Usage Guidelines

The following are some guidelines for using the NGDA archive data model.

  1. The data model provides two grouping mechanisms.
    • Use archival objects to group components that cannot be used separately, or at least are not ordinarily thought of as being separately usable. Example: the files that make up a Shapefile.
    • Use collections (i.e., a collection object and collection-member relationships) to group objects that are related but can still be used independently. The collection object then provides a place to put collection-level metadata, other common metadata, indexes and overviews, etc. Examples: DOQs, DRGs, DEMs...
  2. Define objects and components using the most specific formats available.
    • Defining a file as XML will almost always be insufficient because XML is only a generic encoding mechanism. Instead, use a format that defines the meanings of the tags. Example: define an FGDC metadata record (encoded in XML) using the FGDC/XML format.
    • If a "text" file has a regular, program-interpretable structure, define the file using a format that defines that structure. Example: define a TIFF world file using the ESRI world file specification.
  3. Propagate definitions up the component tree as far as possible. For example, given a "subcomponents" directory containing the multiple files that make up a Shapefile, define the directory as a whole as a Shapefile, not just the individual files. Associating the Shapefile format with both the directory and the individual files is permissible, but redundant.
  4. Archival objects and components may have more than one definition.
    • Use multiple definitions per object or component if there are multiple applicable formats that are not related by the format registry. For example, given a file that contains Dublin Core metadata elements encoded in RDF/XML syntax, define the file using both the Dublin Core specification (which defines the meaning of the elements, independent of any encoding) and the RDF/XML format (which defines a generic encoding mechanism). [Actually, there is a specification that defines a standard encoding of Dublin Core in RDF, but for the purposes of this example we're pretending it doesn't exist or is not applicable.]
    • Use multiple definitions when a file has a nominal format and has subsequently been compressed, or when multiple files have been packaged and compressed into a single file. In such cases, associate the file with both the nominal format(s) and the compression format. Example: for a ZIP archive containing TIFF and TFW files, define the archive using the ZIP, TIFF, and TFW formats.
    • Do not use multiple definitions when the definitions are already related by the format registry. For example, defining a GeoTIFF file using both the GeoTIFF and TIFF formats is unnecessary and discouraged because the two formats are already related (namely, GeoTIFF is a subformat of TIFF).
    • Order of definitions is not significant.
  5. Definitions that apply to all objects in a collection, and that are unlikely to be reused by objects in other collections, should be placed in the collection object as opposed to the format registry. Example: for a collection of DOQs, define the DOQ components (TIFF image, TIFF world file, etc.) using formats, but place the DOQ specification and other, DOQ-specific metadata in the collection object.

last modified 2009-01-12 14:40