On Tue 2003/01/28 16:31:03 PDT, Rob Seaman wrote
in a message to: LEAPSECS_at_ROM.USNO.NAVY.MIL
>A useful exercise from other mailing lists and conferences and such
>is simply to get to know each other and how we all fit into the puzzle
>under discussion.
I am a lapsed astrophysicist, currently a programmer with the Australia
Telescope National Facility (ATNF,
http://www.atnf.csiro.au), a radio
observatory.
These days I am mainly engaged in developing software for the Parkes
radiotelescope (
http://www.parkes.atnf.csiro.au) in two spheres:
1) All aspects of general data reduction for the Multibeam system
(
http://www.atnf.csiro.au/research/multibeam/) including the
21cm HIPASS, ZOA, and HVC surveys of the southern hemisphere.
2) The GUI used for the Telescope Control System (TCS),
I also have a finger in several other pies, mostly related to
astronomical data reduction. In particular, I am a co-author of the
two latest FITS standards documents which deal with astronomical
coordinates (
http://www.atnf.csiro.au/~mcalabre/WCS).
In past years I wrote the high-precision ephemeris routines, ATELIB,
(
ftp://ftp.atnf.csiro.au/pub/software/atelib/) for the Australia
Telescope Compact Array (ATCA,
http://www.narrabri.atnf.csiro.au).
As an adjunct to this I wrote a software system to collect and parse
IERS-A email bulletins and update a database primarily containing
information on leap-seconds, the predicted difference between UTC and
UT1, and the celestial pole offset. This has been running since 1989
and I continue to maintain it.
The IERS-A digest is also used to update the AT Distributed Clock (ATDC,
http://www.atnf.csiro.au/technology/electronics/docs/atdc/atdc.html)
used at all ATNF observatories and is also sent to Jodrell Bank for
their use.
My professional interest in UTC:
1) The ATELIB routines require Greenwich apparent sidereal time
(GAST) to high precision. The observatory maintains its
approximation of TAI using an atomic clock and uses the leap-
second and UTC-UT1 tables to compute UT1, and thence GMST and
GAST.
2) The TCS GUI provides a separate graphical display of the sky
showing a source catalogue and current telescope position.
Observers may use this directly to drive the telescope, or,
typically, just monitor the telescope position and decide what
best to observe next. This display uses unix system time and
relies on UTC approximating UT1. A couple of seconds accuracy
is enough, but an error of several minutes would be unacceptable.
This is especially so for sources near the zenith; the display
is based on an equi-rectangular projection scaled by the azimuth
and elevation drive rates so sources near the zenith appear to
move swiftly across it.
My perspective on UTC:
1) As far as ATELIB is concerned, TAI and UT1 are the times of
interest and UTC is just a bloody nuisance. ATELIB has to deal
with it only because IERS-A tabulates UTC-UT1 rather than
TAI-UT1. If the latter was tabulated, then UTC would only be
calculated for the sake of our NTP servers and the ATDC display
(
http://www.atnf.csiro.au/technology/electronics/ -
docs/atdc_display/atdc_display.html).
And yes, I have had to confront the tedium of engineering
software so that it handles leap-second updates that occur
during a 12 hour observation. No, I didn't get it right the
first time.
2) UTC (civilian time) ought to track the diurnal motion of the
Earth just as the Gregorian calendar tracks it's orbital
motion - and for the same reason.
A practical alternative to the Gregorian calendar - Julian
Date, is not used for civilian timekeeping because it is not
tied to the annual cycle of the seasons. Even the Julian
calendar was not considered accurate enough over the millennia,
so, yes, we do have to do some tedious bookkeeping and
calculation sometimes. Likewise, a system of diurnal time as
a numerical count of SI seconds with no relation to the day, the
most important of climatic cycles, would be equally unwelcome.
In this context, I should point out an important difference
between leap-seconds and leap-days (i.e. leap-years). It would
not be practical to construct a calendar containing a fractional
number of days because then the new-year and all later days in
the year would start at a strange time as measured by the
position of the Sun. However, there is no similar impediment to
a day containing a fractional number of SI seconds because an SI
second is not tied directly to any phenonenon of practical daily
interest.
I note that, as yet, we have not heard a reasoned argument against my
proposal for UTC to measure the true length of day in SI seconds.
Mark Calabretta
ATNF
Received on Wed Jan 29 2003 - 17:08:53 PST