Combining Existing Monitoring Sites with a Probability Survey Design--Examples from U.S. EPA's National Coastal Assessment

Abstract

U.S. Environmental Protection Agency's (EPA's) National Coastal Assessment was envisioned as a research effort led by EPA's Office of Research and Development to evaluate assessment methods for ecosystem condition monitoring. The program was conducted through strategic partnerships with the coastal states. These states conducted the survey in their waters with a common set of indicators. The resources targeted for initial monitoring were estuarine waters. A flexible probability survey design was used to incorporate, to the extent possible, existing state monitoring program sites. Three criteria were developed to evaluate existing monitoring program sites in the northeastern United States for possible incorporation into the national design: (1) the sites were selected to be representative, (2) the variables sampled at the sites were similar in distribution with variables from a probability design, and (3) the correlation structure of variables was equivalent to that for a probability design. Detailed examples were presented for Long Island Sound water quality sites, New Jersey coastal water quality sites, and Casco Bay, ME, sediment sites to illustrate the approach.

Combining potentially incompatible community datasets when harmonizing forest inventories in subarctic Alaska, USA

“…Paul et al. () suggested combining data that exhibit similar univariate probability distributions and correlation structures, though this doesn't directly address the multivariate nature of multi‐species datasets. Combining multivariate datasets is also a common problem in comprehensive long‐term monitoring because monitoring programs often measure multi‐attribute (plant/animal/nutrient) outcomes in the face of ever‐evolving research questions and sampling methods that change over time (Lindenmayer & Likens, ).…”