WFC: Year-2000 issue (again)

[Don Wells]

"SA" == Steve Allen <sla at ucolick.org> writes:
SA> The FITS standard could be amended to mandate that UTC be used for all
SA> DATExxx keywords for earth-based observations subsequent to the
SA> adoption date.  

That is what I am recommending.

SA> I would support that.  

Good. 

My bet is that we will be able to conduct interoperability trials of this
simple agreement quickly, that a world-wide consensus will support it,
and that we will have no trouble getting the IAU-FWG to approve it promptly.
I hope that this can occur before the Kyoto General Assembly, so that we
can have a resolution in support of our agreement passed by the GA at Kyoto.

SA> But if so it must also contain
SA> explicit language indicating that the UTC mandate should not be
SA> presumed for pre-1972 data.

I agree. This is a subject which Barry should discuss in the User's
Guide.  It is clear that there are a number of knowledgeable people in
the FITS community who can help him to generate the explanatory text.

SA> This would permit future earth-based FITS data to be unambiguous, and
SA> it could not invalidate existing archival data.  It would retain the
SA> current level of time-system ambiguity for existing data, and for data
SA> which cannot satisfy the mandate.

I agree.

			       -=-=-=-

I certainly agree with Arnold etal that we should pay careful attention
to what timescale(s) we use in an era in which we are conducting
multi-wavelength observations of objects in high gravitational fields
with dimensions of light-seconds, an era in which our detectors resolve
the times of such observations to milliseconds, an era in which some of
those detectors are on rapidly-moving platforms remote from the Earth. It
is entirely appropriate that Arnold and others who have been studying
these time problems should help Barry to get advice-to-FITS-implementors
into the User's Guide. In the end, we will probably need to agree on some
new FITS conventions to properly represent the physics of such
observations.

I suggest that we consider a class of keywords (column labels, axis
types) of the form TIMExxxx, e.g. TIME-OBS, which would be used for
continuous timescales (no leap seconds!)  measured in "seconds", and for
which a TIMESYS keyword would apply.  An agreement on time along these
lines would be a valuable addition to the celestial coordinate WCS
agreement on which we have been working for the past 15 years.

Meanwhile, we need to get the year-2000 DATExxxx issue behind us ASAP.

(http://www.year2000.com/cgi-bin/y2k/year2000.cgi tells how much time we
have left to fix DATExxxx; it just now told me: "Only 2 years, 230 days,
8 hours, 17 minutes, 1 seconds until Jan. 1, 2000 (UTC)".)

-Don
-- 
  Donald C. Wells         Associate Scientist         dwells at nrao.edu
                    http://fits.cv.nrao.edu/~dwells
  National Radio Astronomy Observatory                +1-804-296-0277
  520 Edgemont Road,   Charlottesville, Virginia       22903-2475 USA