[R-sig-Geo] anisotropy gstat

[Edzer Pebesma]

Dave Depew wrote:
> Hi All,
> I suppose this is a rather simple question...however, I'm managing to 
> get more confused the more I read.
> I'm doing some OK and UK using the R-gstat package...I have some data 
> that is moderately anisotropic. Reading the gstat literature, it would 
> seem that to specify the appropriate parameters for anisotropic 
> variograms I need the set the angle equal to the direction of maximal  
> data continuity? or maximal range?. The ratio of the minimum to 
> maximum range appears to be straightforward enough, but the first part 
> is confusing me.
> As an example;
> data(meuse)
> e<-variogram(dist~1, loc=~x+y,data=meuse, alpha=c(0,45,90,135))
> plot(e)
Dave, try

require(gstat)
data(meuse)
coordinates(meuse) = ~x+y
e<-variogram(dist~1, meuse, alpha=c(0,45,90,135))
plot(e, vgm(.08, "Sph", 5000, anis=c(45, .2)))

I got there with a bit of trial and error. Clearly, for the 45 degrees 
most happens beyond 1500 m.

Of course this is just for the idea; a variogram model that has some 
parabolic behaviour at the origin would be prefered.
>
> Looks to me anyways, that the direction of maximum range is 45 deg, 
> and the minRange/maxRange is ~ 1000/1500 = 0.67.
the max correlation direction is 45, the range is, in my model, 5000, 
the anisotropy ratio 0.2.
>
> Could more experienced gstat users tell me if I'm out to lunch on this 
> one??
You weren't far off.
--
Edzer