[R-sig-Geo] variance estimation (gstat, geoR, automap)

[Sébastien Durand]

Have you found a way to compute the variance estimation for the surface? 

I would be happy to see how?

I'm trying to estimate the variance of a global abundance estimation 
computed by kriging interpolation and I am stuck.

I have tried to used and test the block options but I don't get the same answer.

x1=c(-1,1,9)
x2=c(-1,0,3)
x3=c(2,0,4)
x0=c(0,0)
d=rbind(x1,x2,x3)
colnames(d)=c("x","y","z")
c0=1
c=10
a=3
dist(rbind(x0,d[,1:2]))
d=data.frame(d)
coordinates(d)=~x+y
vario=variogram(z~1, d)
v=vgm(1,model="Lin",0,nugget=1)

x0=data.frame(x=0,y=0)
coordinates(x0)=~x+y
krige(formula=z~1, locations=d, newdat, model=v)
# The point estimation σ2 = var1.var, match the example value of 2.65


But for an area it is another issue and I cannont get the proper answer.



Any answers.