Hi Mark,
Thanks for the reproducible example. The problem is that when I look at
the sample variogram, the semivariance values start high and end low.
This is best illustrated by:
plot(variogram(theta_percent~1, sparse))
You see that there are outliers in the data that cause high
I just uploaded a new version of automap (1.07) that fixes this problem.
It deletes fitted variogram models with a negative sill/range/nugget.
cheers,
Paul
Paul Hiemstra wrote:
Hi Mark,
Thanks for the reproducible example. The problem is that when I look
at the sample variogram, the
I took the expedient of detecting the negative range and recording the
offending time and space slice. Then I went back and checked each model
type individually.
This is perfectly acceptable but I did want to alert you to the
possibility that autofitVariogram and therefore autoKrige could
Thanks. I'll pick it up as it becomes available.
Does the possibility exist that you would delete all models? In such a
case, what does the function return?
On 05/04/2010 05:49 AM, Paul Hiemstra wrote:
I just uploaded a new version of automap (1.07) that fixes this
problem. It deletes
I am using autofitVariogram during the process of interpolating a large
set of daily observations through a volume. Each volume is decomposed
into 2D layers prior to selecting a model to use for interpolation. I
made it through 2010 interpolations and then ran into a failed
interpolation