Re: GSTAT: Anisotropy

["Edzer J. Pebesma" <edzer@pangea.Stanford.EDU>]

Dear all,

tol_hor and alpha are both listed under the variogram modelling
options, and as such they only affect the modelling of the
variogram (or more specific: the selection of point pairs
for inclusion in estimates of the sample variogram, based on
direction).

After calculating the sample variograms in several directions
(using tol_hor and alpha combinations) you can fit, by trial
and error I guess, an anisotropic variogram, which would
typically involve choosing a nugget and sill, and an ellipse
for the range (being direction dependent). The ellipse is
characterised in 2D by three parameters: the major (maximum)
range; the direction of the maximum range and the ratio of
the minimum to the maximum range (0 < ratio <= 1).

Anisotropy in kriging is obtained by defining an anisotropic
variogram in the command file. How this is done should be
clear from the manual.

Hope this helps,
--
Edzer