2D NUMERICAL Forward and Inverse MODELLING FOR INTERPRETATION

OF RESISTIVITY DATA USING FINITE DIFFERENCE

APPROACH
 Work Plan
Ohms law & Eq. of Continuity
Simplification
Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Matrix Formulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Equation of Continuity
► Principle of conservation of charge over a volume
and the equation of continuity is given by

Charge density specified at a point in the cartesian

space by the Dirac delta function.

► Ohm’s law which governs the flow of current into

the ground and is given by
 Work Plan
Equation of Continuity
Simplification
Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Numerical simulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Mathematical model

► Stationary electric fields are conservative

►Applying the above equation in Ohm’s law,

► From conservation of charge over a volume and

equation of continuity
 Mathematical model
► Equation of continuity for a generalized three-
dimensional space

►By application of elementary vector calculus, above

equation can be written as
 Mathematical model
► Assumption
● Change in the conductivity
distribution in the y (strike) direction

► Above equations can be written as

And
 Mathematical model
►Vector relation
●

► Above equation becomes

ϕ and the source term

are functions of whereas the conductivity is a
function of and .
 Work Plan
Equation of Continuity Simplification

Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Numerical simulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Fourier Transform
► Fourier Transform
● Potential distribution ϕ is transformed from the 3D
space into a 2D space (,,) using Fourier Transform
dy
and
=
 Fourier Transform
► So, transforming above equations in Fourier space,
=Q̃δ()

{σ(x,z)}+σ(x,z)-σ(x,z)-2
=-2 Q̃δ()
Where Q̃ =
 Work Plan
Equation of Continuity
Simplification
Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Matrix Formulation
Potential Distribution ϕ
Inverse Fourier Transform

Pseudo-section
 Discretization of the Problem
►Generalized form of above equations can be written as,

; (x,z) ε R
which is assumed to be closed and connected, to have a non-
void interior and to have a sufficiently regular boundary Ӷ with
outward normal η on which

P(x,z) > 0 and σ(x,z) > 0 ; (x,z) ε R

and , ;
 Discretization of the Problem
►To define the semi-infinite lower half
space with arbitrary conductivity distribution,
the set R is designed with artificial
boundaries simulating infinitely distant
planes in the horizontal (x-direction) and
vertical (z-direction) extent. Such a lower
half-space is indicated by the grid shown
in the figure

Figure shows Finite Difference Mesh and Dipole- Dipole

Electrode Configuration Measurements; Working Zone and
Far Zone; Domain Boundaries; Cells of Different Physical
Properties and Nodes.
 BOUNDARY CONDITION
►ϕ (x, z) must be continuous across each element boundary
of the contrasting physical property distribution of σ (x, z).
►Normal component of J()must be continuous across each
boundary.
►Boundary conditions along the left, right and bottom edges
become Dirichlet-type.
►At the ground surface with z=0, it is implemented by
applying the Neumann type condition.
i.e.,
►Q every where except at the location of the current
electrodes. Here source and sink of + Qand − Qare inserted.
 Work Plan
Equation of Continuity
Simplification
Mathematical model

Fourier Transform

Discretization of the
problem and BCs

Numerical algorithms and

implementation

Matrix Formulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Numerical Algorithm
► Finite Difference Method
● For interior Nodes
□ At any node in the set R, Φ̃ (x,) must satisfy the equation

{σ(x,z)}+σ(x,z)-σ(x,z)-2
=-2 Q̃δ()
The two-dimensional partial differential operator

is applied on and
 Numerical Algorithm
□ At any node (i,j) with irregular grid-spacing in x- and
z-direction the operator on any distribution of , could
be approximated by the finite difference equation as
 Numerical Algorithm
□Application of the difference operator of
generalized equation to the above equation for
any interior node
(i,j) of the grid results in the discretized form
given below.

++

=2Q̃δ()δ()
 Numerical Algorithm
□ In notational form above equation can be written as

Where ,
,
, ,
 Numerical Algorithm
► ɸ̃ at node (i, j) is dependent only on the values of ɸ̃ at
the adjacent nodes (i,j-1),(i,j+1),(i-1,j), and (i+1,j). C
are functions of the geometry of the discretization grid
and the physical property values , and hence known at
all nodes in the set R.
 Numerical Algorithm
● For boundary nodes
a) For nodes located on the line z=0
For all nodes (i,j) with j=2,3,…..M-1, i=1,

Where ,
, and
 Numerical Algorithm
b) For the top, left, and right corner nodes
located on the line z=0
□ For the top left corner node (1,1)

Where ,
–
and α=
□ For the top right corner node (1,M)

Where ,
and -
 Numerical Algorithm

c) For nodes located on the bottom edge of the mesh (z)

□ For all nodes (N,j), with j=2,3,…..M-1,

Where ,
,
-
 Numerical Algorithm

d) For the bottom left and right corner nodes

□ For the bottom left corner node (N,1)

Where ,
and -
-
□ for the bottom right corner node (N,M

Where ,
and -
 Numerical Algorithm
e) For nodes located on the left edge of the
mesh
The difference equation for the nodes (i,1), for
i=2,3,…..N-1, has the form given below

Where ,

and -
 Numerical Algorithm
f) For nodes located on the right edge of the
mesh
The difference equation for the nodes (i,M), for
i=2,3,…..N-1, that are on the edge
corresponding to , is given below

and -
 Work Plan
Equation of Continuity
Simplification
Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Matrix Formulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Matrix Formulation
► All the above equations for interior nodes and boundary
nodes can be written in matrix form as

The
matrix C is called the
Capacitance matrix
which is a function
of the geometry and
the physical property
distribution in the grid. Sample nodes numbering convention for rectangular grids.
 Matrix Formulation
► The matrix C is blocktridiagonal, sparse and banded.
 Matrix Formulation
► For ‘C’ as non-symmetric and banded, the best way
of solving these system of linear equations is the
Gaussian elimination algorithm.
 Work Plan
Equation of Continuity
Simplification
Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Matrix Formulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Inverse Fourier Transform
►In order to find out potential in three-dimensional
space, the potential which has been calculated in the
Fourier space will have to go through the inverse Fourier
transform

►Solutions for a discrete set of -values are calculated

and used to evaluate the potential numerically.
► By choosing the optimum number and exact values of
, computational costs can be reduced and accurate
results can be found for less computer time.
 Inverse Fourier Transform
►By testing on different geological models, Cardelli,
E., and Fischanger, F., 2006, decided to use 10 discrete
values for , which are specially valid for dipole-dipole
array and were inversely proportional to the smallest
spacing D between adjacent electrodes, i.e.
= {0.00.010 0.110 0.265 0.425 0.625 1.250 1.750 2.500
4.400}/D.
► Potentials have been calculated simultaneously at all
the nodes for multiple source injections at 9 discrete
optimal values of .
 Potential Distribution
►Each set φ(x, , z) for the 9 optimal values of has been
calculated. A 2-point quadrature formula is then used for
the evaluation of the integral in each interval; these values
are then summed to give
►Integration of the equation has been calculated using
trapezoidal rule for non-uniform grid and is given by

By using the above formula, ϕ(x,y,z) can be calculated

from
 Potential Distribution

► Numerical Example
grid using the horizontal and vertical values
0

-20 0.018

-40 0.016
variation in vertical direction

-60 0.014

0.012
-80

0.01
-100
0.008
-120
0.006
-140
0.004
-160
0.002
-180

potential plot at two current electrode

0 200 400 600 800
variation in horizontal direction Conductivity
positions for a vertical dyke
Earth Model.
 Work Plan
Equation of Continuity Simplification

Mathematical model

Fourier Transform

Discretization of the problem

and BCs

Numerical algorithms and

implementation

Matrix Formulation

Inverse Fourier Transform Potential Distribution ϕ

Pseudo-section
 Pseudo-section
► Plot the data from a 2-D imaging survey, the pseudo-
section contouring plot is normally used.
►Gives a very approximate picture of the true sub-
surface resistivity distribution but a distorted picture of
the subsurface as the shapes of the contours depend
on the type of array used as well as the true subsurface
resistivity.
 Pseudo-section
pseudo-section plot
0

-20

-40

-60
ps eudo-depth(m )

-80

-100

-120

-140

-160

-180
-500 -400 -300 -200 -100 0 100 200 300 400 500

Horizontal data coverage for Dipole-Dipole array

spread(m)
 Numerical Results and Discussions
►Model 1
Homogeneous half-space of resistivity 1000 ohm-m and the
spread length is 1000m.

Earth model

Pseudo resistivity section for Dipole-Dipole array

 Numerical Results and Discussions
►Model 2
Homogeneous half-space of resistivity 1000 ohm-m and the spread
length is 1000m.

Earth model

Pseudo resistivity section for Dipole-Dipole array

 Numerical Results and Discussions

►Model 3
An assumed model of a vertical two dimensional dyke like
structure having length 20m and width 40m is present 10m below
from the surface. The resistivity of the body is 100 ohm-m and
the background resistivity is 10 ohm-m.

Earth model Pseudo resistivity section for

Dipole-Dipole array
 Remember
First
Assignment

Flow
Chart
of
2D
Inversion
Steps
Field data

Computed Data

Diff. btw Field

and computed Inverted Resistivity model
Data

Inverted Resistivity Model

Inversion approach-1

Inversion approach-2