Laterally Constrained 1Dinversion of 3D TEM data - PowerPoint PPT Presentation

Title: Laterally Constrained 1Dinversion of 3D TEM data


1
Laterally Constrained 1D-inversion of 3D TEM data

• Esben Auken, Anders V. Christiansen, Lars
  Jacobsen and Kurt Sørensen

HydroGeophysics Group Department of Earth
Sciences, University of Aarhus,
Denmark www.hgg.au.dk
2
Outline

• The study
• The Laterally Constrained Inversion (LCI)
  algorithm
• 1D-LCI modeling of 3D buried valley structures
• SkyTEM field data example
• Summary

3
The Study

• Development of enhanced 1-D based layered
  inversion of TEM data
• Natural development - LCI is routinely used for
  inversion of
• geoelectric data in 1-D and 2-D CVES and PACES
• TEM jointly with geoelectric data in 1-D
• Use 3-D models of buried valley structures as
  test models
• Important to correctly model
• 3-D near-surface resistivity variations
• 2-D overall structures
• Natural background noise
• Modeling on dense sampled profile data e.g.
  SkyTEM data

4
Base Model - Buried Valley

• Buried valleys general properties
• Depth 100 250 m, resistivities 2 200 W-m
• Filled with sand and/or gravel and/or clay
  sediments
• Caved into low resistivity clay
• Covered with inhomogeneous tills
• 2D /3D slopes up to 30 - 45 deg

5
Outline

• The Study
• The 1D Laterally Constrained Inversion (LCI)
  algorithm
• 1D-LCI modeling of 3D buried valley structures
• SkyTEM field data example
• Summary

6
LCI method

• Subsurface divided into a large number of 1D
  models (quasi-1D structures)
• Constraints carries information on the geologic
  variability
• Output model sections with smooth lateral
  variations and layer boundaries
• Applied routinely for continuous geoelectric data
  - CVES - PACES

7
LCI method

• Subsurface divided into a large number of 1D
  models
• Constraints carries information on the geologic
  variability
• Output layered model sections with smooth
  lateral variations
• Applied routinely for continuous geoelectric data
  - CVES - PACES

8
1D-LCI - inversion methodology

• Schematic representation
• 
  ,
• G is the Jacobian matrix, data variances in Cobs
• R is the roughening matrix, constraint variances
  in Cc
• mprior is the a priori model, constraints
  variances in Cprior
• e is error vectors

9
1D-LCI - inversion methodology

• Schematic representation
• 
  ,
• The model update for iteration n1

10
Derivatives - with a 1D forward code

• Calculation of G is time-consuming
• 1D derivatives

11
Outline

• The Study
• The Laterally Constrained Inversion (LCI)
  algorithm
• 1D-LCI Modeling of 3D buried valley structures
• SkyTEM field data example
• Summary

12
Base Model - Buried Valley
13
Base Model - Buried Valley

• 3D forward modeling
• Finite difference grid

14
Base Model - Buried Valley

• 3D forward modeling
• Finite difference model
• Stochastic top layer
• Central area divided into 10 m cells. Total model
  size approx. 10 km.
• Sounding calculated for each 20 m
• Accuracy better than 10 on 1D models

15
Noise Simulation of Real Field Data

• Noise-free data, V

16
Noise Simulation of Real Field Data

• Noise-free data, V
• Uniform noise, STDuni

17
Noise Simulation of Real Field Data

• Noise-free data, V
• Uniform noise, STDuni
• Noise level, Vnoise

18
Noise Simulation of Real Field Data

• Noise-free data, V
• Uniform noise, STDuni
• Noise level, Vnoise
• Synthetic data, Vresp

19
Noise Simulation of Real Field Data

• Noise-free data, V
• Uniform noise, STDuni
• Noise level, Vnoise
• Synthetic data, Vresp
• High-moment synthetic data, Vresp

20
Buried Valley Model 1
21
Buried Valley Model 1 - Result
22
Buried Valley Model 2
23
Buried Valley Model 2 - Result
24
Buried Valley Model 2 Data from 2400 m
25
Buried Valley Model 2 Data from 2400 m
26
Buried Valley Model 2 Data from 2400 m
27
Buried Valley Model 3
28
Buried Valley Model 3 - Result
29
Outline

• The Study
• The Laterally Constrained Inversion (LCI)
  algorithm
• 1D-LCI modeling of 3D buried valley structures
• SkyTEM field data example
• Summary

30
The SkyTEM System at a Glance

• SkyTEM Helicopter TEM system
• Resolution capabilities are the same or better
  than achieved from ground based systems

31
The SkyTEM System at a Glance

• Magnetic moment 40,000 50,000 Am2
• Operating altitude 15 m (up to 30 m), speed 20
  km/hour
• One sounding every 40 m

32
Survey
Denmark
Germany
33
Inversion Result
34
Survey
35
Inversion Result
36
Outline

• TEM in groundwater exploration
• The Laterally Constrained Inversion (LCI)
  algorithm
• 1D-LCI modeling of 3D buried valley structures
• SkyTEM field data example
• Summary

37
Summary

• 1D LCI of TEM data is possible
• The resolution of the subsurface is enhanced
  where data carries little information
• A priori information about layer boundaries or
  resistivities greatly enhances the model
  resolution
• Fast and reliable inversion scheme

38
Thank you...

• Knutur Árnason, Energy Department, Iceland, for
  letting us use his 3D TEM FD code
• Flemming Jørgensen, Vejle County, Denmark for
  providing us with the geological buried valley
  models and for numerous discussions of the results

39
Related Presentations Thursday Afternoon

• Integrated inversion of CVES and TEM data using
  lateral constraints A. V. Christiansen,
  N. Foged, E. Auken and K. I. Sørensen
• Interpretation of a hydrogeophysical survey -
  data from the high-resolution SkyTEM system
  
  K. I. Sørensen, B.
  Sørensen, A. V. Christiansen and E. Auken

40
(No Transcript)