|
Abstract:
|
Traditionally, large area natural resource inventories have relied on probability-based sampling and design-based inferences. However, the advent of moderate resolution satellite imagery and ground-based spatial data as sources of ancillary data, geographic information systems for combining data from these multiple sources, and fast computer processing systems have made model-based approaches to inference more attractive. Unfortunately, differences in the conceptual underpinnings of design-based and model-based approaches are often not well-understood within these natural resource inventory communities. The emphases of the presentation are threefold: (1) a brief articulation of some conceptual differences between design-based and model-based approaches, (2) examples that illustrate these differences, and (3) comments on particular advantages and disadvantages of both approaches.
|
