1/332016 IP workshop

The University of British Columbia

Geophysical Inversion Facility

gif.eos.ubc.ca

Seogi Kang and Douglas W. Oldenburg

3D TEM-IP inversion workflow for

galvanic source TEM data

IP workshop 2016

6th June 2016

2/332016 IP workshop

Conceptual model for IP response

Air

Earth

v

3/332016 IP workshop

Air

Earth

v

EM coupling?

4/332016 IP workshop

• Measured voltage includes both EM and IP

effects

– How are we recovering conductivity and

chargeability?

– Are we okay with EM-contamination when

recovering chargeability?

Questions?

5/332016 IP workshop

• Maxwell’s equations:

Simulation of TEM data

Ohm’s law in frequency domain Ohm’s law in time domain

Frequency domain Time domain

6/332016 IP workshop

• Cole-Cole model (Pelton et al., 1978)

Complex conductivity

Inverse Fourier

transform

Frequency domain Time domain

7/332016 IP workshop

• Maxwell’s equations:

Simulation of TEM data with IP

Ohm’s law in frequency domain Ohm’s law in time domain

Frequency domain Time domain

EMTDIP code

(Marchant et al., 2015)

8/332016 IP workshop

Observed response?

• Voltage

9/332016 IP workshop

Observed response?

• Voltage

10/332016 IP workshop

Define IP datum

Observation

FundamentalIP = Observation - Fundamental

• IP datum:

• Voltage

11/332016 IP workshop

Define IP datum

Observation

FundamentalIP = Observation - Fundamental

• IP datum:

• Voltage

12/332016 IP workshop

• Oldenburg and Li (1994)

Two stage IP inversion workflow

Invert DC data,

to recover

Linearized equations

Invert data,

recover pseudo-chargeability Seigel (1959)

13/332016 IP workshop

• Kang and Oldenburg (2015)

Invert TEM data,

to recover

Invert data,

recover pseudo-chargeability

Linearized equations

Compute IP datum

Remove EM responses

Estimate intrinsic

IP parameters

IP = Observation - Fundamental

3D TEM-IP inversion workflow

14/332016 IP workshop

• Decouple EM induction effects in the

observed data

– What conductivity?

• Half-space

• Recovered by inverting DC data,

• Recovered by inverting TEM data

• Invert IP datum, recover 3D chargeability

An objective

15/332016 IP workshop

Forward modelling set up

A1 A2

A3 A4

A1 A2

A3 A4

Measure potential difference (Easting direction)

- 200 m bi-pole (625 mid points)

Time range: 1 – 600 ms

Chargeable objects: A2 and A3

- Cole-Cole parameters:

3D mesh:- Core cells:

100mx100mx50m

100mx100mx100m

- # of cells:

52x52x55

16/332016 IP workshop

Forward modelling set up

A1 A2

A3 A4

Conductivity at

Infinite frequency:

A1 A2

A3 A4

Measure potential difference (Easting direction)

- 200 m bi-pole (625 mid points)

Time range: 1 – 600 ms

Chargeable objects: A2 and A3

- Cole-Cole parameters:

17/332016 IP workshop

Observed DC data

Voltage Apparent conductivity

18/332016 IP workshop

Observed data (off-time)

Off-time at 5 ms

• Decaying curves at A1-A4

19/332016 IP workshop

Observed data (off-time)

Off-time at 80 ms

• Decaying curves at A1-A4

20/332016 IP workshop

Observed data (off-time)

Off-time at 130 ms

• Decaying curves at A1-A4

21/332016 IP workshop

Observed data (off-time)

Off-time at 650 ms

• Decaying curves at A1-A4

22/332016 IP workshop

• Time decaying curves (off-time)

EM decoupling

IP = Observation - Fundamental

No hope Can make a

difference

No need to

do anything

EM

Induction

Intermediate

IP

23/332016 IP workshop

• Off-time at 80 ms

EM decoupling: true

Observation Fundamental IP

24/332016 IP workshop

• Off-time at 80 ms

EM decoupling: true

Observation Fundamental IP

25/332016 IP workshop

What data we have?

Late on-time data (DC) Early off-time data (TEM)

26/332016 IP workshop

• Data misfit:

• Tikhanov style regularization:

• Statement of the inversion:

• Depth weight:

3D inversion methodology

DC-IP inversion: SimPEG-DCIP

TEM inversion: UBC-H3DTD code

27/332016 IP workshop

• Recover 3D conductivity

3D DC inversion

A1 A2

A3 A4

A1

A3 A4

True Estimated

28/332016 IP workshop

• Off-time at 130 ms

EM decoupling:

Observation Predicted

True IP

Raw IP

Half-space

29/332016 IP workshop

• Recover 3D conductivity

3D TEM inversion

EM

Induction

Intermediate

IP

Use uncontaminated EM data

Observed vs. Predicted

Time range: 1-6 ms

(6 channels) A1 A2

A3 A4

A1

A3 A4

Estimated

30/332016 IP workshop

• Off-time at 80 ms

EM decoupling:

Observation Predicted

True IP

Raw IP

Half-spaceDC

31/332016 IP workshop

• Recovered 3D conductivity

Comparison of 3D conductivities

True

A1 A2

A3 A4

A1

A3 A4

Estimated

A1 A2

A3 A4

A1

A3 A4

Estimated

32/332016 IP workshop

• Fundamental data at 80 ms

Comparisons of Fundamental

Fundamental Half-space DC TEM

A3

33/332016 IP workshop

• IP data at 80 ms

Comparisons of IP

True IP Half-space DC TEM

False chargeable target

Wrong amplitude

A3

34/332016 IP workshop

• Chargeability: recovered by inverting:

3D IP inversion

A1 A2

A3 A4

A1

A3 A4

A1 A2

A3 A4

A1

A3 A4

A1 A2

A3 A4

A1

A3 A4

SimPEG-DCIP code

True Half-space

A1 A2

A3 A4

A1

A3 A4

DC TEM

35/332016 IP workshop

• Pseudo-chargeability > 0.015

3D cut-off volume

A1 A2

A3

A1 A2

A3

A1 A2

A3

True Half-space

TEMDC

A1 A2

A3

36/332016 IP workshop

Summary

Invert TEM data,

to recover

Invert data,

recover pseudo-chargeability

Linearized equations

Compute IP datum

Remove EM responses

Estimate intrinsic

IP parameters

3D-TEM IP inversion workflow

Observation Predicted Raw IP

dI Pr aw (t) = F [σ(t)] − F [σE M

est ] + noise(t)

A1 A2

A3

TEMDC

A1 A2

A3

37/332016 IP workshop

“Traditionally, early time TEM data has been discarded”

“By using these discarded TEM signals we can better

estimate both 3D conductivity and chargeability”

Take home

38/332016 IP workshop

Thank you

39/332016 IP workshop

• Special thanks to

– SimPEG developers for utility codes to make

this study, and constructive discussions

– SimPEG-DCIP package for DC-IP inversions

– UBC-GIF for allowing us to use H3DTD codes

– Patrick Belliveau and Eldad Haber for

developing time domain IP code

Acknowledgements