strange variogram envelop

I computed a variogram (function variog, geoR) on the residuals of a

linear model (myvariog). I computed the envelop by permutation of the
data values on the spatial locations (function variog.mc.env;  nsim=999).

  >  myvariog$n
[1]   37 1577   97  116   53 2053  342  550   30  460  248


For the largest distance class, I have got 248 pairs of points and the
envelop does not enlarge but converges to zero (see the attached
figure). Could anybody give me any hint about the process explaining
this convergence?
I suppose, giving the aspect of the variogram, that I could not modelise
it in order to take into account spatial autocorrelation in my linear
model (thanks argument correlation). What do you think about it?

Thanks for your help,
Regards,

Marion.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: variogram lm.pdf
Type: application/pdf
Size: 1993 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20130515/716b125e/attachment.pdf>