Interpolating a Surface

From
Sampled Point Data
Interpolating a Surface From Sampled Point Data
Assumes a continuous surface that is sampled
Interpolation
Estimating the attribute values of locations that
are within the range of available data using
known data values
Extrapolation
Estimating the attribute values of locations
outside the range of available data using known
data values
Interpolating a Surface From Sampled Point Data
Interpolation
Estimating a point
here: interpolation

Sample
data
Interpolating a Surface From Sampled Point Data
Extrapolation
Sample
data

Estimating a point
here: extrapolation
Interpolating a Surface From Sampled Point Data

Sampling Strategies for Interpolation

Regular Sampling Random Sampling

Interpolating a Surface From Sampled Point Data
Linear Interpolation
Sample If
elevation data A = 8 feet and

A B = 4 feet
then
C
C = (8 + 4) / 2 = 6 feet
B

Elevation profile
Interpolating a Surface From Sampled Point Data
Non-Linear Interpolation
Sample Often results in a
elevation data more realistic
interpolation but
A
estimating missing
data values is more
C complex
B

Elevation profile
Interpolating a Surface From Sampled Point Data
Global Interpolation
Uses all known sample points to estimate a value at an
unsampled location

Sample
data
Interpolating a Surface From Sampled Point Data
Local Interpolation
Uses a neighborhood of sample points to estimate a value at
an unsampled location

Sample
data

Uses a local neighborhood to

estimate value, i.e. closest n
number of points, or within a
given search radius
Trend Surface
 Trend Surface
Global method
Inexact
Can be linear or non-linear
predicting a z elevation value [dependent
variable] with x and y location values
[independent variables]
 Trend Surface
1st Order Trend Surface

In one dimension: z varies as a linear function of x

z = b0 + b 1x + e z

x
 Trend Surface
1st Order Trend Surface

In two dimensions: z varies as a linear function of x and y

z
y

z = b0 + b 1 x + b 2 y + e

x
Trend Surface
Inverse Distance Weighted
(IDW)
 Inverse Distance Weighted

Local method
Exact
Can be linear or non-linear

The weight (influence) of a sampled data

value is inversely proportional to its
distance from the estimated value
 Inverse Distance Weighted
(Example)
n z 
 i

 
p
i 1
d
z ( x, y )  i
or z ( x, y )     z     1
with

1 
i i i
n

 
d 
p
i 1
i

100
IDW:
4 Closest 3
3 160 neighbors,
r=2
2
200
 Inverse Distance Weighted
(Example)
Weights
A 1 / (42) = .0625
B 1 / (32) = .1111
C 1 / (22) = .2500

A = 100

4
B = 160
3

2
C = 200
 Inverse Distance Weighted
(Example)
Weights Weights * Value
A 1 / (42) = .0625 .0625 * 100 = 6.25
B 1 / (32) = .1111 .1111 * 160 = 17.76
C 1 / (22) = .2500 .2500 * 200 = 50.00
Total = .4236 A = 100

4
6.25 +17.76 + 50.00 = 74.01
B = 160
3
74.01 / .4236 = 175
2
C = 200
Geostatistics
 Geostatistics
 Geostatistics:The original purpose of geostatistics
centered on estimating changes in ore grade within a
mine.
 The principles have been applied to a variety of areas in
geology and other scientific disciplines.

 A unique aspect of geostatistics is the use of

regionalized variables which are variables that fall
between random variables and completely deterministic
variables.
 Geostatistics
 Regionalized variables describe phenomena
with geographical distribution (e.g.
elevation of ground surface).

 The phenomenon exhibit spatial continuity.

 Geostatistics
 It is notalways possible to sample every location.
 Therefore, unknown values must be estimated
from data taken at specific locations that can be
sampled.

 The size, shape, orientation, and spatial

arrangement of the sample locations are termed
the support and influence the capability to predict
the unknown samples.
Semivariance
 Semivariance
 Regionalized variable theory uses a related
property called the semivariance to express
the degree of relationship between points on a
surface.

 The semivariance is simply half the

variance of the differences between all
possible points spaced a constant distance apart.
Semivariance is a measure of the degree of spatial
dependence between samples (elevation(
 Semivariance
 semivariance :The magnitude of the
semivariance between points depends on the
distance between the points. A smaller distance
yields a smaller semivariance and a larger
distance results in a larger semivariance.
 Calculating the Semivariance
(Regularly Spaced PointsRegularly Spaced Points(
 Consider regularly spaced points distance (d) apart, the
semivariance can be estimated for distances that are
multiple of (d) (Simple form):

1 Nh

 ( h)   (z  z ) i ih
2

2N h
i 1
 Semivariance
1 Nh

 ( h)   (z  z )i ih
2

2Nh
i 1

 Zi is the measurement of a regionalized variable

taken at location i ,
 Zi+h is another measurement taken h intervals
away d
 Nh is number of separating distance = number of
points –Lag (if the points are located in a single
profile)
 Calculating the Semivariance
(Irregularly Spaced PointsRegularly Spaced Points(
 Here we are going to explore directional variograms.
 Directional variograms is defines the spatial variation
among points separated by space lag h.
 The difference from the omnidirectional variograms is that
h is a vector rather than a scalar. For example, if
d={d1,d2}, then each pair of compared samples should be
separated in E-W direction and in S-N direction.
 Calculating the Semivariance
(Irregularly Spaced PointsRegularly Spaced Points(
 In practice, it is difficult to find enough sample points
which are separated by exactly the same lag vector [d].
 The set of all possible lag vectors is usually partitioned into
classes
Variogram
 Variogram

 The plot of the semivariances as a function of

distance from a point is referred to as a
semivariogram or variogram.
 Variogram
 The semivariance at a distance d = 0 should be zero,
because there are no differences between points that are
compared to themselves.
 However, as points are compared to increasingly distant
points, the semivariance increases.
 Variogram
 The range is the greatest distance over which the value at a
point on the surface is related to the value at another point.
 The range defines the maximum neighborhood over which
control points should be selected to estimate a grid node.
 Variogram (Models(

 It is a ‘model’ semi-variogram and is usually called the

spherical model.
 a is called the range of influence of a sample.
 C is called the sill of the semi-variogram.

 3 h 1 h  3

C    whereh  a
 ( h)    2 a 2 a  3

C whereh  a
 Variogram (Models(
Exponential Model

γ(h)  C 1  e  h a


spherical and exponential with the same spherical and exponential with the same
range and sill sill and the same initial slope
 Kriging
Interpolation
 Kriging Interpolation

 Kriging is named after the South African

engineer, D. G. Krige, who first developed the
method.

 Kriging uses the semivariogram, in calculating

estimates of the surface at the grid nodes.
 Kriging Interpolation
 The procedures involved in kriging incorporate
measures of error and uncertainty when determining
estimations.
 In the kriging method, every known data value and
every missing data value has an associated variance. If
‘C’ is constant (i.e. known value exactly), its variance
is zero.

 Based on the semivariogram used, optimal

weights are assigned to known values in order to
calculate unknown ones. Since the variogram
changes with distance, the weights depend on the
known sample distribution.
Ordinary Kriging
 Ordinary Kriging
 Ordinary kriging is the simplest form of kriging.
 It uses dimensionless points to estimate other
dimensionless points, e.g. elevation contour
plots.
 In Ordinary kriging, the regionalized variable is
assumed to be stationary.
 Punctual (Ordinary) Kriging
 In our case Z, at point p, Ze (p) to be calculated
using a weighted average of the known values
or control points:

z ( p)   w  z ( p )
e i i

 This estimated value will most likely differ from the actual
value at point p, Za(p), and this difference is called the
estimation error:
  z ( p)  z ( p)
p e a
 Punctual (Ordinary) Kriging
 If no drift exists and the weights used in the
estimation sum to one, then the estimated value
is said to be unbiased. The scatter of the
estimates about the true value is termed the
error or estimation variance,

[ z e
( p i )  z a ( p i )] 2
i
σ 
2
z
i 1

n
 Punctual (Ordinary) Kriging
 kriging tries to choose the optimal weights that
produce the minimum estimation error .
 Optimal weights, those that produce unbiased
estimates and have a minimum estimation variance, are
obtained by solving a set of simultaneous equations .

w1 γ(h11 )  w2 γ(h12 )  w3 γ(h13 )  γ(h1 p )

w1 γ(h21 )  w2 γ(h22 )  w3 γ(h23 )  γ(h2 p )
w1 γ(h31 )  w2 γ(h32 )  w3 γ(h33 )  γ(h3 p )

w1  w2  w3  1
 Punctual (Ordinary) Kriging
 A fourth variable is introduced called the Lagrange
multiplier
w1 (h11 )  w2  (h12 )  w3  (h13 )     (h1 p )
w1 (h21 )  w2  (h22 )  w3  (h23 )     (h2 p )
w1 (h31 )  w2  (h32 )  w3  (h33 )     (h3 p )
w1  w2  w3  1

  (h )  (h )  (h ) 1  w    (h
11 12 13 1 1p
)
  (h )  (h )  (h ) 1  w    (h )
 21 21
   
23 2 2p

  (h )  (h )  (h ) 1  w    (h
31 32 33 3 3p
)
     
 1 1 1 0     1 
 Punctual (Ordinary) Kriging
 Once the individual weights are known, an estimation
can be made by

ze ( p)  w1 z1  w2 z 2  w3 z3
 And an estimation variance can be calculated by

σ  w1γ(h1 p )  w2 γ(h21 p )  w13 γ(h3 p )  λ

2
z