Discrete Global Grids Applicable to Spatially Balanced Sampling

Anthony Olsen¹, Denis White¹ and Don L. Stevens. Jr.<sup>2

1</sup>US EPA Western Ecology Division
Corvallis, Oregon

²Dynamac Corporation
Corvallis, Orgeon

Email: tolsen@mail.cor.epa.gov,  denis@mail.cor.epa.gov,  dstevens@mail.cor.epa.gov


Developing efficient survey designs for natural resources, such as lakes, streams, forests, requires that the spatial distribution of the target population be explicitly considered. During the past 10 years, a joint effort among geographers and statisticians has resulted in an approach to survey design based on the concept of a discrete global grid. These new spatially-balanced survey designs have been successfully applied to over 100 environmental monitoring programs within the United States as well as in a number of other countries. First, we describe the theoretical requirements for spatially-restricted designs. Second, we illustrate how discrete global grids are applied to complete a hierarchical randomization that results in a spatially-balanced sample. The concepts are illustrated with applications to surveys of lakes in Wyoming, streams in Mid-Atlantic states, and terrestrial lands in western states. The survey designs are based on a new discrete global grid system. The system uses an icosahedral model of the sphere, a diamond hierarchical structure, and Snyder’s equal area projection.