Re: [LEAPSECS] Comments on Civil Time decision tree

From: Rob Seaman <seaman_at_noao.edu>
Date: Tue, 27 Sep 2005 17:25:35 -0700

Perhaps I might expand on some of Bill Thompson's statements in the
context of the great convenience factor of using the current UTC
standard.

> The accuracy requirement for the delivery of UTC to the instruments
> is +/- 0.410 seconds.

High quality, cutting edge science doesn't always require nanosecond
precision. In particular, when you are talking about an
"experimental apparatus" such as the Solar System, in which light
takes 16 minutes to cross the orbit of the Earth, and many hours to
reach the outer planets, the few tenths of a second provided by raw
UTC may well be exactly what is required. UTC provides simplicity in
handling while retaining the option of later converting observed
timestamps to TAI and thus to some dynamical timescale.

> Each instrument team commands its own instrument directly. Time-
> tagging is done via UTC.

UTC - or any future civil time scale - provides a common clock to tie
together both scientific and logistical requirements between a
multiplicity of teams and team members. It is often the case that
different instruments on a spacecraft are operated by different
teams. Those teams have to address scientific concerns - they also
have to interoperate with FedEx and with each other.

> I don't have the number in front of me, but I believe that the
> timing requirements for the SOHO spacecraft is even more stringent,
> on the order of 0.1 seconds. This is to support the
> helioseismology instruments.

I worked on an asteroseismology experiment several years back in
which 40,000 spectra of Procyon had to be acquired on an even
barycentric time grid. Cadencing the exposures required converting
Earth time (UTC) to the time as the wavefronts passed the Sun and
triggering on the predicted clock tick. Sure, I could have
referenced the barycentric corrections to something other than UTC,
but I challenge anybody to engender confidence in such a conversion
from an obscure, moving, Earth clock to a remote location a hundred
million miles away. I can handle one or the other, but not both at
the same time, with anything like an intuitive feel.

> There's a one-second period of ambiguity on the STEREO spacecraft
> whenever a leap second is inserted, and time
> critical operations will be avoided during that second.

So, presumably this is an example where handling a rare - but
necessary - phenomenon "properly" was deemed to be not cost
effective. However, they simply redefined the problem such that the
"outage" was reduced to its minimum duration of one-second. In six
hundred years, one might expect many affected parties to do the same
- but of course, the minimum outage then will be 3600 seconds. On
the other hand, we've heard apocryphal horror tales of GLONASS
falling from the sky and other Ghostbuster scale apocalyptic events
resulting from a leap second. Even if such projects chose not to
handle leap seconds transparently, why should their system outages
persist for more than the second achieved here by another space mission?

> For both missions, one has to deal with a significant light travel
> time, much larger than the required time accuracy.

Much of the discussion to date has implicitly assumed that civil time
here on Earth can simply be transported across the Solar System or
the Galaxy as needed - that it is trivial to correct for distant
locales. On the scale of tenths of seconds, this is likely true for
spacecraft bound to the Sun. One questions if this is achievable -
or desirable - at the level of nanoseconds, or perhaps even of
microseconds or milliseconds, for very many scientific or utility
purposes for Solar System travel.

> The process of taking this into account is essentially the same, no
> matter where you are in the solar system, and no matter whether you
> feed the results back to the spacecraft, or simply take it into
> account on the ground.

And more to the point - why should such projects have to be
responsive to such alien requirements? Presumably there were any
number of factors involved in determining the timekeeping choices of
these space projects as a response to various use cases. Is there
any particular reason that they should be forced to put their system
clock on the ground rather than on the spacecraft simply to meet the
one-size-fits-all expectations of others?

We've heard many times in many ways that atomic clocks are orders of
magnitude more regular than our wobbling Earth. But this extends to
other physical phenomena. We all know that atomic clocks are
sufficiently regular time pieces that detection of relativistic
effects is trivial. Think about that - what used to require
exquisitely sensitive measurements during exceedingly rare
opportunities such as a transit of Venus can now be carried out by a
high school student (albeit a relatively well-heeled student) on a
weekend trip to the mountains. General Relativity for everyman -
Albert in a box.

How nice! Isn't that support for ditching that dirty old UTC for
21st century technology? Well - no. A single atomic clock is a loose
cannon. Its precision outraces its accuracy in tracking TAI,
precisely because it serves at the whim of its environment and the
laws of Physics. But, by definition, there is only one master UT
clock - the Earth itself. This provides a natural interoperability
that TAI can never achieve no matter how excellently synchronized its
ensemble of clocks. As we've seen, any nation could completely
ignore all international timekeeping rules - except one - and still
retain sufficient interoperability to participate in every activity
expected of a sovereign realm. Bank transfers, telecommunications,
air traffic control - all could be synchronized by, say, a junior
high school science teacher operating a transit telescope to recover
mean solar time once a semester.

Rob Seaman
Received on Tue Sep 27 2005 - 17:26:36 PDT