coompana regional surface geochemistry: overview and a...
TRANSCRIPT
Ryan NoblePrincipal Research ScientistCSIRO 26 Dick Perry Ave, Kensington, Western Australia 6151P +61 8 6436 8684 | M 0400 239 [email protected] | www.csiro.au
MINERAL RESOURCES
Coompana regional surface geochemistry: overview and a paradigm shift for
sampling and analysis
Project Team and Acknowledgements
• Ryan Noble, Ignacio Gonzalez Alvarez, Nathan Reid, Tenten Pinchand, Dave Cole, Jesse Robertson, Alistair White, Tania Ibrahimi, Ian Lau, Jens Klump
• Carmen Krapf, Chris Wilcox, Rian Dutch, Amy Lockheed, Liz Broadridge, Lynn Jagodzinski, Ian Hopton
Support from
• James Cleverley, Des Pascoe (Imdex) – Field prep gear
• Genevieve Rueger, Hamish Burns, Matt Stephenson (Aerotech)
• Anton Mundy, Clem Laurie (Traditional Owner Monitors)
• Paul Morris (GSWA)
Surface geochemical mapping7 days310 sitesAnalysis and infill sampling on-the-fly
Characterise the unknown
Biogeochemistry test work (2014)
Develop an optimised (“best”) practice
Six profiles to inform sampling and QAQC
Background and reconnaissance
FROM: Dunn CE and Waldron H 2014. Regional biogeochemical survey of the Eucla Basin, South Australia, Report Book 2014/00025. Department of State Development, South Australia, Adelaide.
• 6 mins per site (+4 mins travel/refuel)• 280 samples on 4 km spacing 80 x50 km area • 5 sample collected at each site • IGSN/FAIMS – all integrated in tablets and QR
code labels
Optimised Sampling
• Data capture and backup (FAIMS) –no more lost or missing sites or matching up photos/bags
• Vegetation
• Bladder salt bush Atriplex vesicaria
• Pearl blue bush Maireana sedifolia
• Soils x 2 (crust and 5-10 cm)
• Rock
In the field
How it’s done (~10 mins into 20 seconds)
Meanwhile back at base camp
How it’s done (~5 mins into 20 seconds)
24 hours lag timeOptimised QAQC especially for pXRF• 5 mins per soil sample – prep/pXRF/ASD• Geochem maps/ model updated daily• Machine learning with existing
precompetitive data• Added infill sampling with 36 points on 1
km spacing
Results on-the-fly from the portable lab
% K by sampling day
Evening analysis interpretation/discussion
Some results after Day 1
Results after Day 7
Infill area identified and sampled on Day 7
The model driven approach
• Maps and models done, gear and samples packed, home time
Day 8
Using a data driven probabilistic model to assist regional characterisation
• A Gaussian Process Regressor models the relationship between geospatial data (160 features) and pXRF geochemistry data
• ASD was not incorporated into the model this time
Day 1 Model
Day 3 Model
Day 5 Model
Day 7 Model
The field geochemical results
Increase of clays?
Estimate of kaolinite using the 2200nm absorption
ASD does not clearly support that idea
Change in limestone unit (guided from the model)
Does this match the S biogeochemistry?
Does Fe match regolith landform features?
Based on absorption in the 800-1000nm region
ASD more supportive here
Based on the wavelength of the iron oxide absorption, lower is hematite, higher is goethite
ASD also gives clues to shift in Fe phases that pXRF does not
The challenge of pXRF (sorry explorers… for now)
Work to be done
Soil, plant and rock analysis in “regular” time
• Includes ultrafine fraction workflow
• Landscape model/setting with some down hole and mineralogical analyses
Optimisation testing with “smart sampling” and related cost benefit analysis
Integration of spectral/particle size with “other data”
Data and report release with industry workshop day (early June TBA)
The future of a regional geochemical workflow
24 hrs
Element X
Infill target identified
Samples collected, analysed and modelled (in batches) to adjust sampling
• Could this be done between sites?
• A variant of “smart sampling” becomes the industry norm instead of grids
• Anomaly recognition for exploration is the next challenge (not background variation) once regional background is determined
• Model/field lab to include other analyses ASD, LIBS, pXRD
Ryan NoblePrincipal Research ScientistMineral ResourcesPhone: 08 6436 8684Email: [email protected]: www.csiro.au
On behalf of the project team, thank you for your time