[R-sig-Geo] multivariable variogram modeling

[Faye Anderson]

You might consider posting to r-sig-geo next time, so others may enjoy
the discussion and may even answer.

The most common situation for this is that you have duplicate
observations (use function zerodist to find out), or you have perfectly
correlated variables; use correct.diagonal argument in fit.lmc.

Pls let me know if either of the two worked.
--
Edzer
Hi everyone,
Below is a question I sent to Edzer and he thankfully helped me with.
Any comments?
thanks
Faye
_____
Hi Edzer,

I'd appreciate it if you can comment or provide me with any kind of
guidance on the following:
In your book (Applied Spatial Data Analysis with R), I'm trying to
apply the code of section 8.4.5 to a multivariate data set of similar
structure (groundwater constituents) but I get an error message of
"singular covariance matrix". I think the covariance matrix of my
variables is not positive definite. How can I confirm that? If so,
what is the route for me to get a model for my data set?

Thank you so much.

Faye