Defining the Universe: Creating a Data Model for a Geospatial Data Repository

摘要

For the last several years, many people have been working on solving the problem of interoperability among heterogeneous simulation systems. Some of this effort has been focused on the creation of data models for such federations by aggregating the models of each system. This is predicated on a standardized framework for data models and a common dictionary of terms. Initiatives such as SEDRIS and the Common Data Model Framework (CDMF) provide the foundation for this work. This work, however, has profound implications on a different problem: the creation of a data repository to facilitate the rapid generation of geospatial datasets. Eliminating the duplication of effort between and among the different user domains within the M&S and C4I communities requires a repository which can store, not only source data, but also the results of complex operations performed during the process of creating a dataset for a specific user. The definition of a consistent data model which can store information from many different users is similar to creating a data model for a federation of many different systems. Choosing a framework for creating such a data model is necessary in order to ensure that the repository can adapt to an evolving data model. Because the purpose is slightly different, the CDMF does not fit this application. A similar framework, the Fused Data Metamodel, is more conducive to the creation of a data model for a repository. The Fused Data Metamodel, which builds on the CDMF, adds three key components necessary for a repository data model: 1) it does not require that all attributes in the data model have values in a compliant dataset, 2) it incorporates both raster and vector data, and 3) it allows for the persistence of correlations between data elements. Utilizing the framework we introduced in [1], this paper describes the creation of a data model specifically designed for use in a repository serving the M&S and C4I communities. We examine the advantages of creating a unified data model for such a repository, and describe how the hurdles of creating this data model are alleviated by the tools already developed to analyze and create data models for the interoperability of heterogeneous systems. Finally, we show that the creation of a geospatial data repository, along with the definition of its data model, is a prerequisite to the rapid generation of synthetic natural environments.