Skip to content

taking zonal anisotropy into account in gstat

2 messages · István Gábor Hatvani, Edzer Pebesma

Original
István Gábor Hatvani
# 
Dear Colleagues,

I would like to ask for some help.

Does anyone have an idea, is it possible to take zonal anisotropy into
account during kriging in *gstat*, such as in the paper found in the link
below in my GoogleDrive? I did not find any argument concerning this
question.

https://drive.google.com/folderview?id=0B0GWyljhqBgKYnM4LWR1SjFlNU0&usp=sharing

I did manage to plot it in another program, but I am not satisfied with it.

Thank you for your help in advance!

Best regards,

Istvan
Edzer Pebesma
# 
You can "fake" it with a geometrically anisotropic model that has a very
large range in the direction where you want it to disappear, as in:

library(sp)
demo(meuse,ask=FALSE,echo=FALSE)
library(gstat)
v = variogram(log(zinc)~1, meuse, alpha = c(0,45,90,135))
vm = vgm(.25, "Sph", 1000, anis = c(45, 0.5))
plot(v, vm, main = "geometric")
zonal = vgm(.5, "Sph", 1e9, anis = c(45, 1/1e6))
# range is 1e9, effectively infinity, in 45 direction;
# it is 1e9/1e6 = 1000 in 135 direction.
vm = vgm(.25, "Sph", 1000, add.to = zonal)
plot(v, vm, main = "zonal")
On 05/03/2015 01:39 PM, Istv?n G?bor Hatvani wrote: