Echelons and PHASES as Candidates for Ranking in Spatial Ranked Set Sampling
by
Wayne L. Myers
The Pennsylvania State University

Ranked set sampling improves efficiency of estimation by ensuring better representation across the distributional range than conventional occurrence-based randomization. Geospatial investigations offer opportunities for using ranked set sampling in a double sampling sense if synoptic information is available for purposes of ranking. Databases for geographic information systems (GIS) typically constitute the more readily available sources of spatial variables by which to accomplish ranking. Once approach considered here facilitates cluster-based combination of several spatial variables for joint use in the ranking operation. Multi-band image data and other pseudo-band information can thereby be mobilized for the purpose of geospatial ranked set sampling. A second analytical approach called 'echelons' is applicable for surface or pseudo-surface variables. Echelons serve to give rankings relative to regional context and spatial complexity. By using echelons, one can segregate rankings for spatial elements that are part of upper-level formations from those that have similar values but occur in the context of lower-level formations.

Date received: October 22, 2002