Uncertainty and Interoperability: The Areal Interpolation Problem

摘要

Discussions of geographic interoperability normally address two sets of issues: syntactic interoperability, having to do with compatibility of data formats, and semantic interoperability, having to do with the meaning ascribed to data by the sender and the receiver. In this paper we propose two additional dimensions, both related to data quality: issues associated with data accuracy, and affecting the ability to work with data from sources of different accuracies; and spatial support, having to do with differences of spatial sampling. The case of areal interpolation is explored, and it is shown that all known methods of areal interpolation can be brought within a single, unified framework based on geostatistics, and defined by the assumptions the user is willing to make about a hypothesized, underlying continuous field. Examples are used to illustrate the approach. The nominal case is discussed, and an empirical method presented that also appears to provide some promise in overcoming differences of spatial support.