This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: University of California, Los Angeles Department of Statistics Statistics C173/C273 Instructor: Nicolas Christou Ordinary kriging Kriging (Matheron 1963) owes its name to D. G. Krige a South African mining engineer and it was first applied in mining data. Kriging assumes a random field expressed through a variogram or covariance function. It is a very popular method to solve the spatial prediction problem. Let Z = ( Z ( s 1 ) ,Z ( s 2 ) ,...,Z ( s n )) be the vector of the observed data at known spatial locations s 1 ,s 2 ,...,s n . The objective is to estimate the unobserved value Z ( s ) at location s . The model: The model assumption is: Z ( s ) = μ + δ ( s ) where δ ( s ) is a zero mean stochastic term with variogram 2 γ ( · ). The variogram was discussed in previous handouts in detail. The Kriging System The predictor assumption is ˆ Z ( s ) = n X i =1 w i Z ( s i ) i.e. it is a weighted average of the sample values, and ∑ n i =1 w i = 1 to ensure unbiasedness. The w i ’s are the weights that will be estimated. Kriging minimizes the mean squared error of prediction min σ 2 e = E [ Z ( s ) ˆ Z ( s )] 2 or min σ 2 e = E " Z ( s ) n X i =1 w i Z ( s i ) # 2 For intrinsically stationary process the last equation can be written as: σ 2 e = 2 n X i =1 w i γ ( s s i ) n X i =1 n X j =1 w i w j γ ( s i s j ) (1) See next page for the proof: 1 Let’s examine ( Z ( s ) ∑ n i =1 w i Z ( s i )) 2 : z ( s ) n X i =1 w i z ( s i ) ! 2 = z 2 ( s ) 2 z ( s ) n X i =1 w i z ( s i ) + n X i =1 n X j =1 w i w j z ( s i ) z ( s j ) = n X i =1 w i z 2 ( s ) 2 n X i =1 w i z ( s ) z ( s i ) + n X i =1 n X j =1 w i w j z ( s i ) z ( s j ) 1 2 n X i =1 w i z 2 ( s i ) 1 2 n X j =1 w j z 2 ( s j ) + n X i =1 w i z 2 ( s i ) = 1 2 n X i =1 n X j =1 w i w j [ z ( s i ) z ( s j )] 2 + n X i =1 w i [ z ( s ) z ( s i )] 2 If we take expectations on the last expression we have 1 2 n X i =1 n X j =1 w i w j E [ z ( s i ) z ( s j )] 2 + n X i =1 w i E [ z ( s ) z ( s i )] 2 = 1 2 n X i =1 n X j =1 w i w j var [ z ( s i ) z ( s j )] + n X i =1 w i var [ z ( s ) z ( s i )] But var [ z ( s i ) z ( s j )] = 2 γ ( · ) is the definition of the variogram, and therefore the previous expression is written as: 2 ∑ n i =1 w i γ ( s s i ) ∑ n i =1 ∑ n j =1 w i w j γ ( s i s j ) Therefore kriging minimizes σ 2 e = E [( Z ( s ) n X i =1 w i Z ( s i )] 2 = 2 n X i =1 w i γ ( s s i ) n X i =1 n X j =1 w i w j γ ( s i s j ) subject to n X i =1 w i = 1 The minimization is carried out over ( w 1 ,w 2 ,...,w n ), subject to the constraint ∑ n i =1 w i = 1....
View
Full
Document
This note was uploaded on 02/11/2012 for the course STATS c173/c273 taught by Professor Nicolaschristou during the Spring '11 term at UCLA.
 Spring '11
 NicolasChristou
 Statistics

Click to edit the document details