Spatially Balanced Survey Designs for Large
Scale Monitoring Programs
Don L. Stevens, Jr.
Statistics Department
Oregon State University
Corvallis, Oregon
Spatial pattern is a dominant feature of environmental response
variables. The pattern is shaped by
geographic, metrological, and anthropogenic stressors that all have pattern
themselves. If the response and stressor patterns are well-understood, a
sampling design to monitor a response can utilize that information to locate
sample points at key sites. However,
in a period of large-scale climate change, relationships will not be static,
and a more robust design may be preferable. The Generalized Random Tessellation Stratified (GRTS) design
was developed in the USEPA's Environmental Monitoring and Assessment Program in
the 1990's. Spatially balanced designs, such as GRTS, ensure that spatial
distribution of the sample points closely matches the spatial distribution of
the domain. Thus, patterns in the
response will be sampled in proportion to their extent in the domain. The GRTS design has features, such as
dynamic sample adjustment and easy accommodation of variable probability, that
make it well-suited for large-scale monitoring programs. This talk will give a brief description
of GRTS, describe its functionality, and illustrate its application to
large-scale monitoring programs in the U.S. and Australia.