4. Geostatistics

Question 1

What is the symbol used for autocovariance?

Answer 1

C(h)

Question 2

What is the symbol used for semivariance?

Answer 2

γ(h)

Question 3

What is the equation for autocovariance

Answer 3

C(h) = [ U(s), U(s’) ]

Question 4

What is the equation for semivariance?

Answer 4

γ(h) = 1/2 * Var [ U(s), U(s’) ]

Question 5

What is the range of spatial autocorrelation?

Answer 5

It is the distance after which there is no more autocorrelation.
(Autocovariance line hits 0)

Question 6

Where do you read the range on the plot?

Answer 6

On the x-axis

Question 7

What is the relationship between the range and the map?

Answer 7

The longer the range, the smoother the map

Question 8

When is the maximum value of autocovariance reached [ C(h) ]

Answer 8

At the zero distance lag

Question 9

What is the nugget effect?(3ish)

Answer 9

• The drop in the autocovariance graph or the jump in the semivariance graph.
• It is the NON autocorrelated part of the random component of the stochastic process

Question 10

What are the three types of pairs of distances that can be used?

Answer 10

• true distances
• equal-frequency distance classes
• equally spaced distance classes

Question 11

Which the the types of pairs of distances is the best method?

Answer 11

equal-frequency distances

Question 12

Why is it an issue if there is a lack of precision in the small distances?

Answer 12

It is a critical point because of the nugget effect

Question 13

Which method is preferably used to fit a variogram model?

Answer 13

EGLS fitting procedure

Question 14

Why is the EGLS fitting procedure prefered?

Answer 14

Because it takes into account the inequality of variances and the correlation between ordinates of the experimental variogram, especially at the first distance lags, and thus provides model parameter estimates with suitable properties.

Question 15

What is the drift?

Answer 15

the mean function

Question 16

In Phase I of CRAD (coregionalization analysis with a drift) what are the two main criteria applied in performing the decomposition?

Answer 16

1. Modeling spatial heterogeneity of the mean while incorporating spatial autocorrelation of the random component through EGLS fitting
2. Modeling spatial autocorrelation from the direct experimental variogram of residuals

Question 17

In PROC VARIOGRAM, what does maxlag= number do?

Answer 17

It allows you to set the maximum number of distance classes that you want.

Question 18

In PROC VARIOGRAM, what does lagd = number do?

Answer 18

It is the length of the distance classes

Question 19

What is the difference between an experimental variogram, an empirical semivariogram, and an estimated semivariance plot?

Answer 19

They are all the same thing