You are here:
Spatially balanced survey designs for natural resources
Citation:
OLSEN, A. R., T. KINCAID, AND Q. C. PAYTON. Spatially balanced survey designs for natural resources. Chapter 6, Robert Gibson, Joshua Millspaugh, Andrew Cooper, Daniel Licht (ed.), Design and Analysis of Long-Term ecological Monitoring Studies. Cambridge University Press, Cambridge, Uk, , 126-150, (2012).
Impact/Purpose:
Ecological resource monitoring programs typically require the use of a probability survey design to select locations or entities to be physically sampled in the field.
Description:
Ecological resource monitoring programs typically require the use of a probability survey design to select locations or entities to be physically sampled in the field. The ecological resource of interest, the target population, occurs over a spatial domain and the sample selected is intended to be a “representative” sample of that target population. A sample that has the same spatial distribution as the target population would typically be considered a “representative” sample. Spatially balanced survey designs address this spatial “representativeness” feature. This chapter describes spatial balance and how it applies to ecological features that are polygonal areas (e.g., ecoregions, watersheds, land cover types, estuaries), linear networks (e.g., streams and rivers, road networks), and points (e.g. small lakes, 12 digit hydrologic units). The generalized random tessellation stratified (GRTS) approach to spatially balanced survey design is described and illustrated. Then the spsurvey library for the R Statistical Software is applied to points, linear networks and polygonal areas monitoring designs. The applications include use of GRTS when the survey design involves stratification and unequal probability of selection. A simulation study of spatial balance properties of GRTS compared to simple random sampling (IRS) is used to show that GRTS performs better than IRS.