Using wavelet tools to estimate and assess trends in atmospheric data NRCSE.
NRCSE
description
Transcript of NRCSE
Nonstationary covariance structures I Deformations
Peter Guttorp amp Paul D Sampson
University of Washington
NRCSE
The spherical correlation
Corresponding variogram
ρ(v) =1minus 15v + 05 v
φ( )3 h lt φ
0 otherwise
⎧⎨⎪
⎩⎪
( )φ φ
στ + minus le leφ
τ + σ gt φ
22 3
2 2
3 ( ) 02
t t t
t
nugget
sill
range
Review Descriptive characteristics of (stationary) spatial covariance expressed in a variogram
Nonstationary spatial covariance
Basic idea the parameters of a local variogram model---nugget range sill and anisotropy---vary spatially
Look at some pictures of applications from recent methodology publications
Swall amp Higdon Process convolution approachSoil contamination example --- Piazza Rd site
Swall amp Higdon Process convolution approachPosterior mean and covariance kernel ellipses
Pintore amp Holmes 2005 Spatially adaptive non-stationary covariance functions via spatially adaptive spectra
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
The spherical correlation
Corresponding variogram
ρ(v) =1minus 15v + 05 v
φ( )3 h lt φ
0 otherwise
⎧⎨⎪
⎩⎪
( )φ φ
στ + minus le leφ
τ + σ gt φ
22 3
2 2
3 ( ) 02
t t t
t
nugget
sill
range
Review Descriptive characteristics of (stationary) spatial covariance expressed in a variogram
Nonstationary spatial covariance
Basic idea the parameters of a local variogram model---nugget range sill and anisotropy---vary spatially
Look at some pictures of applications from recent methodology publications
Swall amp Higdon Process convolution approachSoil contamination example --- Piazza Rd site
Swall amp Higdon Process convolution approachPosterior mean and covariance kernel ellipses
Pintore amp Holmes 2005 Spatially adaptive non-stationary covariance functions via spatially adaptive spectra
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Nonstationary spatial covariance
Basic idea the parameters of a local variogram model---nugget range sill and anisotropy---vary spatially
Look at some pictures of applications from recent methodology publications
Swall amp Higdon Process convolution approachSoil contamination example --- Piazza Rd site
Swall amp Higdon Process convolution approachPosterior mean and covariance kernel ellipses
Pintore amp Holmes 2005 Spatially adaptive non-stationary covariance functions via spatially adaptive spectra
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Swall amp Higdon Process convolution approachSoil contamination example --- Piazza Rd site
Swall amp Higdon Process convolution approachPosterior mean and covariance kernel ellipses
Pintore amp Holmes 2005 Spatially adaptive non-stationary covariance functions via spatially adaptive spectra
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Swall amp Higdon Process convolution approachPosterior mean and covariance kernel ellipses
Pintore amp Holmes 2005 Spatially adaptive non-stationary covariance functions via spatially adaptive spectra
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Pintore amp Holmes 2005 Spatially adaptive non-stationary covariance functions via spatially adaptive spectra
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Nott amp Dunsmuire 2002 Biometrika Fig 2 Sydney wind pattern data Contours of equal estimated correlation with two different fixed sites shown by open squares (a) location 33middot85degS 151middot22degE and (b) location 33middot74degS 149middot88degE The sites marked by dots show locations of the 45 monitored sites
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Kim Mallock amp Holmes JASA 2005 Piecewise Gaussian model for groundwater permeability data
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Nonstationary covariance models 1 Deformations
bull P Guttorp and P D Sampson (1994) Methods for estimating heterogeneous spatial covariance functions with environmental applications In G P Patil C R Rao (editors) Handbook of Statistics XII Environmental Statistics 663-690 New York North HollandElsevier bull W Meiring P Guttorp and P D Sampson (1998) Space-time Estimation of Grid-cell Hourly Ozone Levels for Assessment of a Deterministic Model Environmental and Ecological Statistics 5 197-222 bull PD Sampson (2001) Spatial Covariance In Encyclopedia of Environmetricsbull PD Sampson D Damian and P Guttorp (2001) Advances in Modeling and Inference for Environmental Processes with Nonstationary Spatial Covariance In GeoENV 2000 Geostatistics for Environmental Applications P Monestiez D Allard R Froidevaux eds Dordrecht Kluwer pp 17-32bull PD Sampson D Damian P Guttorp and DM Holland (2001) Deformationmdashbased nonstationary spatial covariance modelling and network design In Spatio-Temporal Modelling of Environmental Processes Coleccioacute laquoTreballs DrsquoInformagravetica I Tecnologiaraquo Nuacutem 10 J Mateu and F Montes eds Castellon Spain Universitat Jaume I pp 125-132bull D Damian PD Sampson and P Guttorp (2003) Variance Modeling for Nonstationary Spatial Processes with Temporal Replications Journal of Geophysical Research ndash Atmosphere 108 (D24) bull F Bruno P Guttorp PD Sampson amp D Cocchi (2004) Non-separability of space-time covariance models in environmental studies In The ISI International Conference on Environmental Statistics and Health conference proceedings (Santiago de Compostela July 16-18 2003) a cura di Jorge Mateu David Holland Wenceslao Gonzaacutelez-Manteiga Universidade de Santiago de Compostela Santiago de Compostela 2003 pp 153-161bull John Kent Statistical Methodology for Deformations
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
General space-time setup
Z(xt) = (xt) + (x)12E(xt) + (xt) trend + smooth + error
We shall assume that is known or constantt = 1T indexes temporal replicationsE is L2-continuous mean 0 variance 1 independent of the error C(xy) = Cor(E(xt)E(yt))D(xy) = Var(E(xt)-E(yt)) (dispersion)
euro
Cov(Z(xt)Z(yt)) =ν(x)ν(y)C(xy) x ne y
ν(x) + σε2 x = y
⎧ ⎨ ⎩
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Geometric anisotropy
bull Recall that if we have an isotropic covariance (circular isocorrelation curves) bull If for a linear transformation A we have geometric anisotropy (elliptical isocorrelation curves) bull General nonstationary correlation structures are typically locally geometrically anisotropic
euro
C(xy) = C( x minus y )
euro
C(xy) = C( Ax minus Ay )
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
The deformation ideaIn the geometric anisotropic case write
where f(x) = Ax This suggests using a general nonlinear transformation
G-plane rarr D-space
Usually d=2 or 3We do not want f to fold
Remark Originally introduced as a multidimensional scaling problem find Euclidean representation with intersite distances monotone in spatial dispersion D(xy)
euro
C(xy) = C( f (x) minus f(y) )
2 df R Rrarr
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Implementation
Consider observations at sites x1 xn Let be the empirical covariance between sites xi and xj Minimize
where J(f) is a penalty for non-smooth transformations such as the bending energy
euro
ˆ C ij
euro
(θf ) a wijˆ C ij minus C(f(xi )f (x j)θ)( )
ijsum
2+ λJ(f )
J (f)=part2f
partx2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+2part2f
partxparty
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2
+part2f
party2
⎛
⎝ ⎜ ⎜
⎞
⎠ ⎟ ⎟
2⎡
⎣
⎢ ⎢
⎤
⎦
⎥ ⎥ dxdyintint
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
SARMAP
An ozone monitoring exercise in California summer of 1990 collected data on some 130 sites
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Transformation
This is for hr 16 in the afternoon
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Fig 7 Precipitation in Southern France - an example of a non-linear deformation
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
G-plane Equicorrelation Contours
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
D-plane Equicorrelation Contours
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Theoretical properties of the deformation model
IdentifiabilityPerrin and Meiring (1999) Let
If (1) and are differentiable in Rn(2) γ(u) is differentiable for ugt0then (fγ) is unique up to a scaling for γand a homothetic transformation for f(rotation scaling reflection)
( )( ) ( ) ( ) ( ) n nD x y f x f y x y R Rγ= minus isin times1f minusf
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Thin-plate splines
euro
f(s) = c + As + W ˜ σ (s)
Linear part
euro
˜ σ (s) = σ(s minus x1)σ(s minus xn)( )
euro
σ(h) = h 2 log( h )
euro
1 W = 0 X W = 0
euro
J(f) = tr(W ˜ S W)
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
A Bayesian implementation
Likelihood
Prior
Linear part ndashfix two points in the G-D mapping ndashput a (proper) prior on the remaining two parameters
Posterior computed using Metropolis-Hastings
euro
L(S | Σ) = (2πΣ )minus(Tminus1) 2 exp minusT2
trΣminus1S ⎧ ⎨ ⎩
⎫ ⎬ ⎭
euro
p(W) prop exp minus1
2τWi
˜ S Wii=1
2
sum ⎛
⎝ ⎜
⎞
⎠ ⎟
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
California ozone
12
3
4
5
6
789
10
1112
1314
15
16
1718
19
20
21
22
23
24
25
26
27
2829
30
31
32
33
34
35
36
37
3839
40
4142
43
444546
47
48
49
50
51
52
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
13
14
15
16
17
18
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
63 Region 6 monitoring sites and their representation in a deformed coordinate system reflecting spatial covariance
Thu Oct 30 001236 PST 2003
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Posterior samples
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
44
4546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63
12
3
4
5
67 8
9
10
1112
1314
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
5960 61
62
63 12
3
4
5
67 89
10
1112
13
14
15
16
1718
19
20
21
22
2324
25
26
27
2829
30
31
32
3334
35
36
37
38
39
404142
43
444546
47
48
49
50
5152
53
54
55
56
57
58
596061
62
63
N=63 S Calif 4 samples from the posterior distribution of deformations reflecting spatial covarianceTue Oct 28 221829 PST 2003
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Other applications
Point process deformation (Jensen amp Nielsen Bernoulli 2000)
Deformation of brain images (Worseley et al 1999)
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Isotropic covariances on the sphere
Isotropic covariances on a sphere are of the form
where p and q are directions pq the angle between them and Pi the Legendre polynomials
Example ai=(2i+1)ρi
euro
C(pq) = aii= 0
infin
sum Pi (cosγpq )
euro
C(pq) =1minus ρ2
1minus 2ρcos γpq + ρ2 minus 1
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
A class of global transformations
Iteration between simple parametric deformation of latitude (with parameters changing with longitude) and similar deformations of longitude (changing smoothly with latitude) (Das 2000)
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Three iterations
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Global temperature
Global Historical Climatology Network 7280 stations with at least 10 years of data Subset with 839 stations with data 1950-1991 selected
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Isotropic correlations
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Deformation
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Assessing uncertainty
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Other approaches Haas 1990 Moving window kriging Nott amp Dunsmuir 2002 Biometrikamdashcomputationally convenient but hellip Higdon amp Swall 1998 2000 Gaussian moving averages or ldquoprocess convolutionrdquo model Fuentes 2002 Kernel averaging of orthogonal locally stationary processes Kim Mallock amp Holmes 2005 Piecewise Gaussian modeling Pintore amp Holmes 2005 Fourier and Karhunen-Loeve expansions
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Gaussian moving averages
Higdon (1998) Swall (2000)Let be a Brownian motion without drift and
This is a Gaussian process with correlogram
Account for nonstationarity by letting the kernel b vary with location
euro
X(s) = b(s minus u)dξ (u)R2int
euro
ρ(d) = b(u)R2int b(u minus d)du
euro
ρ(s1s2 ) = bs 1R2int (u)bs 2(u)du
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds
Kernel averagingFuentes (2000) Introduce orthogonal local stationary processes Zk(s) k=1K defined on disjoint subregions Sk and construct
where wk(s) is a weight function related to dist(sSk) Then
A continuous version has
euro
Z(s) = wk (s)Zk (s)k=1
K
sum
euro
ρ(s1s2 ) = wk(s1)wk(s2 )ρkk=1
K
sum (s1 minus s2 )
euro
Z(s) = w(x minus s)Zθ (s )int (x)ds