VARIOGRAMS

EXAMPLE 1: CONDUCTIVITY DATA

First read the data in (if you haven't done already)

cru <- read.geodata("Cruciani.dat", head=T, coords.col=2:3, data.col=4)
cru.borda <- read.table("Cruciani.border", head=T)[,2:3]
cru.borda <- rbind(cru.borda, cru.borda[1,])
      

1. Variogram cloud

cru.cloud <- variog(cru, option="cloud")
plot(cru.cloud)
      

Inspect the alternative options for this function and try some of them.

args(variog)
options(helphtml = TRUE)
help.start()
help(variog)

cru0.cloud <- variog(cru, option="cloud", lambda=0)
plot(cru0.cloud)

cru.cloud.m <- variog(cru, option="cloud", est="modulus")
plot(cru.cloud.m)

cru.cloud0.m <- variog(cru, option="cloud", lam=0, est="modulus")
plot(cru.cloud0.m)
      

2. Binned variogram

par.ori <- par(no.readonly=TRUE)
par(mfrow=c(3,2), mar=c(3,3,0,0), mgp=c(1.5,.7,0))

cru.v1 <- variog(cru)
plot(cru.v1, ylim=c(0,3.5))
cru.v1

cru0.v1 <- variog(cru, lambda=0, max.dist=12)
plot(cru0.v1, ylim=c(0,3.5))
cru0.v1

cru0.v2 <- variog(cru, uvec=seq(0,9, l=8), lambda=0)
plot(cru0.v2, ylim=c(0,3.5))

cru0.v3 <- variog(cru, uvec=seq(1,15, l=8), lambda=0)
plot(cru0.v3, ylim=c(0,3.5))

cru0.v4 <- variog(cru, uvec=seq(0,9, l=12), lambda=0)
plot(cru0.v4, ylim=c(0,3.5))

cru0.v5 <- variog(cru, uvec=seq(1,15, l=12), lambda=0)
plot(cru0.v5, ylim=c(0,3.5))

par(par.ori)
      

3. Fitting variograms

FITTING "BY EYE"

cru.v1 <- variog(cru, uvec=seq(0,9, l=8), lambda=0)
plot(cru.v1, ylim=c(0,2.5))
lines.variomodel(list(nugget=0, cov.pars=c(2, 2), cov.model="exp"), max.dist=9)
      

Now try different models changing the arguments above
Some examples:

lines.variomodel(list(nugget=0.5, cov.pars=c(2, 2), cov.model="exp"), max.dist=9, col="red")

lines.variomodel(list(nugget=0.5, cov.pars=c(2, 3), cov.model="exp"), max.dist=9, col="blue")

lines.variomodel(list(nugget=0.5, cov.pars=c(2, 2), cov.model="gau"), max.dist=9, col="green")
      

NOTE: details on correlation model is given in the documentation for cov.spatial.

cru.v1 <- variog(cru, uvec=seq(0,9, l=8), lambda=0)
plot(cru.v1)

cru.v1.exp <- variofit(cru.v1, ini=c(2,2), cov.model="exp")
lines(cru.v1.exp)

cru.v1.gau <- variofit(cru.v1, ini=c(2,2), cov.model="gau")
lines(cru.v1.gau, lty=2)
      

Explore the options in variofit :

args(variofit)
help(variofit)

cru.v1.exp1 <- variofit(cru.v1, ini=c(2,2), cov.model="exp", wei="eq")
lines(cru.v1.exp, col="blue")

cru.v1.gau1 <- variofit(cru.v1, ini=c(2,2), cov.model="gau", wei="eq")
lines(cru.v1.gau, lty=2, col="blue")
      

EXAMPLE 2: a simulated data

Load the data included in the package distribution

data(s100)
help(s100)
      

Try to analyse this data performing exploratory and variogram analysis.
In particular try with different options for the arguments direction and tolerance when using the function variog.
Try also the variogram envelopes:

cru.v1 <- variog(cru, uvec=seq(0,9, l=8), lambda=0)
cru.env1 <- variog.mc.env(cru, obj=cru.v1)
plot(cru.v1, env=cru.env1)
      

EXAMPLE 3: Wolfcamp data

data(wolfcamp)
help(wolfcamp)
	

Try to analyse this data performing exploratory and variogram analysis.
In particular, try with the argument trend.