Standards Problem Report
                          ========================

        Title: PDS Lacks Time Standard
         Date: 23 April 2003
 Submitted by: Anne Raugh, SBN


Problem:

  START_TIME is a fundamental search parameter in nearly all PDS data
  products, yet there is no formal definition of what time should be
  recorded in START_TIME.  In fact, time is not well defined in most,
  if not all, elements with a type of TIME.


Issues:

  Internally, it is not possible to make positive statements about the
  results of time-based searches even within a single mission.  The 
  uncertainty in START_TIME definition also casts doubt on any other
  calculations and parameters that might depend on START_TIME, including
  pointing information and target location.
  
  Externally, it effectively allows each data preparer to define his own
  local time standard for each data set, or - worse - to not define any
  time standard.  The long-term usefulness and reliability of the 
  resulting archive is significantly diminished by lack of a well-defined
  time standard.


Suggested Solution?

  Examine all elements currently in the PSDD with a GENERAL_DATA_TYPE of
  TIME.  Determine the appropriate time standard for each and insert this
  explicitly into each definition.

  Note that legacy data is unlikely to be reprocessed to correct START_TIME,
  even when that is possible.  A method must be provided to alert users
  to poorly-defined START_TIME values in legacy data.


Prioritization Considerations:

  Lack of a time standard clearly has not impeded the archiving activities
  of the PDS to date.  However, it is - or at least should be - embarrassing
  that an organization that exists to preserve scientific data does not have
  a well-defined time standard for fundmental measurements like START_TIME.


Impact/Schedule Considerations:

  Requiring well-defined time standards will improve the reliability and
  long-term usefulness of the archived data.

  Requiring consistent time standards across all data sets will improve the
  reliability of time-based searching algorithms.

  Time definitions for legacy data may not be recoverable, and legacy data
  are unlikely to be corrected for this problem alone.  This may corrupt
  the results of time-based searches. 

  It may be difficult to ensure that users of legacy data are informed that
  current time standards do not apply to data archived prior to a particular
  date.

  Discussion of what constitutes an appropriate time standard for any given
  value, particularly one that is applicable across all data types, may need
  to be undertaken at a very high level by the science managers.
  
  All data preparers must be informed of the explicit time standard requirements
  early in the design process, and the time values provided in product labels
  must be carefully checked for standards adherence.