14 Lab: Variogram properties

Influence of variogram parameters on universal kriging predictions

Load the geoR package which provides the wolfcamp data. Create by the following code a data.frame that contains a grid of prediction points.

library(geoR)
d.wolfcamp.grid <- expand.grid(
  x=seq(from=-250, to=250, length=52), 
  y=seq(from=-150, to=150, length=32)
)

Then use the following code to explore the effect of the variogram parameters on the UK predictions:

persp(
  krige.conv(geodata=wolfcamp, locations=d.wolfcamp.grid,
    krige=krige.control(trend.d="1st", trend.l="1st", 
    cov.model="spherical", cov.pars=c(sill=1000, range=100),
    nugget=3000)), 
  theta=150, phi=30, d=5, scale=FALSE,
  expand=0.3, shade=0.5, xlab="x", ylab="y", zlab="pressure"
)

Task 1

Vary the nugget-to-sill ratio for a total sill (= nugget + sill) equal to 4000.

Task 2

Using a constant nugget-to-sill ratio of 1:3 and a range equal to 100, compare the predictions for a total sill of 400 and 4000.

Task 3

Vary the range for a nugget-to-sill ratio equal to 1:3.