OH)4 3CO2 ? Mg3Si4O10(OH)23MgCO3
  3H2O
• serpentine carbon
  dioxide talc
  magnesite

7
Komatiite?

• Ultramafic lava flow
• Rich in high Mg olivine
• Olivine alters to serpentine
• Only occurs in Archaean terrains
• High temperatures - associated with mantle plumes

8
Characteristics of komatiite flow in Pilbara

• Probably formed when komatiite erupted into
  Archaean seawater and was subsequently buried and
  heated by subsequent flows
• Lateral extent - 20km x 30m
• Possible pillow flows
• Ultrabasic lava
• High MgO (gt18 wt.)
• Chloritised basalts overly the komatiite flow and
  provide a more easily detected target

2mm
9
VNIR Spectra of serpentine
10
Rule Based Detection of serpentine with
hyperspectral data
Serpentine (note 2.3-2.4 micron region)
Chlorite
Shallow depth of 2.4 feature could be due to
lack of Fe
11
Detection of serpentine with hyperspectral data
N is up, 2.5km across
Serpentine in black (picks out 2 komatiite layers)
Chlorite in red
12
Photomontage
Hyperspectral 126-band Image cube
Principal Component Analysis PCA
Feature Extraction
Dendrogram
Spectra
PCAHCA Classes
Classification
Bayesian Estimates
Hierarchical Cluster Analysis HCA
Artificial Neural Network ANN
For more information M. Storrie-Lomardi,
http//www.kinohi.org or see us at poster
location 583, Thursday PM
13
Applications for CRISM

• Multispectral survey of CRISM uses 59 bands with
  coarse resolution
• Band coverage should naturally be as intense as
  possible in SWIR, but particularly in 2.3-2.4 to
  detect serpentine
• OH fundamentals at 2.7-2.8 microns will be
  instructive
• Strategy find the chlorite first, then the
  serpentine on high resolution runs

14

• http//aca.mq.edu.au/abrown.htm