sample.geodata: Sampling from geodata objects
In rundel/geoR: Analysis of Geostatistical Data

Description Usage Arguments Details Value See Also Examples

This functions facilitates extracting samples from geodata objects.

1 2	sample.geodata(x, size, replace = FALSE, prob = NULL, coef.logCox, external)

`x`	an object of the class `geodata`.
`size`	non-negative integer giving the number of items to choose.
`replace`	Should sampling be with replacement?
`prob`	A vector of probability weights for obtaining the elements of the data points being sampled.
`coef.logCox`	optional. A scalar with the coeficient for the log-Cox process. See DETAILS below.
`external`	numeric values of a random field to be used in the log-Cox inhomogeneous poisson process.

If prob=NULL and the argument coef.logCox, is provided, sampling follows a log-Cox proccess, i.e. the probability of each point being sampled is proportional to:

exp(b * Y(x))

with b given by the value passed to the argument coef.logCox and Y(x) taking values passed to the argument external or, if this is missing, the element data of the geodata object. Therefore, the latter generates a preferential sampling.

a list which is an object of the class geodata.

as.geodata, sample.

par(mfrow=c(1,2))
S1 <- grf(2500,  grid="reg", cov.pars=c(1, .23))
image(S1, col=gray(seq(0.9,0.1,l=100)))
y1 <- sample.geodata(S1, 80)
points(y1$coords, pch=19)
## Now a preferential sampling
y2 <- sample.geodata(S1, 80, coef=1.3)
## which is equivalent topps
## y2 <- sample.geodata(S1, 80, prob=exp(1.3*S1$data))
points(y2$coords, pch=19, col=2)
## and now a clustered (but not preferential)
S2 <- grf(2500,  grid="reg", cov.pars=c(1, .23))
y3 <- sample.geodata(S1, 80, prob=exp(1.3*S2$data))
## which is equivalent to
## points(y3$coords, pch=19, col=4)
image(S2, col=gray(seq(0.9,0.1,l=100)))
points(y3$coords, pch=19, col=4)