[R-sig-Geo] spautolm or errorsarlm (and gstat question)

[Zev Ross]

Hi All,

Just wondering if A) there is a prediction function
 associated with spautolm or B) if there is a way
to get the errorsarlm to accept the "weights" argument
similar to lm. As far as I can tell spautolm accepts 
non-spatial-weights but doesn't predict and 
errorsarlm predicts but doesn't accept "weights" 
is this true? And a somewhat unrelated question 
-- can someone tell me how, in R, to specify a nugget
that is half microscale variation and half measurement
error. I see in the GSTAT manual that it's possible, 
but I can't figure out the coding in R.

So if I have the following:

vgmRaw<-vgm(35, "Sph",range=1000,nugget=10)
vFitRaw<-fit.variogram(variogram(no2.ppb~1, 
data=x.oehha, width=250), model=vgmRaw,
  fit.method=2)

How do I make the nugget half and half?

Thank you! Zev