Review Discrete Global Grid Systems with OGC Develop semantic data models supporting feature fusion Phase 1 Standardize metadata for provenance and uncertainty Phase 1

50 Copyright © 2010 Open Geospatial Consortium data instances are also needed – a topic closely related to ADSD and fusion of unstructured data.

9.4.8 Continue to improve methods for GML schema handling

OGC has extensive experience in working with information schemas, e.g., GML schemas. Even so, OGC can continue to improve and refine methods for schema handling, harmonization and run-time mapping with adaptors. Further efforts on improving handling of associations are needed.

9.4.9 Review Discrete Global Grid Systems with OGC

Tessellation systems have been discussed previously in OGC in particular in the OGC Abstract Specification, Topic 6 Coverages and Topic 2 Coordinate Reference Systems. During OGC TC meetings, there have been recent discussions about the need for coordinate reference systems without singularities at the poles and date line. The DGGS system presented by Pyxis has desirable properties and may meet the OGC needs. It is recommended that discussion of DGGS be taken up with the OGC membership to determine if a movement toward consensus can be achieved. A testbed might be established which demonstrates the advantages of DGSS through a WCS.

9.4.10 Develop semantic data models supporting feature fusion Phase 1

Common data models and encoding patterns are needed for representing feature semantics, feature associations, and their geometry, topology, and temporal properties in standard ways for enabling interoperable ObjectFeature fusion. Elements of the data model include common ontologies, vocabularies and taxonomies, association types, link encoding mechanismspatterns, and the means for publishingsharingprocessing of “fused” features. Clearly GML and OM are the starting-point for an ObjectFeature fusion model. Such models and patterns are essential for interoperable transformation and automated processing of data in fusion workflows. With stable and finite representations of features i.e., their structure, associations, and semantics derived from a common data model, come the means for discovery, transformation, and reasoning in support of compose-able and higher-order ObjectFeature fusion capabilities needed to solve increasingly more complex problems and to share the results.

9.4.11 Standardize metadata for provenance and uncertainty Phase 1

Fusion is a hard problem in part because we are drowning in a volume of data from multiple sources, all with different levels of detail and uncertainty. The challenge in ObjectFeature fusion is not getting data but making sense of them While it is important to minimize the introduction of uncertainty during processing and handling of data, it is equally important to recognize and quantitatively characterize the uncertainty in a result. Quantitative representations of uncertainty support provenance history of a data product within the feature lifecycle model. The use of metadata, and specifically uncertainty metadata, must be shown in real-worldpractical fusion scenarios. UncertML should be considered with evaluation of the Gaussian model for the variety of types of uncertainty. Demonstration of methods and tools for creating and interoperably using provenance and uncertainty metadata to support automation and improved faster, valid, more accurate fusion results are needed. Copyright © 2010 Open Geospatial Consortium 51 10 Observation sensor Fusion

10.1 Introduction