Sampling Designs for Environmental Monitoring - Introduction to Ecological Sampling - page 152

TABLE 10.2
Common Spatial Designs Suitable for Research and Monitoring Studies, Along with
References and a Brief Description or Comment

Spatial Design References Description/Comment

Research
Simple random

Cochran (1977, Chap. 2) Sites selected completely at random.

Covered by many sampling texts and
articles.

Scheaffer et al . (1979, Chap. 4)
Särndal et al . (1992, Chap. 3)
Lohr (2010, Chap. 2)

Two stage Cochran (1977, Chap. 11) Large or primary sites (polygons)

selected first. Small or secondary sites
within the selected primary sites are
sampled second. In general, the design
and sample size can vary by stage.

Scheaffer et al . (1979, Chap. 9)
Särndal et al . (1992, Chap. 4)
Lohr (2010, Chap. 6)

Stratified Cochran (1977, Chap. 5, 5A) A special case of two-stage sampling in

which strata define primary sites, and
all primary sites are sampled (census
at stage 1). A special case is the
one-per-strata sampling (Breidt, 1995).
In general, design and sample size can
vary by strata.

Scheaffer et al . (1979, Chap. 5)
Särndal et al . (1992, Chap. 3)
Lohr (2010, Chap. 3)

Cluster Cochran (1977, Chap. 9, 9A) A special case of two-stage sampling

that occurs when all secondary sites
within a primary site are sampled
(census at stage 2).

Scheaffer et al . (1979, Chap. 9)
Särndal et al . (1992, Chap. 4)
Lohr (2010, Chap. 5)

Monitoring
Systematic Cochran (1977, Chap. 8), Also, grid sampling. Suitable for two-

dimensional resources. The geometry of
the grid is usually square (square cells)
or triangular (hexagonal cells).

Scheaffer et al . (1979, Chap. 8)
Särndal et al . (1992, Chap. 3)
Lohr (2010, Section 2.7)

General random
sample (GRS)

This chapter Suitable for one-dimensional resources.

Produces equal probability or unequal
probability samples, ordered or
unordered samples, simple random
and systematic samples of fixed size.

Balanced
acceptance
sampling (BAS)

Robertson et al . (2013) Suitable for n -dimensional resources.

Produces equal probability or unequal
probability spatially balanced samples
in n dimensions.

Note: See corresponding section of the text and references for a more thorough description.

vertical y range of coordinates. Random site coordinates are then generated
one at a time as random coordinate pairs within the bounding box. Random
coordinate pairs are generated by choosing a random deviate from a uniform
distribution for the x coordinate, and an (independent) random deviate from