31
Mark Jessell WA Fellow & Winthrop Professor Laurent Aillères, Tom Carmichael, Monash Uni Eric de Kemp, Mike Hillier, Geol. Survey Canada Roland Martin, CNRS, Toulouse Mark Lindsay CET UWA Stéphane Perrouty, Uni Toulouse Next Generation 3D Modelling & Inversion After Dante 1315

Mark Jessell - Next Generation 3D Modelling

Embed Size (px)

DESCRIPTION

Developing better integration of geological constraints into 3D regional modelling Identify ways to carry geological meaning through the geophysical inversion process

Citation preview

Page 1: Mark Jessell - Next Generation 3D Modelling

Mark JessellWA Fellow & Winthrop Professor

Laurent Aillères, Tom Carmichael, Monash UniEric de Kemp, Mike Hillier, Geol. Survey CanadaRoland Martin, CNRS, ToulouseMark Lindsay CET UWAStéphane Perrouty, Uni Toulouse

Next Generation 3D Modelling & Inversion

After Dante 1315

Page 2: Mark Jessell - Next Generation 3D Modelling

Aims of WA Fellowship Program

• Better integrate geological constraintsinto 3D regional modelling

• Carry geological meaning through the geophysical inversion process

Page 3: Mark Jessell - Next Generation 3D Modelling

3D Model(s) of AustraliaGautier Laurent et al 2013

ARC DP1096409 Betts, Aillères, Jessell & de Kemp

Active MinesExisting or planned 3D models

Page 4: Mark Jessell - Next Generation 3D Modelling

A good question

Page 5: Mark Jessell - Next Generation 3D Modelling

Sedimentary Basins

Mines Regional Lithosphere

3D Constraints

RICH (3D seismic, deep boreholes, gravity)

RICH (dense boreholes,magnetics, seismic, electromagnetics)

POOR (rare boreholes, surface outcrops, gravity, magnetics)

RICH(Teleseismic, seismic, gravity, MT)

Structural Complexity

SIMPLE(R) COMPLEX COMPLEX SIMPLE(R)

Dedicated Software

Gocad 1989, Geomodeller 1999…

MicroMine 1986,Leapfrog 2003...

Noddy 1981 Gocad 1989

3D geomodelling scenarios

Page 6: Mark Jessell - Next Generation 3D Modelling

Jessell, 1981

Knowledge Data Data + Knowledge Data + Knowledge + Uncertainty

Short history of 3D modelling

Page 7: Mark Jessell - Next Generation 3D Modelling

3D geology is an under-constrained problem

We do not have sufficient geological and/or geophysical data to define a unique 3D model

We should not restrict ourselves to a single 3D model

3D Modelling tools that require continualmanual intervention are a dead end

Page 8: Mark Jessell - Next Generation 3D Modelling

Ashanti Belt

3.6 - 2.5 Ga

2.3 - 2.0 Ga

after Milési et al., 2004, BRGM SIGAfrique

Gold deposits

Orogenic

Placer

______10 km

Perrouty et al., submittedEcon Geol 2013

West African Craton

3D Prospectivity Analysis

Page 9: Mark Jessell - Next Generation 3D Modelling

Tarkwa Basin

Kumasi Basin Akyem

BasinVolume : 160*160*15 kmResolution : 200 m

Model (after inversion)

0 (m)

-14000

Depth(Faults)

3D Prospectivity Analysis

Perrouty et al., submittedEcon Geol 2013

Page 10: Mark Jessell - Next Generation 3D Modelling

BVC2BV1

BVC1

xy

- Measure the minimum distance between deposits andthe different Sefwi Group units

- Count the number of deposits found less than 1500 mfrom each unit

BVC2 BV1 BVC1

Distance 0 m x y

Fault or unconformity

3D Prospectivity Analysis

Number of deposits less than 1500m from each unit

Stra

tigra

phic

Dep

th(m

)

Perrouty et al., submittedEcon Geol 2013

Page 11: Mark Jessell - Next Generation 3D Modelling

Challenges

Better Inputs

Better use of geology duringmodelling & inversion

Better analysis of results of modelling

a) Trainingb) Structural analysisc) Geophysical imaging

Page 12: Mark Jessell - Next Generation 3D Modelling

a) Training:Structural Geophysics Course

Page 13: Mark Jessell - Next Generation 3D Modelling

κ’ Kent Distribution(measure of clustering)

b) Structural analysis:Intelligent upscaling

Tom Carmichael, Monash Uni

Limousin, France

Page 14: Mark Jessell - Next Generation 3D Modelling

c) Geophysical Imaging: Full Tensor Gravity

Daniel Wedge, CET (Data from First Quantum) →ARC Linkage

Page 15: Mark Jessell - Next Generation 3D Modelling

Challenges

Better Inputs

Better use of geology duringmodelling & inversion

Better analysis of results of modelling

a) Better use of field datab) Geologically Appropriate interpolation schemesc) Integrated Inversion

Page 16: Mark Jessell - Next Generation 3D Modelling

a) Better use of field data Alsop et al., 1996(Near Moine Thrust)

Page 17: Mark Jessell - Next Generation 3D Modelling

Structural Interpolation

Geology 101

?

?GeomodellerSKUALeapfrog

Jessell, Aillères, de Kemp & 1000Structural geologists

Page 18: Mark Jessell - Next Generation 3D Modelling

a

b

bedding

d

c

?

Geology 201

Bedding Only

Page 19: Mark Jessell - Next Generation 3D Modelling

a

b

d

c

F1F2

Need vergence of minorfolds or Sn/Sn+1 to constrain fold model

F1

bedding

S2 cleavage

Geology 201

Bedding-CleavageRelationships with Relative Timing

S1 cleavage

Page 20: Mark Jessell - Next Generation 3D Modelling

Multiquadric RBF (stratigraphy, faults?)

Michael Hillier (Geol Survey Canada)

Same input data:

Geologically Appropriate

interpolation schemes

Geology 501

RBF=Radial Basis Function

Page 21: Mark Jessell - Next Generation 3D Modelling

Geologically Appropriate

interpolation schemes

Gaussian RBF (salt domes?)

Michael Hillier (Geol Survey Canada)

Same input data:

Geology 501

RBF=Radial Basis Function

Page 22: Mark Jessell - Next Generation 3D Modelling

Thin plate spline RBF (folds?)

Michael Hillier (Geol Survey Canada)

Same input data:

Geologically Appropriate

interpolation schemes

Geology 501

RBF=Radial Basis Function

Page 23: Mark Jessell - Next Generation 3D Modelling

Doesn’t explaingeophysical signal

c) Integrated Inversion • Structures• Age Relationships• Petrology• Geophysics• Petrophysics• Prior Knowledge•••

Petrophysics

• Need inversion schemes thatretain geological meaningthrough the inversion process

• So we can test the resultsagainst the original geologicalAND geophysical data

• Currently working on speedingup inversion so that in the future we can include bettergeological constraints

Page 24: Mark Jessell - Next Generation 3D Modelling

Challenges

Better Inputs

Better use of geology duringmodelling & inversion

Better analysis of results of modelling

a) Uncertaintyb) Geodiversity

Page 25: Mark Jessell - Next Generation 3D Modelling

Vary inputs:• Orientations• Position• Age

relationships

Original Inputs

Perturbed Inputs 1

Perturbed Inputs 2

Perturbed Inputs 3

Perturbed Inputs 4

Perturbed Inputs N

•••

Implicit Modelling

Engine

Wellman et al., 2010, 1011Jessell et al., 2010Lindsay et al., 2012,2013

a) Uncertainty & Simulation

Page 26: Mark Jessell - Next Generation 3D Modelling

Stratigraphic Variability = number of possible lithologies

Lindsay et al., 2012

• Where do we need to collect more data?

• Use variability to weight petrophysical inversions

1 Lithologies per voxel 6

Page 27: Mark Jessell - Next Generation 3D Modelling

0

0

0.40.30.2

0.4

-0.1-0.2-0.3-0.4

-0.3

-0.4

-0.2

-0.1

0.1

0.3

0.2

0.1

0.5

-0.5

0.5-0.5

b) Geodiversity via Principal component analysis

Original Model: 11th closest to

barycentre

Studying variation between plausible models

• Initial model NOT the most representative model

• Use diverse range of models to seed geophysical inversions

Gippsland Basin model suite

Lindsay et al., 2013

Based on multiple geologicalattributes: unit volume, unit depth, surface complexity, geophysical misfit…

Principal Component 1

Prin

cipa

l Com

pone

nt 2

Page 28: Mark Jessell - Next Generation 3D Modelling

Better Inputsa) Trainingb) Structural analysisc) Geophysical imaging

Better use of geology during modelling & inversiona) Better use of field datab) Geologically Appropriate interpolation schemesc) Integrated Inversion

Better analysis of results of modellinga) Uncertaintyb) Geodiversity

3D Modelling Challenges

Page 29: Mark Jessell - Next Generation 3D Modelling

ARC Linkage Submitted: Reducing 3D uncertainty via improved data interpretation methodsMark Jessell, Eun-Jung Holden, Mark Lindsay, Klaus Gessner, Jon Hronsky

ARC Centre of Excellence Submitted: Computational Geoscience and Earth ModellingMultiscale Analysis & Modelling

Western Australian Fellowship: WA_In3DMark Jessell

Page 30: Mark Jessell - Next Generation 3D Modelling

The problem that emerges when a model of a phenomenon is just as hard to understand as the phenomenon that it is supposed to explain.

Everything simple is false, everything complex is unusable

Paul Valéry, 1937

Bonini’s Paradox

The challenge we have set ourselves is to make our models (and modelling systems) less false, without becoming unuseable

Page 31: Mark Jessell - Next Generation 3D Modelling