[R-sig-Geo] problems in anisotropy detection

Sebastiano Trevisani @trev|@@n| @end|ng |rom |u@v@|t
Sat Jul 13 09:43:26 CEST 2019


Dear Emanuele,

Did you first checked manually the presence of anisotropy with gstat
variogram function? Then consider also that automatic fitting procedure
generally are not capable to fit a zonal anisotropy but only a geometric
one.

Sebastiano


Il sab 13 lug 2019, 07:32 Emanuele Barca <emanuele.barca using ba.irsa.cnr.it> ha
scritto:

> Dear friends,
>
> I have a dataset of hydraulic heads, mydata = <X, Y, HH, Z> and i would
> like to check for anisotropy.
>
> In R I found 3 functions to carry out such task:
>
> 1. likfit
>
> x_geodata <- as.geodata(mydata, coord.cols = 1:2, data.col = 3,
>                          covar.col = 4)
> fit_mle  <- likfit(x_geodata,
>                     fix.nugget = FALSE,
>                     cov.model = "exponential", psiA = 0, psiR = 1,
>                     ini = c(var(Y), 1), fix.psiA = FALSE, fix.psiR =
> FALSE,
>                     hessian = TRUE)
>
> that detects no anisotropy.
>
> 2. estimateAnisotropy
>
> mydata.sp <- mydata
> coordinates(mydata.sp) = ~ X + Y
> estimateAnisotropy(mydata.sp, depVar = "LivStat", "LivStat ~ Z")
>
> that returns the following
>
> [generalized least squares trend estimation]
> $`ratio`
> [1] 1.340775
>
> $direction
> [1] -35.29765
>
> $Q
>                Q11          Q22          Q12
> [1,] 1.926136e-05 2.329241e-05 5.721893e-06
>
> $doRotation
> [1] TRUE
>
> finally,
>
> 3. estiStAni
>
> vmod1 <- fit.variogram(vgm1, vgm(18, "Ste", 1300, 0.78, kappa = 1.7))
> estiStAni(vgm1, c(10, 150), "vgm", vmod1)
>
> that returns an error:
>
> Error in `$<-.data.frame`(`*tmp*`, "dir.hor", value = 0) :
>    replacement has 1 row, data has 0.
>
> I am very puzzled, can anyone help me understanding if there is
> anisotropy in my dataset?
>
> thanks
>
> emanuele
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

	[[alternative HTML version deleted]]



More information about the R-sig-Geo mailing list