Filling Out the File Area Observational Classes

From The SBN Wiki
Revision as of 02:28, 25 January 2013 by Raugh (talk | contribs) (→‎Variable-Width Table Structures: Added note about collection inventories using delimited tables)
Jump to navigation Jump to search

The <File_Area_Observational> class is a specific flavor of the more general <File_Area> parent class. It contains the essential file description class, <File>, followed by data structure descriptions. Each observational product must have at least one <File_Area_Observational>, and may have more than one if the core observational data are contained in multiple files.

There should be one <File_Area_Observational> for each data file. Each of these File_Area_Observational classes may contain multiple data structures, but you should never repeat the <File_Area_Observational> for each data structure in a single file. In other words, there should never be two <File> sub-classes with the same <file_name> value.

For additional explanation, see the PDS4 Standards Reference or contact your PDS node consultant.

Note that in the PDS4 master schema, all classes have capitalized names; attributes never do.


This class is handled identically in all the File_Area_* classes. It is described in detail on the Filling Out the File Class page.

Data Structures

Following the <File> class are the data structure definitions. There is one of these for each data object in the file. You can list them in any order, though users generally prefer to see them in the same order in which they occur in the file.

As of this writing, these are the data structures allowed in an observational product. You should use the most specific data structure name that pertains to your data, to be able to take advantage of any transformation or analysis tools for that data type that would be applicable to your data.

Contact your PDS consultant if you don't see a suitable structure for your data to get advice on how to modify your data to fit archival structures. The data structures themselves are described on separate pages.

2D Array Structures

  • Array_2D_Image: Use this for your standard 2D image data.
  • Array_2D_Spectrum: Use this if your "image" is a 2D spectrum.
  • Array_2D_Map: Use this if your "image" is a map - especially if there is associated cartographic information.
  • Array_2D: In general, you should use a more specific flavor of Array_2D when possible, to get access to additional support software where it exists and would be useful.

3D Array Structures

  • Array_3D_Image: Use this for banded images that aren't spectral cubes (RGB color images, for example), or image stacks for which there is a cohesive third dimension. If your image stack is really a 2D image with a series of masks or backplanes, the masks/backplanes should generally be described individually (or as their own sequence).
  • Array_3D_Movie: Use this structure for a time sequence of 2D images. Note that this is still an array structure, not a contemporary movie format like MPEG.
  • Array_3D_Spectrum: Use this structure for spectral cubes to get access to transformation and analysis tools associated with this data type.
  • Array_3D: Use this generic 3D array structure if your data is sufficiently different from the other 3D array types that you don't need or want access to associated software for those types.

Fixed-Width Table Structures

  • Table_Character: Use this for fixed-width character tables. Note that for the time being, only ASCII characters are allowed in Table_Character data, so the fact that "width" is measured in bytes is equivalent to measuring it in characters.
  • Table_Binary: Use this for fixed-width binary data tables. Note that SBN experience indicates that end users hate binary tables, and they are very difficult to work with for users trying to compare data from numerous sources. SBN routinely advises its data preparers to avoid binary tables for these reasons, and on occasions reviewers have demanded that binary tables be translated to character tables before archiving.

Variable-Width Table Structures

  • Table_Delimited: This structure is used when parsing rules must be followed in order to convert data in the input file into programmatic memory structures. This is not a format favored by SBN users, so it is not generally accepted by our reviewers for archival data, but other nodes do use it and from time to time an SBN data preparer submits supporting data in a comma-separated variable text file format. In order to use this structure, the parsing rules must be documented, reviewed and accepted by PDS so that the data format can be supported by the standard PDS transformation tools. Contact your PDS node consultant for details.

A Table_Delimited structure is used with the PDS/DSV parsing standard (still in development) to define the inventory tables which list the products comprising a Collection.

Header Structures

  • Header: Use this for labelling FITS and VICAR headers in your data. If you have some other kind of header in your data, talk to your PDS node consultant about which data structure is appropriate.
  • Header_Encoded: No SBN observational product should ever use this structure. If you disagree, let us know.

Other Structures

  • Stream_Text: No SBN observational product should ever use this structure. If you disagree, let us know.