Which spatial data quality can be meta-propagated?

Abstract

As a mid-way between the usability criteria proposed by the upcoming ISO 19157 and error propagation methods, the concept of meta-propagation allows derivation of an estimate of the propagated uncertainties using only metadata information about the quality of the datasets and processes used in a scientific workflow. The principle of meta-propagation has been illustrated using the thematic accuracy with the quantitative attribute accuracy sub-element. This paper explores other data quality elements that can be meta-propagated. Using a similar approach as for thematic accuracy, we address further the metadata quality for the processes and their links with data quality. The paper focuses on generic principles and a few specific situations where appropriate quality measures can be fully described. Basic meta-propagation, based on separability of the assessments (i.e. one input to one output), is presented but the paper also discusses the potential of using quality measures for non-separable approaches. Spatiality of the measures, i.e. taking the benefit of a spatial data quality value being a map, is also investigated. Practical considerations between the burden of deriving needed quality metadata and the will to derive useful uncertainty assessments for large scientific workflows are also expressed.