block and output grid size in block kriging

Dear Edzer, 

To understand why block kriging results vary,with different grid sizes, to
keep gridsize identical to grid cells, and smaller than grid cells.

library(gstat)
data(meuse)
coordinates(meuse) = ~x+y
data(meuse.grid)
gridded(meuse.grid) = ~x+y

lzn.vgm<- variogram(log(zinc)~1, meuse)
lzn.fit1 <- fit.variogram(lzn.vgm, model=vgm(psill=1, model="Sph",
range=900, nugget=1))
plot(lzn.vgm, lzn.fit1)

# Ordinary kriging
lzn.ok <- krige(log(zinc)~1, meuse, meuse.grid, model = lzn.fit1)

#Block kriging 
lzn.bok <- krige(log(zinc)~1, meuse, meuse.grid, model = lzn.fit1,
block=c(40,40))

# Choosing a random location with the centre coordiantes(xx,yy) and find the
corresponding block kriged estimate
xx = 181140
yy = 333100
est.bk = lzn.bok$var1.pred[(lzn.bok$x==xx)&(lzn.bok$y==yy)]
[1] 6.499158
est.ok =lzn.ok$var1.pred[(lzn.ok$x==xx)&(lzn.ok$y==yy)]
[1] 6.499624

Why are the kriging estimates at a point different in block and ordinary
kriging, when the grid size is isentical to (40 x 40) or less than(20x20)
the meuse data.

Thanks







diff= m.bk - est.bk



--
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/block-and-output-grid-size-in-block-kriging-tp7587770p7587780.html
Sent from the R-sig-geo mailing list archive at Nabble.com.