Complete geospatial data infrastructures (SDIs) consist of contemporary, comparable and integrated GI at global, regional or national levels along with services that enable an efficient use of the information. There are numerous research issues associated with the design, implementation and use of SDIs. Spatial data infrastructures policy including the political and administrative procedures required to initiate and maintain SDIs can be studied in order to enhance their utility. In practical terms there are problems such as copyright and pricing policies. Harmonization of databases can be based on appropriately applied ontology schemas and developed similarity measures. The fact that detailed geographic data are collected at different levels (municipal, regional, national) means that SDIs are likely to contain multiple representations in order to obtain the vertical integration. Effective generalization of maps as well as organization of multiple representations in databases could rationalize the production of topographic maps and assist in updating of databases. Such generalization requires significant consideration of conceptual schema, geometrical and spatial properties and visual appearance. It can be undertaken in real time (on-the-fly generalization) and it has links with Geovisualization and with the modelling described in the next section.
Metadata is the key for geospatial data infrastructures at both national and global levels, and the derivation, storage, scope and use of metadata have been addressed through mature national and ISO standards on metadata of GI (ISO 19115:2003) as well as its extension to gridded and imagery data (ISO 19115-2:2007). A special part of metadata describes quality information. There is also an ISO standard on geographic data quality (ISO 19113:2002) with definitions of quality elements and measures to be used. However, the uncertainty issues are not solved only by publishing standards and by forcing the data producers to document metadata of the produced datasets. The users need to be able to evaluate also the uncertainty of the results of the analyses in which they combine several datasets of different quality. Thus evaluation of the uncertainty of the GI analysis results and estimating the risks of subsequent decision-making are further research issues of importance.
Metadata is inherently multivariate and metadata representation by multivariate visualization methods, along with the usability of such visualizations needs to be examined. The linkages among metadata, data quality and visualization are potentially valuable. The metadata standard for gridded and imagery data, for example, introduces the ‘two-dimensional quality coverage concept’ and the ‘spatially varying quality concept’. These could be used for other data set types as well.
The visualization of data quality in general, and such spatially varying quality in particular, are examples of how map quality – including generalization quality – can be addressed.