Metadata should be created and submitted to an NGDS catalog for any resource that is meant to be accessible individually.
Individual documents require one metadata record per document. Some document types may consist of a bundle of files, e.g. ESRI shape file. In general these should be bundled into a single file like a zip archive or UNIX tar file. The metadata must include the URL at which the document can be accessed. These documents might be scans of well logs, scanned reports or publications, or data in a spreadsheet, such as an Excel file.
Datasets (data products) are typically considered as individual works (see FRBR), unified by the compilation activity that brings information together into a single data structure, editing and verifying content as necessary. This approach is based on our interest in facilitating data access by users, and the observation that such a user is first interested in a particular kind of data; upon finding a fit-for-purpose dataset, they will next want to know how to get the data. From this perspective a dataset work (data product, resource) will have a single metadata record that may include specification of multiple distributions of the resource. For instance a borehole temperature dataset may be available with all the records in an Excel spreadsheet, or visualized through a web map service, or individual observations may be accessed through a web feature service (see for example Montana Hot Springs).
At a more granular level, individual records (features, objects) in a dataset may include source information, documenting details of observation or measurement procedure and other information specific to a particular data type. This might include information such as location, data and time of observations, and the source of the data. These feature-level metadata are delivered with the data, and only summarized in the work-level dataset metadata that are published to the NGDS-compliant catalog. This granularity issue can be difficult because of differing perspectives on what is data or metadata, differing granularity of documentation available, and different use-case priorities.
The required metadata content is explained in USGIN metadata recommendations. These requirements proscribe the content of the metadata, but not the delivery format. ISO19139 xml is the preferred encoding based on its expanding adoption in the community. USGIN guidelines for implementing the content recommendations in ISO19139 XML are available, as well as a detailed USGIN profile for interoperable metadata using ISO19139.
FGDC xml is widely used and if participants already have workflow in place using this format and can provide the requested metadata content, this can be harvested using an FGDC to ISO transformation. FGDC XML should be tested to validate against the official XML schema at http://www.fgdc.gov/schemas/metadata/fgdc-std-001-1998.xsd. Please confer with the NGDS developer team about metadata formatting to facilitate import of metadata into the NGDS catalog.