Data Break 6: Meuse River Zinc, Empirical

Semivariogram

BIOS 737 Spring 2004

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.1/11
 Meuse River

• library(gstat)
• Data included in gstat library.
• data(meuse) makes the data available.
• Soil samples from flood plain of Meuse River, near
village of Stein (Belgium).
• Geostatistical data: locations s and attributes Z(s).

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.2/11
 Visualizing data

• Create a “bubble plot” (dots at locations, size

represents attribute).
• bubble command in gstat library.

library(gstat)
data(meuse)

bubble(meuse, xcol="x",ycol="y",
zcol="cadmium", maxsize = 2.5,
main = "cadmium concentrations (ppm)",
key.entries = c(.5,1,2,4,8,16))

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.3/11
 Bubble plot: Cadmium
cadmium concentrations (ppm)

333000

332000
0.5
1
2
y

4
8
16

331000

330000

178500 179000 179500 180000 180500 181000 181500

x

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.4/11
 Bubble plot: Zinc
zinc concentrations (ppm)

333000

332000
100
200
400
y

800
1600

331000

330000

178500 179000 179500 180000 180500 181000 181500

x

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.5/11
 Our goal

• Use (empirical) semivariogram to describe patter

in zinc concentrations.
• Question 1: Do the zinc concentrations appear to
be normally distributed?
• Log(zinc)?

par(mfrow=c(2,2)
hist(meuse$zinc)
qqnorm(meuse$zinc)
hist(log(meuse$zinc))
qqnorm(log(meuse$zinc))

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.6/11
 Normally distributed?
Histogram of meuse$zinc Normal Q−Q Plot

40

1500
Sample Quantiles
30
Frequency

1000
20

500
10
0

0 500 1000 1500 2000 −2 −1 0 1 2

meuse$zinc Theoretical Quantiles

Histogram of log(meuse$zinc) Normal Q−Q Plot

7.5
40

7.0
Sample Quantiles

6.5
30
Frequency

6.0
20

5.5
10

5.0
0

4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 −2 −1 0 1 2

log(meuse$zinc) Theoretical Quantiles

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.7/11
 Using gstat

• Function variogram calculates the empirical

variogram.
• Lots of options, we’ll illustrate the most basic.
• plot recognizes output of variogram and plots
with nice labels.

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.8/11
 R code

logzinc.var <- variogram(log(zinc)˜1,

loc=˜x+y,
data=meuse)

print(logzinc.var)

plot(logzinc.var)

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.9/11
 Output in logzinc.var

np dist gamma dir.hor dir.ver i

1 57 79.29244 0.1234479 0 0
2 299 163.97367 0.2162185 0 0
3 419 267.36483 0.3027859 0 0
4 457 372.73542 0.4121448 0 0
5 547 478.47670 0.4634128 0 0
6 533 585.34058 0.5646933 0 0
7 574 693.14526 0.5689683 0 0
8 564 796.18365 0.6186769 0 0
9 589 903.14650 0.6471479 0 0
10 543 1011.29177 0.6915705 0 0
11 500 1117.86235 0.7033984 0 0
12 477 1221.32810 0.6038770 0 0
13 452 1329.16407 0.6517158 0 0
14 457 1437.25620 0.5665318 0 0
Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.10/11
 Resulting plot

0.6

0.4
semivariance

0.2

0.0
0 500 1000 1500
distance

Data Break 6: Meuse River Zinc, Empirical SemivariogramBIOS 737 Spring 2004 – p.11/11