Astronomical timekeeping strategies to
accommodate possible changes to UTC

Steve Allen (UCO/Lick Observatory)
Rob Seaman (National Optical Astronomy Observatory)
Peter Bunclark (University of Cambridge/Institute of Astronomy)
Mark Calabretta (ATNF/CSIRO)
Jonathan McDowell (Smithsonian Astrophysical Observatory)
Clive Page (University of Leicester)
Arnold Rots (CfA/Harvard)
Christian Veillet (CFHT)

UTC may be redefined without leap seconds.  This would significantly
affect many astronomical software systems.

No time scale has been immune to change

MNRAS vol 2, # 1, pp 11-12 (1831)

    The most prominent subject of public interest, (and one that has
    engaged much of the attention of your Council during the past
    year,) was the proposing an amended form of the _Nautical
    Almanac_.  The Lords Commissioners of the Admiralty, being
    desirous that the National Ephemeris should be remodelled, so as
    to meet the increasing wants and intelligence of the navy, and
    also the demands of astronomers, referred the consideration of the
    subject to the Council of the Astronomical Society.  A Committee
    was formed, with the co-operation of the most distinguished
    navigators and astronomers of the empire, and an elaborate Report
    drawn up, recommending various alterations and additions.  This
    has been approved of by the proper authorities; and the _Nautical
    Almanac_ of 1834 will appear in strict conformity with the plan
    advised.

    The most important _alteration_ is the substitution of _mean_ for
    apparent solar time in all the data of the Ephemeris.  It was with
    considerable reluctance, and after viewing the subject in every
    light, that the Committee resolved upon this material departure
    from the mode adopted by Maskelyne, and rendered familiar by the
    practice of more than half a century; but, as chronometers are to
    be found in perhaps every ship which relies upon astronomical
    means for her guidance, and as mean time must necessarily be
    obtained where chronometers are used, it was deemed safer to
    dismiss apparent solar time altogether as unnecessary, and as
    being a source of confusion.  The advantages, as to simplicity, of
    exchanging a varying for a constant measure of duration, and of
    assimilating the computations of the navigator and of the
    astronomer, are too obvious to be dwelt upon; and, in the opinion
    of all the naval officers present, no serious or lasting
    inconvenience from the change is likely to be felt.

The Admiralty of 1831 understood the different time scales.  They
recognized a distinction between a rarely-if-ever-reset private ship's
chronometer, measuring time intervals with its log of offsets and
rates, and a public clock, continually reset by astronomy to record an
epoch.  They deemed that most navigators already had the requisite
tools for a change in time scale.

Despite objection by the IAU, the Admiralty redefined GMT by 12 hours
in 1925.

The BIH and BIPM have made adjustments to TAI since its inception,
including major changes as of 1977 and 1995.

The IAU has redefined UT (UT1) as of 1956, 1984, 1997, and 2003; TT
(TDT) in 1991 and 2000; and TDB in 1991 and 2006.

Using the language of the 1831 Admiralty, the monthly issue of BIPM
Circular T demonstrates that the best chronometers in the world are
only clocks after they compare how different they are and reset
themselves, and TT(BIPMnn) retrospectively resets TT(TAI).

In conjunction with other international agencies, the ITU-R have been
seriously reconsidering the existence of leap seconds in UTC for six
years.  Several proposals have suggested abandoning leap seconds,
either immediately or after an interval of a few years.  Despite
considerable effort in surveys, polls, conferences, correspondence,
and public review processes the fate of leap seconds in UTC is still
not clear.

The AAS DDA recently reviewed implications and consequences
http://www.aas.org/policy/DDA-UTCreport.pdf
This resulted in the AAS Leap Second Committee
http://www.aas.org/policy/LeapSecondCommittee.html
Other jurisdictions still have open input channels to the ITU-R process.

The 1884 International Meridian Conference was called to address a new
need in technological civilization:
We must all be able to agree on what time it is.
Questions that conference did not answer are faced by the ITU-R now:
What do we mean "What time is it?", and how closely do we need to agree?

For most civil purposes time is only relevant to the nearest minute;
leap seconds are unnoticeable.  For many astronomical purposes time is
only relevant to the nearest second; leap seconds are inconsequential
(in the moment, but they add consequentially).

Applications that require time to better than one second must
recognize that ascertaining the true time is a process dependent on
external sources and systems.  Correct operation of the process
demands verifying how each element in the chain of provenance,
real-time and ex post facto data acquisition and reduction, obtains
and handles its notion of time.  UTC, with or without leap seconds,
may not be the optimal time scale.  As did the navigators of the
Admiralty's fleet in 1833, each element of the system may have to keep
a log of how its chronometer differs from "true" time.

Applications such as telescope pointing and satellite tracking
inherently rely on UT1.  Much existing software for these applications
dates to the 1960s when the only conceivable time scale was explicitly
tracking Earth rotation as closely as possible.  Spacecraft tracking
software that does not acknowledge UT1 - UTC introduces a time of
arrival error of up to 1us.  In the absence of leap seconds this
error becomes unbounded, and telescope pointing will also be affected.

The IAU 2000 resolutions decreed changes in conventional models for
Earth rotation which were implemented in 2003.  In conjunction with
changes for celestial coordinate systems, almost no telescope pointing
or satellite tracking software currently conforms to IAU convention.
During this century all software must be rewritten to conform, but
absent leap seconds many systems that presume UT1 = UTC will break
within a decade.

Know time scale needs & dependencies of applications & systems

Be prepared for change from external sources of time scales

Participate in the creation of standards and infrastructure, such as a
navigator's log of offsets between our system clocks^H^H^H^H^H^H
chronometers and master clock(s), with or without leap seconds

Change happens.  Are your systems ready?