Knowledge reference for national forest assessments - sample designs
Construction of an appropriate sample design for an NFA is a crucial step if estimates are to be sufficiently precise and scientifically defensible. One of the first steps in this process is to define the target population and select a sampling frame. We recommend an infinite population sampling frame in which observations and measurements of a field plot support area are attributed to the point at the field plot center. Because NFAs often are expected to produce estimates of change, we recommended that the sample design include at least some permanent plots. The next step is to distribute the field plots throughout the population to be sampled. We have presented information on and discussed several popular sample designs: simple random sampling, stratified sampling, systematic sampling, and cluster sampling. If the sample design includes a systematic component, we recommend caution when using rectangular grids for target populations with large north-south components. Although the selection of the particular sample design depends on a variety of considerations, we recommend that if stratified sampling is not used, consideration be given to post-sampling stratification and stratified estimation. Finally, additional information on these and more complex sample design issues is available in the reference material.