In this paper we review approaches which have been taken to characterizing the uncertainty in spatial information. This includes both continuous predictions of spatial variables, and thematic maps of landcover classes. We contend that much work in this area has failed to engage adequately with the decision processes of the end-user of information, and that the engagement of spatial statisticians is essential to achieve this. We examine generalized measures of uncertainty, and those focussed on particular decision models. We conclude that the latter are likely to be the most fruitful, particularly if they emerge from a formal decision analysis. We outline the principles of value of information theory, and suggest that this represents an ideal framework in which to develop measures of uncertainty which can support both the rational collection of data and the interpretation of the resulting information.

在本文中，我们回顾了用于表征空间信息不确定性的方法。这包括空间变量的连续预测和土地覆盖类别的专题图。我们认为，该领域的许多工作未能充分参与信息最终用户的决策过程，而空间统计学家的参与对于实现这一目标至关重要。我们研究了不确定性的广义度量，以及那些专注于特定决策模型的度量。我们得出结论，后者可能是最富有成效的，特别是如果它们来自正式的决策分析。我们概述了信息论的价值原则，