Meng-ying,
I did mean the clustering of locations, if the sample is evenly and or randomly
spread your s^2 estimate will be no problem, it's when the data is clustered in
locations I believe removal of these data improves the estimate, the less clustering
the less improvment, the higher the clustering, the higher the improvment. Spatially
clustered data has correlation which is picked up in the sub-range portion of the
variogram.
So the variance is 93% of the sill for the population which adds credence to your
argument. What would be interesting is to see the results we get from a sample with
some clustering i.e. a sample variogram sill closer to 18.63. I'm interested in this
method for use with mine sampling data. I have GSLIB, but no compiler. Are you
developing a sample subset with some clustering, or can you send me the coordinated
SGSIM.OUT?
Digby
All right, if you think the clustering of data values (I'm not talking about clustering of locations) are not be part of the representation of population.
* By using the ai-geostats mailing list you agree to follow its rules ( see http://www.ai-geostats.org/help_ai-geostats.htm )
* To unsubscribe to ai-geostats, send the following in the subject or in the body (plain text format) of an email message to [EMAIL PROTECTED] Signoff ai-geostats
