[R-sig-Geo] practical range for kriging interpolation or not

[Edzer Pebesma]

On 05/14/2014 05:53 PM, Moshood Agba Bakare wrote:
> Dear All,
> The range obtained from fitting an exponential function to empirical
> variogram using gstat library of R is 5.5 m. This indicates that the
> practical range is  5.5 x 3 = 16.5 m indicating a pair of sample yield
> points whose distance is beyond 16.5 m is uncorrelated.
> 
> what will be the value of maxdist option of krige function to produce
> estimated values within the grid? I thought of using maxdist=16.5 as my
> search radius in interpolation so that a data point beyond 16.5 m will
> not be used in interpolation. Am I right to use the practical range
> (16.5) or the original range (5.5) obtained directly from fitting the
> exponential model?
> 
> Thank you.
> 

Using 16.5 makes more sense than 5.5, but I've never heard statisticians
confirm that this would be good practice.

The exponential variogram decreases asymptotically to zero correlation,
so correlation is not zero at distance 16.5.

Points far away, at very low spatial correlation distances, are still
taken into account for estimating the mean (ordinary kriging) or mean
function (universal/external drift/regression kriging).

It is pretty simple to try things out e.g. by cross validation.
-- 
Edzer Pebesma
Institute for Geoinformatics (ifgi), University of Münster
Heisenbergstraße 2, 48149 Münster, Germany. Phone: +49 251
83 33081 http://ifgi.uni-muenster.de GPG key ID 0xAC227795

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: OpenPGP digital signature
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20140514/a1fef2f9/attachment.bin>