SlideShare uma empresa Scribd logo
1 de 21
1
Chapter 11 – Kriging
Kriging is a spatial prediction method of nice statistical properties: BLUE (“best linear unbiased
estimator”). The method was first developed by G. Matheron in 1963, two volumes published in
French. Matheron named the method after the South African mining engineer, D.G. Krige, who in
the 50’s developed methods for determining ore grades, although the specific prediction method of
Matheron has not much to do with Krige (see Cressie 1990 for the history).
Kriging shares the same weighted linear combination estimator as those given in the last chapter:
where zi is the sample value at location i, wi is a weight, n is the number of samples.
As we will show next that estimators of the above form are unbiased if the sum of the weights is 1.
The distinguishing feature of kriging, therefore, is its aim of minimizing the error variance.
Many kriging methods have been developed for different prediction purposes, e.g., block kriging,
universal kriging, cokrigin, etc. Here we will only concentrate on the most basic one: ordinary
kriging.
* Cressie, N. 1990. The origins of kriging. Mathematical Geology 22:239-252.
* Diggle, P.J. & Tawn, J.A. 1998. Model-based geostatistics (with Discussion). Applied Statistics 47:299-350.
∑=
=
n
i
iizwz
1
ˆ
2
Kriging - unbiasedness
Assume we have a model:
Z(s) = µ + ε(s),
where ε(s) is a zero mean second-order stationary random field with covariogram function C(h)
and variogram γ(h). Also σ2
=C(0).
The weighted linear estimator for location s0 is: (*)
The estimation error at location s0 is the difference between the predictor and the random variable
modeling the true value at that location:
The bias is:
So, as long as the weighted linear estimator (*) is unbiased. All the methods in
chapter 10 meet this condition, thus are unbiased. However, the unbiasedness tells us nothing
about how to determine the weights wi’s.
∑=
=
n
i
iiZwZ
1
0
ˆ
∑ −=−= 0000
ˆ ZZwZZR ii
( ) ∑ −=∑ −= )()()( 000 ZEZwEZZwERE iiii
( )∑ −=∑ −=∑ −= 1)()( 0 iiii wwZEZEw µµµ
,1∑ =iw
3
Minimizing error variance
Kriging is such a method that determines the weights so that the mean squared error (MSE) is
minimized:
subject to the unbiasedness constraint
Once we have chosen a data generating model (through a covariogram or variogram), the
minimization of MSE can be achieved by setting the n partial first derivatives to 0, then the n
weights wi’s can be obtained by solving the n simultaneous equations. However, this procedure
does not quite work for our problem because we only want the solutions that meet the
unbiasedness condition.
( )2
00 )ˆ( ZZEMSE −=
.1∑ =iw
( ) ( )00
2
00
ˆvar)ˆ( ZZZZE −=−
),ˆcov(2)var()ˆvar( 0000 ZZZZ −+=
( ) ),cov(2)var(var 00 ZZwZZw iiii ∑−+∑=
0
2
2 ii
i j
ijji CwCww ∑−+∑∑= σ
4
Minimizing error variance using the Lagrange multiplier
The Lagrange multiplier is a useful technique for converting a constrained minimization problem
into an unconstrained one.
Take the first term w1 as an example:
The final ordinary kriging system is:
( )∑ −+∑−+∑∑= 122 0
2
iii
i j
ijji wCwCwwMSE λσ
0222 10
1
1
1
=+−∑=
∂
∂
=
λCCw
w
MSE n
j
jj
10
1
1 CCw
n
j
jj =+∑
=
λ












=












•












1
......
01...1
1...
............
1...
0
101
1
111
nnnnn
n
C
C
w
w
CC
CC
λ
C w = D
(n+1)×(n+1) (n+1)×1 (n+1)×1
w = C
-1
D
(**)
5
Estimating the variance of errors
Because kriging predictor is unbiased, the variance of the prediction errors is just the MSE:
The first term on the right hand side – From the the equation of the first derivative (i.e., the (**)
equation on the previous page), we have
Therefore, the error variance is of the form:
The variance is often called the ordinary kriging variance, expressed in a matrix form:
Note: σ2
is simply C(0).
0
2
2 ii
i j
ijji CwCwwMSE ∑−+∑∑= σ
.)( 00 ∑ −=∑ −=∑ ∑=∑∑
i
ii
i
ii
i j
ijji
i j
ijji CwCwCwwCww λλ
λσ −∑−= 0
2
iiCwMSE
DwOK '22
−=σσ
6
Interpretation of kriging
The kriging system may be better understood through the following intuitive interpretation. Two steps are
involved in determining the linear weight of kriging:
1. The D vector provides a weighting scheme similar to that of the inverse distance method. The higher the
covariance between a sample (denoting i = 1, 2, …, n) and the location being estimated (denoting 0), the
more that sample would contribute to the estimation. Like an inverse distance method, the covariance
(thereof weight) between sample i and location 0 generally decreases as the sample gets farther away.
Therefore, D vector contains a type of inverse distance weighting in which the “distance” is not the
geometric distance to the estimating sample but a statistical distance.
2. What really makes kriging differ from the inverse distance method is the C matrix. The multiplication of D
by C-1
does more than simply rescale D so that w sums to 1. C records (covariance) distances between all
sample pairs, providing the OK system with information on the clustering of the available sample data. So
C helps readjust the sample weight according to their clustering. Clustered samples will be declustered by
C. Therefore, OK system takes into account of two important aspects of estimation problem: distance and
clustering.












=












•












1
......
01...1
1...
............
1...
0
101
1
111
nnnnn
n
C
C
w
w
CC
CC
λ
C w = D
(n+1)×(n+1) (n+1)×1 (n+1)×1
w = C
-1
D
7
Ordinary kriging in terms of variogram γ(h)
In practice, kriging is usually implemented using variogram rather than covariogram because it
has better statistical properties (unbiased and consistent). From chapter 9 (page 6): γ(h) = C(0) -
C(h), we have C(h) = C(0) - γ(h). Substituting this covariogram into the unconstrained MSE on
page 4 leads to
Similar to the covariogram, the weights can be solved by setting the equations of the 1st
derivatives w.r.t. wi’s to zero. The final kriging equation in matrix notation is:
( )∑ −+−∑−+∑∑ −= 12)(2)( 0
222
iii
i j
ijji wwwwMSE λγσσγσ
( ).122 0 ∑ −+∑+∑∑−= iii
i j
ijji wwww λγγ
Γ w = D
(n+1)×(n+1) (n+1)×1 (n+1)×1
w = Γ -1
D














=














−
•
















1
......
01...11
10...
...............
1...0
1...0
0
20
10
2
1
21
221
112
nnnn
n
n
w
w
w
γ
γ
γ
λ
γγ
γγ
γγ
8
Ordinary kriging variance in terms of variogram γ(h)
Following the same steps as for the variance based on the covariogram, we have the ordinary
kriging variance in terms of variogram:
where w and D are the vectors given on the previous page.
Steps for kriging –
1. EDA exploration, removing trend, checking for stationarity and isotropy
2. Computing the empirical variogram
3. Fitting and selecting a theoretical variogram model to the empirical variogram
4. Computing the weight w using the fitted theoretical variogram, i.e., kriging.
5. Predicting the values at the locations of interest
6. Validation
7. Plotting kriging surfaces
,'0
2
Dww iiOK ∑ =−= λγσ
9
Checking and removing trends (make the data stationary)
Example: soil pH value in the Gigante plot of Panama, using the full data set (soil.dat, has 349 data
points).
The data appear to have a trend in the northwest-southeastern direction. To remove such a trend, we
fit the data using using model: z = 5.67 - 0.003295x + 0.001025y + 4.521e-6x2
+ ε. Terms y2
and x×y
are not significant. It seems that the trend surface analysis has detrended the data.
0 100 200 300 400 500
0200400600800
0 100 200 300 400 5000200400600800
Before detrended After detrended
High
Low
10
Has the trend really been removed?
We further check it using variograms. The comparison of the variograms before and after detrending
confirms that there is no trend in the residuals. We are confident that the residuals of the trend surface
analysis are likely stationary. We can now go on to do kriging.
>soil.geodat=as.geodata(soil.dat,coords.col=2:3,data.col=5,borders=T)
>variog.b0=variog(soil.geodat,uvec=seq(0,500,by=5), max.dist=500)
>plot(variog.b0)
>variog.b2=variog(soil.geodat,uvec=seq(0,500,by=5),trend="2nd",max.dist=500)
>plot(variog.b2)
11
Spherical model
Logistic model
Fitting a variogram
Several variogram models can be fitted to the data. For illustration purpose, only two models
(the spherical and logistic models) are shown here. By visual inspection, it seems that the
logistic model may capture the spatial autocorrelation better than the spherical model,
particularly at short distance lag. However, the sigmoid shape of the logistic model may not
reflect the intrinsic feature of the data. We will use the spherical model for kriging.
2
2
0009258.0
0001072.0
1065.0)(
h
h
h +=γ
Logistic model:
12
R implementation using geoR- ordinary kriging
1. Compute variogram by directly considering trend (i.e., removing 2nd
order trend. Kriging will
automatically put back the trend in the final prediction):
>variog.b2=variog(soil.geodat,uvec=seq(0,500,by=5),trend="2nd",max.dist=500)
2. Model variogram using spheric variogram model:
>pH.sph=variofit(variog.b2,cov.model="spherical")
>pH.sph # also try summary(pH.sph) to see the output
variofit: model parameters estimated by WLS (weighted least squares):
covariance model is: spherical
parameter estimates:
tausq sigmasq phi
0.1240 0.0955 130.7104
3. Fitting logistic model:
> u=variog.b2$u;v=variog.b2$v
> logist.nls=nls(v~c0+a*u^2/(1+b*u^2),start=c(c0=0.05,a=0.25,b=0.1))
>logist.nls
model: y ~ c0 + a * x^2/(1 + b * x^2)
data: parent.frame()
c0 a b
0.1064792 0.0001072 0.0009258
residual sum-of-squares: 0.04605
nugget (c0) c1 range
for 0 < h ≤ 130.71
for h ≥ 130.71






−+= 3
)
71.130
(
2
1
71.1302
3
0955.01240.0)(
hh
hγ
0955.01240.0)( +=hγ
2
2
0009258.0
0001072.0
1065.0)(
h
h
h +=γ
Logistic model:
13
R implementation - ordinary kriging
1. Generate locations at which interpolation is needed:
>x=soil.dat$gx; y=soil.dat$gy
>prd.loc=expand.grid(x=sort(unique(x)),y=sort(unique(y)))
2. Run krige.conv for spatial interpolation:
>pH.prd=krige.conv(soil.geodat,loc=prd.loc,krige=krige.control(cov.model="spherical",cov.pars
=c(0.09549404,130.71043698)))
3. View the prediction: You can directly apply image to pH.prd.
Here we want to have more control over the features of the image,
We create matrix for the pH.prd$predict and then apply image:
>pH.prd.mat=matrix(pH.prd$predict,byrow=T,ncol=84)
>image(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.mat),
xlim=c(-20,500),ylim=c(-20,820),xlab="x",ylab="y")
>lines(gigante.border,lwd=2,col=“green”)
>contour(pH.prd,add=T)
We can do the same thing to view the variation in the prediction:
pH.prd$krige.var. Taking the squared root, it is prd.se.
>pH.prd.se.mat=matrix(sqrt(pH.prd$krige.var),
byrow=T,ncol=84)
>image(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.se.mat),
xlim=c(-20,500),ylim=c(-20,820),xlab="x",ylab="y")
>lines(gigante.border,lwd=2)
14
Plot prediction variance
It is desirable to view the variation of the prediction: pH.prd$krige.var. Taking the squared root for
prd.se.
>pH.prd.se.mat=matrix(sqrt(pH.prd$krige.var),byrow=T,ncol=84)
>image(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.se.mat),xlim=c(-20,500),
ylim=c(-20,820),xlab="x",ylab="y")
>lines(gigante.border,lwd=2,col="blue")
>contour(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.se.mat),xlim=c(-20,500),
ylim=c(-20,820),add=T)
pH surface pH std error surface pH std error with contour
15
1. Independent data validation:
Compare the predicted with the observed
data. As shown in the left table, these 13
data samples were not included in the
kriging analysis. The predictions were
generated from:
>pH.prd13=krige.conv(soil.geodat,loc=
prd.loc13,krige=krige.control(cov.model=
"spherical",cov.pars=c(0.09549,130.71043)))
2. Cross-validation:
Deleting one observation each time from
the data set and then predicting the deleted
observation using the remaining
observations in the data set. This process is
repeated for all observations. Residuals are
then analyzed using standard techniques of
regression analysis to check the underlying
model assumptions.
Evaluating the outputs
of kriging prediction:
SrfID gx gy Site pH pred var se.fit
31 240 8 site8mN 4.92 4.304 0.0088 0.0936
123 240 232 site8mS 5.97 5.933 0.0109 0.1046
124 240 220 site20mS 5.57 5.762 0.0203 0.1426
128 300 260 site20mN 5.41 5.375 0.0198 0.1406
190 358 420 site2mW 5.28 5.268 0.0033 0.0573
242 362 540 site2mE 5.23 5.437 0.0039 0.0627
243 368 540 site8mE 5.04 5.343 0.0104 0.1020
290 238 600 site2mW 4.63 5.086 0.0042 0.0647
291 232 600 site8mW 5.45 5.177 0.0145 0.1202
292 220 600 site20mW 5.4 5.371 0.0254 0.1595
310 422 660 site2mE 5.33 5.545 0.0033 0.0573
312 440 660 site20mE 6.17 5.747 0.0185 0.1361
360 478.6 538.6 site2mSW 5.55 5.310 0.0033 0.0572
16
Block kriging
In many occasions, we are interested in estimating the value in a block (cell) rather than that
at a point. The block kriging system is similar to that of the
OK, of the form:
where i.e., the covariogram between
block A and sample point i is the average of the covariograms between the points
locating within A and i.
The block kriging variance is: where
+
+
+
+
+
+
Block A
. .
. .
+ observed samples
. regularly spaced locations
set up within the block












=












•












1
......
01...1
1...
............
1...
11
1
111
nA
A
nnnn
n
C
C
w
w
CC
CC
λ
C w = D
(n+1)×(n+1) (n+1)×1 (n+1)×1
,
1
∑=
∈Aj
ijiA C
A
C
,'2
DwCAAOK −=σ .
1
2 ∑ ∑=
∈ ∈Ai Aj
ijAA C
A
C
17
R implementation - block kriging
Block kriging is achieved by using OK:
1. Create a systematical grid lattice (as dense as you want) using expand.grid.
2. Use krige.conv for OK to do spatial interpolation for the grids.
3. Average the values of those grids falling within the defined block.
+
+
+
+
+
+
Block A
. .
. .
+ observed samples
. regularly spaced locations
set up within the block
18
Spatial estimation: additive/nonadditive variables
Some precaution is necessary before applying geostatistical analysis to your data. The method
does not universally apply to any type of data.
Additive variable:
Nonadditive variable:
Nonadditive variables include: number of species in a block, ratio data (e.g., number of cars
per household in a city block). Geostatistics is invalid for analyzing nonadditive variables
because subtraction makes no sense here.
5 balls 3
7 4
8
11
scaled up
3 colors
(1b, 2r, 2 w)
5 colors
4 colors
scaled up
3 colors
(1b, 1g, 1y)
1 color
(4g)
3 colors
(3b, 2r, 2 w)
19
 
Spatial estimation: scale effect
Few spatial data (point process is an exception) can avoid the problem of the size of sample area 
(called support in geostat, or modifiable areal unit in geography, or grain size in landscape 
ecology).
In many practical applications, the support of the samples is not the same as the support of the 
estimates we are trying to calculate. For example, when assessing gold ore grades in an area, we 
take samples from drill hole cores, but in mining operation we treat truckloads as the size of 
sample (consider a truckload either as ore or as waste).
So a critical and difficult question is: can we infer about the properties of a variable at different 
levels of supports from the observations sampled at a particular support? In other words, can we 
scale down or up a spatial process?
0 1 2 3 4
010203040
rg am m a(100, 1)
0 2 4 6 8
0510152025
rg am m a(100, 3)
?
20
 
Grain
size (m)
No. stems/m2
(std. error)
No. species/m2
(std. error)
5×5 0.671 (0.244) 0.585 (0.197)
10×10 0.671 (0.167) 0.475 (0.095)
20×20 0.671(0.130) 0.318 (0.038)
25×25 0.671 (0.121) 0.267 (0.026)
50×50 0.671  (0.100) 0.129 (0.008)
100×100 0.671 (0.085) 0.049 (0.001)
250×250 0.671 (0.048) 0.011 (0.0004)
500×500 0.671 (0.041) 0.003 (< 0.001)
500×1000 0.671 0.0016
Spatial estimation: scale effect
Number of stems and number of species per 
m2
 at different sampling scales (grain size) in 
a 1000×500 m rain forest of Malaysia. The 
entire plot has 335,356  trees belonging to 814 
species. The densities at each grain size were 
computed as follows: (1) divide the plot into a 
grid system using a given scale (e.g., 5×5 m), 
(2) count the total number of stems and the 
number of species in each cells, respectively, 
(3) average these two quantities across all the 
cells, and (4) then divide the averages by the 
scale. 
The results clearly show how sampling scale 
profoundly affects the species diversity. They 
suggest that diversity based on per unit area 
(the last column) is a misleading measurement 
for comparing diversity between two 
ecosystems.
21
 
Spatial estimation: scale effect
“This problem of the discrepancy between the support of our samples and the 
intended support of our estimates is one of the most difficult we face in 
estimation.”
Isaaks & Srivastava (1989, page 193)

Mais conteúdo relacionado

Mais procurados

Spatial analysis and modeling
Spatial analysis and modelingSpatial analysis and modeling
Spatial analysis and modeling
Tolasa_F
 
Inverse distance weighting
Inverse distance weightingInverse distance weighting
Inverse distance weighting
Penchala Vineeth
 
Fundamentals of Remote Sensing
Fundamentals of Remote Sensing Fundamentals of Remote Sensing
Fundamentals of Remote Sensing
Pallab Jana
 

Mais procurados (20)

Geodetic systems (earth, ellipsoid)
Geodetic systems (earth, ellipsoid)Geodetic systems (earth, ellipsoid)
Geodetic systems (earth, ellipsoid)
 
Spot satellite
Spot satelliteSpot satellite
Spot satellite
 
Thermal remote sensing and its applications
Thermal remote sensing and its applicationsThermal remote sensing and its applications
Thermal remote sensing and its applications
 
Spatial Autocorrelation
Spatial AutocorrelationSpatial Autocorrelation
Spatial Autocorrelation
 
Spatial analysis and modeling
Spatial analysis and modelingSpatial analysis and modeling
Spatial analysis and modeling
 
Interpolation techniques in ArcGIS
Interpolation techniques in ArcGISInterpolation techniques in ArcGIS
Interpolation techniques in ArcGIS
 
Inverse distance weighting
Inverse distance weightingInverse distance weighting
Inverse distance weighting
 
GIS Map Projection
GIS Map ProjectionGIS Map Projection
GIS Map Projection
 
GIS - Topology
GIS - Topology GIS - Topology
GIS - Topology
 
Projections and coordinate system
Projections and coordinate systemProjections and coordinate system
Projections and coordinate system
 
Plateform and sensors of remote sensing
Plateform and sensors of remote sensingPlateform and sensors of remote sensing
Plateform and sensors of remote sensing
 
GIS
GISGIS
GIS
 
Datum
DatumDatum
Datum
 
Fundamentals of Remote Sensing
Fundamentals of Remote Sensing Fundamentals of Remote Sensing
Fundamentals of Remote Sensing
 
Geographical information system
Geographical information systemGeographical information system
Geographical information system
 
georeference
georeferencegeoreference
georeference
 
Map Projections ―concepts, classes and usage
Map Projections ―concepts, classes and usage Map Projections ―concepts, classes and usage
Map Projections ―concepts, classes and usage
 
Scale and resolution
Scale and resolutionScale and resolution
Scale and resolution
 
Relief displacement
Relief displacementRelief displacement
Relief displacement
 
Introduction to remote sensing and gis
Introduction to remote sensing and gisIntroduction to remote sensing and gis
Introduction to remote sensing and gis
 

Destaque

Student_Garden_geostatistics_course
Student_Garden_geostatistics_courseStudent_Garden_geostatistics_course
Student_Garden_geostatistics_course
Pedro Correia
 
The Correlogram Explained
The Correlogram ExplainedThe Correlogram Explained
The Correlogram Explained
Ed Isaaks
 
Functional dependencies
Functional dependenciesFunctional dependencies
Functional dependencies
Kumar
 
Yam regev - hard core local seo tips & tricks
Yam regev  - hard core local seo tips & tricksYam regev  - hard core local seo tips & tricks
Yam regev - hard core local seo tips & tricks
Barry Schwartz
 
Mazapanes bañados en chocolate alejandro botero y tomas mesa 8 c
Mazapanes bañados en chocolate alejandro botero y tomas mesa 8 cMazapanes bañados en chocolate alejandro botero y tomas mesa 8 c
Mazapanes bañados en chocolate alejandro botero y tomas mesa 8 c
tomasmesagiraldo
 
Herramientas de comunicación web 2.0 en la dirección de proyectos cio 2009 ...
Herramientas de comunicación web 2.0 en la dirección de proyectos   cio 2009 ...Herramientas de comunicación web 2.0 en la dirección de proyectos   cio 2009 ...
Herramientas de comunicación web 2.0 en la dirección de proyectos cio 2009 ...
Ramon Costa i Pujol
 
Diarrea aguda
Diarrea agudaDiarrea aguda
Diarrea aguda
Ana Prado
 

Destaque (19)

Student_Garden_geostatistics_course
Student_Garden_geostatistics_courseStudent_Garden_geostatistics_course
Student_Garden_geostatistics_course
 
The Correlogram Explained
The Correlogram ExplainedThe Correlogram Explained
The Correlogram Explained
 
Resumen de los orígenes y evolución de las redes
Resumen de los orígenes y evolución de las redesResumen de los orígenes y evolución de las redes
Resumen de los orígenes y evolución de las redes
 
Como crear un perfil en kiwibox
Como crear un perfil en kiwiboxComo crear un perfil en kiwibox
Como crear un perfil en kiwibox
 
Future of the Contact Centre
Future of the Contact CentreFuture of the Contact Centre
Future of the Contact Centre
 
Agradecimientos a mis compañeros y maestros
Agradecimientos a mis compañeros y maestrosAgradecimientos a mis compañeros y maestros
Agradecimientos a mis compañeros y maestros
 
Present simple
Present simplePresent simple
Present simple
 
Functional dependencies
Functional dependenciesFunctional dependencies
Functional dependencies
 
Origen del lenguaje
Origen del lenguajeOrigen del lenguaje
Origen del lenguaje
 
Warp #2 benedikt germanier - building a brand
Warp #2   benedikt germanier - building a brandWarp #2   benedikt germanier - building a brand
Warp #2 benedikt germanier - building a brand
 
Analisi Social Tv - The Apprentice 2
Analisi Social Tv - The Apprentice 2Analisi Social Tv - The Apprentice 2
Analisi Social Tv - The Apprentice 2
 
Yam regev - hard core local seo tips & tricks
Yam regev  - hard core local seo tips & tricksYam regev  - hard core local seo tips & tricks
Yam regev - hard core local seo tips & tricks
 
Googol pta780 g1a 200kw 250kva
Googol pta780 g1a 200kw 250kva Googol pta780 g1a 200kw 250kva
Googol pta780 g1a 200kw 250kva
 
Curculum Vitae English
Curculum Vitae EnglishCurculum Vitae English
Curculum Vitae English
 
Mazapanes bañados en chocolate alejandro botero y tomas mesa 8 c
Mazapanes bañados en chocolate alejandro botero y tomas mesa 8 cMazapanes bañados en chocolate alejandro botero y tomas mesa 8 c
Mazapanes bañados en chocolate alejandro botero y tomas mesa 8 c
 
Herramientas de comunicación web 2.0 en la dirección de proyectos cio 2009 ...
Herramientas de comunicación web 2.0 en la dirección de proyectos   cio 2009 ...Herramientas de comunicación web 2.0 en la dirección de proyectos   cio 2009 ...
Herramientas de comunicación web 2.0 en la dirección de proyectos cio 2009 ...
 
Tics y Ntics
Tics y NticsTics y Ntics
Tics y Ntics
 
Diarrea aguda
Diarrea agudaDiarrea aguda
Diarrea aguda
 
Lumitek pro pt
Lumitek pro ptLumitek pro pt
Lumitek pro pt
 

Semelhante a Ch11.kriging

matlab program for Smith chart
matlab program for Smith chartmatlab program for Smith chart
matlab program for Smith chart
Dnyanesh Patil
 

Semelhante a Ch11.kriging (20)

A Rapid Location Independent Full Tensor Gravity Algorithm
A Rapid Location Independent Full Tensor Gravity AlgorithmA Rapid Location Independent Full Tensor Gravity Algorithm
A Rapid Location Independent Full Tensor Gravity Algorithm
 
StructuralTheoryClass2.ppt
StructuralTheoryClass2.pptStructuralTheoryClass2.ppt
StructuralTheoryClass2.ppt
 
Steven Duplij, Raimund Vogl, "Polyadic Braid Operators and Higher Braiding Ga...
Steven Duplij, Raimund Vogl, "Polyadic Braid Operators and Higher Braiding Ga...Steven Duplij, Raimund Vogl, "Polyadic Braid Operators and Higher Braiding Ga...
Steven Duplij, Raimund Vogl, "Polyadic Braid Operators and Higher Braiding Ga...
 
New test123
New test123New test123
New test123
 
Cs229 notes9
Cs229 notes9Cs229 notes9
Cs229 notes9
 
PART I.3 - Physical Mathematics
PART I.3 - Physical MathematicsPART I.3 - Physical Mathematics
PART I.3 - Physical Mathematics
 
mds.pdf
mds.pdfmds.pdf
mds.pdf
 
matlab program for Smith chart
matlab program for Smith chartmatlab program for Smith chart
matlab program for Smith chart
 
Comparing the methods of Estimation of Three-Parameter Weibull distribution
Comparing the methods of Estimation of Three-Parameter Weibull distributionComparing the methods of Estimation of Three-Parameter Weibull distribution
Comparing the methods of Estimation of Three-Parameter Weibull distribution
 
Serr calculation
Serr calculationSerr calculation
Serr calculation
 
Propagation of Uncertainties in Density Driven Groundwater Flow
Propagation of Uncertainties in Density Driven Groundwater FlowPropagation of Uncertainties in Density Driven Groundwater Flow
Propagation of Uncertainties in Density Driven Groundwater Flow
 
Solution kepler chap 1
Solution kepler chap 1Solution kepler chap 1
Solution kepler chap 1
 
Regression.pptx
Regression.pptxRegression.pptx
Regression.pptx
 
Regression.pptx
Regression.pptxRegression.pptx
Regression.pptx
 
QMC: Transition Workshop - Density Estimation by Randomized Quasi-Monte Carlo...
QMC: Transition Workshop - Density Estimation by Randomized Quasi-Monte Carlo...QMC: Transition Workshop - Density Estimation by Randomized Quasi-Monte Carlo...
QMC: Transition Workshop - Density Estimation by Randomized Quasi-Monte Carlo...
 
Ch03 ssm
Ch03 ssmCh03 ssm
Ch03 ssm
 
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
 
MSE.pptx
MSE.pptxMSE.pptx
MSE.pptx
 
Different Types of Machine Learning Algorithms
Different Types of Machine Learning AlgorithmsDifferent Types of Machine Learning Algorithms
Different Types of Machine Learning Algorithms
 
A study on number theory and its applications
A study on number theory and its applicationsA study on number theory and its applications
A study on number theory and its applications
 

Último

Biogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune Waterworlds
Biogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune WaterworldsBiogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune Waterworlds
Biogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune Waterworlds
Sérgio Sacani
 
Chemical Tests; flame test, positive and negative ions test Edexcel Internati...
Chemical Tests; flame test, positive and negative ions test Edexcel Internati...Chemical Tests; flame test, positive and negative ions test Edexcel Internati...
Chemical Tests; flame test, positive and negative ions test Edexcel Internati...
ssuser79fe74
 
Seismic Method Estimate velocity from seismic data.pptx
Seismic Method Estimate velocity from seismic  data.pptxSeismic Method Estimate velocity from seismic  data.pptx
Seismic Method Estimate velocity from seismic data.pptx
AlMamun560346
 
Asymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 b
Asymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 bAsymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 b
Asymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 b
Sérgio Sacani
 
SCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptx
SCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptxSCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptx
SCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptx
RizalinePalanog2
 

Último (20)

module for grade 9 for distance learning
module for grade 9 for distance learningmodule for grade 9 for distance learning
module for grade 9 for distance learning
 
Proteomics: types, protein profiling steps etc.
Proteomics: types, protein profiling steps etc.Proteomics: types, protein profiling steps etc.
Proteomics: types, protein profiling steps etc.
 
Biogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune Waterworlds
Biogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune WaterworldsBiogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune Waterworlds
Biogenic Sulfur Gases as Biosignatures on Temperate Sub-Neptune Waterworlds
 
Clean In Place(CIP).pptx .
Clean In Place(CIP).pptx                 .Clean In Place(CIP).pptx                 .
Clean In Place(CIP).pptx .
 
High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...
High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...
High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...
 
Zoology 5th semester notes( Sumit_yadav).pdf
Zoology 5th semester notes( Sumit_yadav).pdfZoology 5th semester notes( Sumit_yadav).pdf
Zoology 5th semester notes( Sumit_yadav).pdf
 
PSYCHOSOCIAL NEEDS. in nursing II sem pptx
PSYCHOSOCIAL NEEDS. in nursing II sem pptxPSYCHOSOCIAL NEEDS. in nursing II sem pptx
PSYCHOSOCIAL NEEDS. in nursing II sem pptx
 
Vip profile Call Girls In Lonavala 9748763073 For Genuine Sex Service At Just...
Vip profile Call Girls In Lonavala 9748763073 For Genuine Sex Service At Just...Vip profile Call Girls In Lonavala 9748763073 For Genuine Sex Service At Just...
Vip profile Call Girls In Lonavala 9748763073 For Genuine Sex Service At Just...
 
Chemical Tests; flame test, positive and negative ions test Edexcel Internati...
Chemical Tests; flame test, positive and negative ions test Edexcel Internati...Chemical Tests; flame test, positive and negative ions test Edexcel Internati...
Chemical Tests; flame test, positive and negative ions test Edexcel Internati...
 
Seismic Method Estimate velocity from seismic data.pptx
Seismic Method Estimate velocity from seismic  data.pptxSeismic Method Estimate velocity from seismic  data.pptx
Seismic Method Estimate velocity from seismic data.pptx
 
9999266834 Call Girls In Noida Sector 22 (Delhi) Call Girl Service
9999266834 Call Girls In Noida Sector 22 (Delhi) Call Girl Service9999266834 Call Girls In Noida Sector 22 (Delhi) Call Girl Service
9999266834 Call Girls In Noida Sector 22 (Delhi) Call Girl Service
 
FAIRSpectra - Enabling the FAIRification of Spectroscopy and Spectrometry
FAIRSpectra - Enabling the FAIRification of Spectroscopy and SpectrometryFAIRSpectra - Enabling the FAIRification of Spectroscopy and Spectrometry
FAIRSpectra - Enabling the FAIRification of Spectroscopy and Spectrometry
 
Justdial Call Girls In Indirapuram, Ghaziabad, 8800357707 Escorts Service
Justdial Call Girls In Indirapuram, Ghaziabad, 8800357707 Escorts ServiceJustdial Call Girls In Indirapuram, Ghaziabad, 8800357707 Escorts Service
Justdial Call Girls In Indirapuram, Ghaziabad, 8800357707 Escorts Service
 
Asymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 b
Asymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 bAsymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 b
Asymmetry in the atmosphere of the ultra-hot Jupiter WASP-76 b
 
GBSN - Microbiology (Unit 3)
GBSN - Microbiology (Unit 3)GBSN - Microbiology (Unit 3)
GBSN - Microbiology (Unit 3)
 
Forensic Biology & Its biological significance.pdf
Forensic Biology & Its biological significance.pdfForensic Biology & Its biological significance.pdf
Forensic Biology & Its biological significance.pdf
 
Locating and isolating a gene, FISH, GISH, Chromosome walking and jumping, te...
Locating and isolating a gene, FISH, GISH, Chromosome walking and jumping, te...Locating and isolating a gene, FISH, GISH, Chromosome walking and jumping, te...
Locating and isolating a gene, FISH, GISH, Chromosome walking and jumping, te...
 
SCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptx
SCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptxSCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptx
SCIENCE-4-QUARTER4-WEEK-4-PPT-1 (1).pptx
 
Call Girls Alandi Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Alandi Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Alandi Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Alandi Call Me 7737669865 Budget Friendly No Advance Booking
 
STS-UNIT 4 CLIMATE CHANGE POWERPOINT PRESENTATION
STS-UNIT 4 CLIMATE CHANGE POWERPOINT PRESENTATIONSTS-UNIT 4 CLIMATE CHANGE POWERPOINT PRESENTATION
STS-UNIT 4 CLIMATE CHANGE POWERPOINT PRESENTATION
 

Ch11.kriging

  • 1. 1 Chapter 11 – Kriging Kriging is a spatial prediction method of nice statistical properties: BLUE (“best linear unbiased estimator”). The method was first developed by G. Matheron in 1963, two volumes published in French. Matheron named the method after the South African mining engineer, D.G. Krige, who in the 50’s developed methods for determining ore grades, although the specific prediction method of Matheron has not much to do with Krige (see Cressie 1990 for the history). Kriging shares the same weighted linear combination estimator as those given in the last chapter: where zi is the sample value at location i, wi is a weight, n is the number of samples. As we will show next that estimators of the above form are unbiased if the sum of the weights is 1. The distinguishing feature of kriging, therefore, is its aim of minimizing the error variance. Many kriging methods have been developed for different prediction purposes, e.g., block kriging, universal kriging, cokrigin, etc. Here we will only concentrate on the most basic one: ordinary kriging. * Cressie, N. 1990. The origins of kriging. Mathematical Geology 22:239-252. * Diggle, P.J. & Tawn, J.A. 1998. Model-based geostatistics (with Discussion). Applied Statistics 47:299-350. ∑= = n i iizwz 1 ˆ
  • 2. 2 Kriging - unbiasedness Assume we have a model: Z(s) = µ + ε(s), where ε(s) is a zero mean second-order stationary random field with covariogram function C(h) and variogram γ(h). Also σ2 =C(0). The weighted linear estimator for location s0 is: (*) The estimation error at location s0 is the difference between the predictor and the random variable modeling the true value at that location: The bias is: So, as long as the weighted linear estimator (*) is unbiased. All the methods in chapter 10 meet this condition, thus are unbiased. However, the unbiasedness tells us nothing about how to determine the weights wi’s. ∑= = n i iiZwZ 1 0 ˆ ∑ −=−= 0000 ˆ ZZwZZR ii ( ) ∑ −=∑ −= )()()( 000 ZEZwEZZwERE iiii ( )∑ −=∑ −=∑ −= 1)()( 0 iiii wwZEZEw µµµ ,1∑ =iw
  • 3. 3 Minimizing error variance Kriging is such a method that determines the weights so that the mean squared error (MSE) is minimized: subject to the unbiasedness constraint Once we have chosen a data generating model (through a covariogram or variogram), the minimization of MSE can be achieved by setting the n partial first derivatives to 0, then the n weights wi’s can be obtained by solving the n simultaneous equations. However, this procedure does not quite work for our problem because we only want the solutions that meet the unbiasedness condition. ( )2 00 )ˆ( ZZEMSE −= .1∑ =iw ( ) ( )00 2 00 ˆvar)ˆ( ZZZZE −=− ),ˆcov(2)var()ˆvar( 0000 ZZZZ −+= ( ) ),cov(2)var(var 00 ZZwZZw iiii ∑−+∑= 0 2 2 ii i j ijji CwCww ∑−+∑∑= σ
  • 4. 4 Minimizing error variance using the Lagrange multiplier The Lagrange multiplier is a useful technique for converting a constrained minimization problem into an unconstrained one. Take the first term w1 as an example: The final ordinary kriging system is: ( )∑ −+∑−+∑∑= 122 0 2 iii i j ijji wCwCwwMSE λσ 0222 10 1 1 1 =+−∑= ∂ ∂ = λCCw w MSE n j jj 10 1 1 CCw n j jj =+∑ = λ             =             •             1 ...... 01...1 1... ............ 1... 0 101 1 111 nnnnn n C C w w CC CC λ C w = D (n+1)×(n+1) (n+1)×1 (n+1)×1 w = C -1 D (**)
  • 5. 5 Estimating the variance of errors Because kriging predictor is unbiased, the variance of the prediction errors is just the MSE: The first term on the right hand side – From the the equation of the first derivative (i.e., the (**) equation on the previous page), we have Therefore, the error variance is of the form: The variance is often called the ordinary kriging variance, expressed in a matrix form: Note: σ2 is simply C(0). 0 2 2 ii i j ijji CwCwwMSE ∑−+∑∑= σ .)( 00 ∑ −=∑ −=∑ ∑=∑∑ i ii i ii i j ijji i j ijji CwCwCwwCww λλ λσ −∑−= 0 2 iiCwMSE DwOK '22 −=σσ
  • 6. 6 Interpretation of kriging The kriging system may be better understood through the following intuitive interpretation. Two steps are involved in determining the linear weight of kriging: 1. The D vector provides a weighting scheme similar to that of the inverse distance method. The higher the covariance between a sample (denoting i = 1, 2, …, n) and the location being estimated (denoting 0), the more that sample would contribute to the estimation. Like an inverse distance method, the covariance (thereof weight) between sample i and location 0 generally decreases as the sample gets farther away. Therefore, D vector contains a type of inverse distance weighting in which the “distance” is not the geometric distance to the estimating sample but a statistical distance. 2. What really makes kriging differ from the inverse distance method is the C matrix. The multiplication of D by C-1 does more than simply rescale D so that w sums to 1. C records (covariance) distances between all sample pairs, providing the OK system with information on the clustering of the available sample data. So C helps readjust the sample weight according to their clustering. Clustered samples will be declustered by C. Therefore, OK system takes into account of two important aspects of estimation problem: distance and clustering.             =             •             1 ...... 01...1 1... ............ 1... 0 101 1 111 nnnnn n C C w w CC CC λ C w = D (n+1)×(n+1) (n+1)×1 (n+1)×1 w = C -1 D
  • 7. 7 Ordinary kriging in terms of variogram γ(h) In practice, kriging is usually implemented using variogram rather than covariogram because it has better statistical properties (unbiased and consistent). From chapter 9 (page 6): γ(h) = C(0) - C(h), we have C(h) = C(0) - γ(h). Substituting this covariogram into the unconstrained MSE on page 4 leads to Similar to the covariogram, the weights can be solved by setting the equations of the 1st derivatives w.r.t. wi’s to zero. The final kriging equation in matrix notation is: ( )∑ −+−∑−+∑∑ −= 12)(2)( 0 222 iii i j ijji wwwwMSE λγσσγσ ( ).122 0 ∑ −+∑+∑∑−= iii i j ijji wwww λγγ Γ w = D (n+1)×(n+1) (n+1)×1 (n+1)×1 w = Γ -1 D               =               − •                 1 ...... 01...11 10... ............... 1...0 1...0 0 20 10 2 1 21 221 112 nnnn n n w w w γ γ γ λ γγ γγ γγ
  • 8. 8 Ordinary kriging variance in terms of variogram γ(h) Following the same steps as for the variance based on the covariogram, we have the ordinary kriging variance in terms of variogram: where w and D are the vectors given on the previous page. Steps for kriging – 1. EDA exploration, removing trend, checking for stationarity and isotropy 2. Computing the empirical variogram 3. Fitting and selecting a theoretical variogram model to the empirical variogram 4. Computing the weight w using the fitted theoretical variogram, i.e., kriging. 5. Predicting the values at the locations of interest 6. Validation 7. Plotting kriging surfaces ,'0 2 Dww iiOK ∑ =−= λγσ
  • 9. 9 Checking and removing trends (make the data stationary) Example: soil pH value in the Gigante plot of Panama, using the full data set (soil.dat, has 349 data points). The data appear to have a trend in the northwest-southeastern direction. To remove such a trend, we fit the data using using model: z = 5.67 - 0.003295x + 0.001025y + 4.521e-6x2 + ε. Terms y2 and x×y are not significant. It seems that the trend surface analysis has detrended the data. 0 100 200 300 400 500 0200400600800 0 100 200 300 400 5000200400600800 Before detrended After detrended High Low
  • 10. 10 Has the trend really been removed? We further check it using variograms. The comparison of the variograms before and after detrending confirms that there is no trend in the residuals. We are confident that the residuals of the trend surface analysis are likely stationary. We can now go on to do kriging. >soil.geodat=as.geodata(soil.dat,coords.col=2:3,data.col=5,borders=T) >variog.b0=variog(soil.geodat,uvec=seq(0,500,by=5), max.dist=500) >plot(variog.b0) >variog.b2=variog(soil.geodat,uvec=seq(0,500,by=5),trend="2nd",max.dist=500) >plot(variog.b2)
  • 11. 11 Spherical model Logistic model Fitting a variogram Several variogram models can be fitted to the data. For illustration purpose, only two models (the spherical and logistic models) are shown here. By visual inspection, it seems that the logistic model may capture the spatial autocorrelation better than the spherical model, particularly at short distance lag. However, the sigmoid shape of the logistic model may not reflect the intrinsic feature of the data. We will use the spherical model for kriging. 2 2 0009258.0 0001072.0 1065.0)( h h h +=γ Logistic model:
  • 12. 12 R implementation using geoR- ordinary kriging 1. Compute variogram by directly considering trend (i.e., removing 2nd order trend. Kriging will automatically put back the trend in the final prediction): >variog.b2=variog(soil.geodat,uvec=seq(0,500,by=5),trend="2nd",max.dist=500) 2. Model variogram using spheric variogram model: >pH.sph=variofit(variog.b2,cov.model="spherical") >pH.sph # also try summary(pH.sph) to see the output variofit: model parameters estimated by WLS (weighted least squares): covariance model is: spherical parameter estimates: tausq sigmasq phi 0.1240 0.0955 130.7104 3. Fitting logistic model: > u=variog.b2$u;v=variog.b2$v > logist.nls=nls(v~c0+a*u^2/(1+b*u^2),start=c(c0=0.05,a=0.25,b=0.1)) >logist.nls model: y ~ c0 + a * x^2/(1 + b * x^2) data: parent.frame() c0 a b 0.1064792 0.0001072 0.0009258 residual sum-of-squares: 0.04605 nugget (c0) c1 range for 0 < h ≤ 130.71 for h ≥ 130.71       −+= 3 ) 71.130 ( 2 1 71.1302 3 0955.01240.0)( hh hγ 0955.01240.0)( +=hγ 2 2 0009258.0 0001072.0 1065.0)( h h h +=γ Logistic model:
  • 13. 13 R implementation - ordinary kriging 1. Generate locations at which interpolation is needed: >x=soil.dat$gx; y=soil.dat$gy >prd.loc=expand.grid(x=sort(unique(x)),y=sort(unique(y))) 2. Run krige.conv for spatial interpolation: >pH.prd=krige.conv(soil.geodat,loc=prd.loc,krige=krige.control(cov.model="spherical",cov.pars =c(0.09549404,130.71043698))) 3. View the prediction: You can directly apply image to pH.prd. Here we want to have more control over the features of the image, We create matrix for the pH.prd$predict and then apply image: >pH.prd.mat=matrix(pH.prd$predict,byrow=T,ncol=84) >image(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.mat), xlim=c(-20,500),ylim=c(-20,820),xlab="x",ylab="y") >lines(gigante.border,lwd=2,col=“green”) >contour(pH.prd,add=T) We can do the same thing to view the variation in the prediction: pH.prd$krige.var. Taking the squared root, it is prd.se. >pH.prd.se.mat=matrix(sqrt(pH.prd$krige.var), byrow=T,ncol=84) >image(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.se.mat), xlim=c(-20,500),ylim=c(-20,820),xlab="x",ylab="y") >lines(gigante.border,lwd=2)
  • 14. 14 Plot prediction variance It is desirable to view the variation of the prediction: pH.prd$krige.var. Taking the squared root for prd.se. >pH.prd.se.mat=matrix(sqrt(pH.prd$krige.var),byrow=T,ncol=84) >image(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.se.mat),xlim=c(-20,500), ylim=c(-20,820),xlab="x",ylab="y") >lines(gigante.border,lwd=2,col="blue") >contour(unique(prd.loc$x),unique(prd.loc$y),t(pH.prd.se.mat),xlim=c(-20,500), ylim=c(-20,820),add=T) pH surface pH std error surface pH std error with contour
  • 15. 15 1. Independent data validation: Compare the predicted with the observed data. As shown in the left table, these 13 data samples were not included in the kriging analysis. The predictions were generated from: >pH.prd13=krige.conv(soil.geodat,loc= prd.loc13,krige=krige.control(cov.model= "spherical",cov.pars=c(0.09549,130.71043))) 2. Cross-validation: Deleting one observation each time from the data set and then predicting the deleted observation using the remaining observations in the data set. This process is repeated for all observations. Residuals are then analyzed using standard techniques of regression analysis to check the underlying model assumptions. Evaluating the outputs of kriging prediction: SrfID gx gy Site pH pred var se.fit 31 240 8 site8mN 4.92 4.304 0.0088 0.0936 123 240 232 site8mS 5.97 5.933 0.0109 0.1046 124 240 220 site20mS 5.57 5.762 0.0203 0.1426 128 300 260 site20mN 5.41 5.375 0.0198 0.1406 190 358 420 site2mW 5.28 5.268 0.0033 0.0573 242 362 540 site2mE 5.23 5.437 0.0039 0.0627 243 368 540 site8mE 5.04 5.343 0.0104 0.1020 290 238 600 site2mW 4.63 5.086 0.0042 0.0647 291 232 600 site8mW 5.45 5.177 0.0145 0.1202 292 220 600 site20mW 5.4 5.371 0.0254 0.1595 310 422 660 site2mE 5.33 5.545 0.0033 0.0573 312 440 660 site20mE 6.17 5.747 0.0185 0.1361 360 478.6 538.6 site2mSW 5.55 5.310 0.0033 0.0572
  • 16. 16 Block kriging In many occasions, we are interested in estimating the value in a block (cell) rather than that at a point. The block kriging system is similar to that of the OK, of the form: where i.e., the covariogram between block A and sample point i is the average of the covariograms between the points locating within A and i. The block kriging variance is: where + + + + + + Block A . . . . + observed samples . regularly spaced locations set up within the block             =             •             1 ...... 01...1 1... ............ 1... 11 1 111 nA A nnnn n C C w w CC CC λ C w = D (n+1)×(n+1) (n+1)×1 (n+1)×1 , 1 ∑= ∈Aj ijiA C A C ,'2 DwCAAOK −=σ . 1 2 ∑ ∑= ∈ ∈Ai Aj ijAA C A C
  • 17. 17 R implementation - block kriging Block kriging is achieved by using OK: 1. Create a systematical grid lattice (as dense as you want) using expand.grid. 2. Use krige.conv for OK to do spatial interpolation for the grids. 3. Average the values of those grids falling within the defined block. + + + + + + Block A . . . . + observed samples . regularly spaced locations set up within the block
  • 18. 18 Spatial estimation: additive/nonadditive variables Some precaution is necessary before applying geostatistical analysis to your data. The method does not universally apply to any type of data. Additive variable: Nonadditive variable: Nonadditive variables include: number of species in a block, ratio data (e.g., number of cars per household in a city block). Geostatistics is invalid for analyzing nonadditive variables because subtraction makes no sense here. 5 balls 3 7 4 8 11 scaled up 3 colors (1b, 2r, 2 w) 5 colors 4 colors scaled up 3 colors (1b, 1g, 1y) 1 color (4g) 3 colors (3b, 2r, 2 w)
  • 19. 19   Spatial estimation: scale effect Few spatial data (point process is an exception) can avoid the problem of the size of sample area  (called support in geostat, or modifiable areal unit in geography, or grain size in landscape  ecology). In many practical applications, the support of the samples is not the same as the support of the  estimates we are trying to calculate. For example, when assessing gold ore grades in an area, we  take samples from drill hole cores, but in mining operation we treat truckloads as the size of  sample (consider a truckload either as ore or as waste). So a critical and difficult question is: can we infer about the properties of a variable at different  levels of supports from the observations sampled at a particular support? In other words, can we  scale down or up a spatial process? 0 1 2 3 4 010203040 rg am m a(100, 1) 0 2 4 6 8 0510152025 rg am m a(100, 3) ?
  • 20. 20   Grain size (m) No. stems/m2 (std. error) No. species/m2 (std. error) 5×5 0.671 (0.244) 0.585 (0.197) 10×10 0.671 (0.167) 0.475 (0.095) 20×20 0.671(0.130) 0.318 (0.038) 25×25 0.671 (0.121) 0.267 (0.026) 50×50 0.671  (0.100) 0.129 (0.008) 100×100 0.671 (0.085) 0.049 (0.001) 250×250 0.671 (0.048) 0.011 (0.0004) 500×500 0.671 (0.041) 0.003 (< 0.001) 500×1000 0.671 0.0016 Spatial estimation: scale effect Number of stems and number of species per  m2  at different sampling scales (grain size) in  a 1000×500 m rain forest of Malaysia. The  entire plot has 335,356  trees belonging to 814  species. The densities at each grain size were  computed as follows: (1) divide the plot into a  grid system using a given scale (e.g., 5×5 m),  (2) count the total number of stems and the  number of species in each cells, respectively,  (3) average these two quantities across all the  cells, and (4) then divide the averages by the  scale.  The results clearly show how sampling scale  profoundly affects the species diversity. They  suggest that diversity based on per unit area  (the last column) is a misleading measurement  for comparing diversity between two  ecosystems.
  • 21. 21   Spatial estimation: scale effect “This problem of the discrepancy between the support of our samples and the  intended support of our estimates is one of the most difficult we face in  estimation.” Isaaks & Srivastava (1989, page 193)