UTC - A Cautionary Tale

In article ,
Jonathan Thornburg -- remove -animal to reply wrote:

In sci.astro.research Rob Seaman wrote:
On July 5th, the head of the Earth Orientation Center of the IERS,
Daniel Gambis, made the first official pronouncement that the next leap
second may be the last leap second.
[[...]]

Steve Allen of Lick Observatory provides an excellent page of UTC and
leap second resources:

http://www.ucolick.org/~sla/leapsecs

I also found the following article to be interesting:

R. A. Nelson et al,
"The Leap Second: Its History and Possible Future"
Metrologia volume 38 (2001), pages 509-529
http://www.iop.org/EJ/article/0026-1...6/6/me1606.pdf

That article is accessible only from "computers" (i.e. IP addresses)
registered as belonging to an institute subscribing to that journal.

[[Mod. note -- Alas true. However, typing
"Nelson leap second Metrologia" into scholar.google.com
quickly found what seems to be the identical article at
http://www.cl.cam.ac.uk/~mgk25/time/...leapsecond.pdf
which (I hope) is free for all. Ich liebe google!
-- jt]]

[[quoted signature snipped]]
--
----------------------------------------------------------------
Paul Schlyter, Grev Turegatan 40, SE-114 38 Stockholm, SWEDEN
e-mail: pausch at stockholm dot bostream dot se
WWW: http://stjarnhimlen.se/

Paul Schlyter wrote:

In article ,
Jonathan Thornburg -- remove -animal to reply wrote:

In sci.astro.research Rob Seaman wrote:

On July 5th, the head of the Earth Orientation Center of the IERS,
Daniel Gambis, made the first official pronouncement that the next leap
second may be the last leap second.

[[...]]

Steve Allen of Lick Observatory provides an excellent page of UTC and
leap second resources:

http://www.ucolick.org/~sla/leapsecs

I also found the following article to be interesting:

R. A. Nelson et al,
"The Leap Second: Its History and Possible Future"
Metrologia volume 38 (2001), pages 509-529
http://www.iop.org/EJ/article/0026-1...6/6/me1606.pdf

That article is accessible only from "computers" (i.e. IP addresses)
registered as belonging to an institute subscribing to that journal.

[[Mod. note -- Alas true. However, typing
"Nelson leap second Metrologia" into scholar.google.com
quickly found what seems to be the identical article at
http://www.cl.cam.ac.uk/~mgk25/time/...leapsecond.pdf
which (I hope) is free for all. Ich liebe google!
-- jt]]

Yes. It is! Thank you very much for posting the open access link.
It is a very nice up-to-date reference article.

Regards,
Martin Brown

In article ,
Rob Seaman wrote:

On July 4th, the International Earth Rotation Service (IERS) announced
that the first leap second in seven years will be issued at the end of
December 2005.

On July 5th, the head of the Earth Orientation Center of the IERS,
Daniel Gambis, made the first official pronouncement that the next leap
second may be the last leap second.

What would this mean? This is such a drastic change in the philosophy
of timekeeping that it is hard to express. This would mean that civil
time worldwide - the clock on the wall, the watch on your wrist, the
time provided by your cell phone and laptop and television and by the
time signals on the radio - that civil time literally everywhere would
cease to have any connection to the rotation of the Earth.

Not quite -- note the condition 3:

# 3 - The difference of UT1 from UTC should not exceed 1 hour.

This imposes some connection between civil time and the rotation of
the Earth. If there would be no connection at all between civil
time and the rotation of the Earth, then no limit would be imposed
on the difference between UT1 and UTC.

And we already have this situation today - the only difference is that
the upper limit of the difference between UT1 and UTC is 0.9 seconds
instead of one hour.

An alternate interpretation could be that the Prime Meridian would begin
to drift from the observatory at Greenwich England.

Astronomers already have that situation. Ever heard about the term
"Ephemeris meridian" or "Ephemeris longitude" ?

For most purposes time-of-day would become just a polite fiction.

If so, it already is "polite fiction". And it became "polite fiction"
when mean solar time was abandoned some 100+ years ago, to be replaced
by standard time zones. And it became even more "polite fiction" when
daylight savings time was introduced. Some places have even occsaionally
used double daylight savings time, i.e. changing the clock two hours.

For others, time-of-day would become a nightmare to calculate and
correct for from tables downloaded from the internet or corrections
typed in manually with all the usual ambiguity of sign and magnitude.

Well, yes, people would have to learn how to add and subtract and to
get the signs right, etc. It could perhaps be a bit of hassle to
some. But would it really be a "nightmare"? The only difference to
most people would be that the times of sun/moon-rise/set would drift
slowly by up to an hour, to suddenly be reset when a "leap hour" was
added. Would this be a "nightmare"? Well if you think so, how are
you able to deal with switching between standard and daylight savings
time - not twice a millennium but TWICE A YEAR ????

A link to the official proposal is available at the bottom of that
page, along with a list of committee members who can be contacted with
comments. I have a hard time expressing how much I detest both this
proposal and the process through which it has been made. Gambis is to
be applauded for finally bringing it to light after six years of
furtive discussions within the precision timing community. Belgian
astronomer and mathematician Jean Meeus comments on the proposal:

http://www.ucolick.org/~sla/leapsecs...vent2005-07-08

Jean Meeus' actual message about this can be found he
http://www.mail-archive.com/leapsecs.../msg00476.html

A few comments:
-------------------------------------------------------------------------
# (1) A first inconvenience of the change would be that the mean Sun
# would no longer transit the meridian of Greenwich at 12:00, "official
# time", and this would be the end of a long tradition. Very regrettable!

That tradition has already ended with the introduction of standard
time zones and daylight savings time. Spain for instance resides at
about the same longitude as Great Britain (the Greenwich Meridian
passes through Spain), yet Spain uses Central European Time. As a
result, the Sun will transit near 13:00 in Spain on standard time.
And in summer, when daylight savings time is used, the sun will
transit in Spain near 14:00.

# (2) Acceptance of the change would result in three separate uniform time
# scales running "parallel" to each other, indeed a ridiculous situation.
# Presently, we already have the Dynamical Time and the Atomic Time (TAI)
# which differ by a constant value, 32.184 seconds. If the US proposal is
# accepted, then we would have a third parallel time scale, the (new) UTC.
# There is no need to have a proliferation of those parallel time scales.
#
# (3) As long as the difference UT-UTC remains smaller than 0.9 second, this
# difference can be neglected for many applications where no extremely high
# accuracy is needed. This the case for the instants given for occultations,
# the phenomena of the satellites of Jupiter, etc., as published in various
# astronomical almanacs. These instants are given in UT, and we can simply
# use the time given by the radio signals which are in UTC, and consider them to
# be equal to UT. That would no longer be the case if a new definition of UTC
# is accepted.

Yep, a new time scale should really be avoided. OK, we already have UTC
but for many purposes UTC and UT1 can be considered equal, but that would
no longer be the case if UTC-UT1 was allowed to become up to 1 hour.

And in addition we also have GPS time ..... there are too many time scales.
One way to deal with that would be to set the future civil time equal to
TAI.

# (4) A newly defined UTC would be a problem when constructing long lists of
# astronomical phenomena such as lunar eclipses or transits of Mercury and
# Venus. Suppose we want to construct such a list for the years 1000 to 3000.
# What time scale should we use?

In such a case the choice is clear: one should always use TT for such
lists.

# If we choose UT, then times given for, say, the year 2500 would not be
# consistent with the official time, which will be the (new) UTC.
# If, instead, we choose UTC, then there would be another problem: of course
# for years before 1900, UTC would be meaningless, as in those years there
# were no leap seconds and even no time signals! And events taking place in,
# say, 1975 and 2500 would not be comparable, because the UTC used in 1975
# would not be the new UTC.
#
# (5) It is proposed that "The difference of UT1 from UTC should not exceed
# 1 hour." This means that a leap *hour* should be introduced when the
# difference between UT and UTC becomes too large, which would be the case
# somewhere between the years 2800 and 3200. Of course the exact year is not
# yet known presently, as it depends on the slowing down of the Earth's
# rotation. Consider, for instance, the transit of Venus of 14 June 2984. First
# exterior contact (for the Earth's center) will take place at 10:10:23 Dynamical
# Time.
# This will be 10:09 UTC if the US proposal is accepted. However, if the
# leap hour is introduced before A.D. 2984, then the instant would become
# 09:09 UTC. Consequently, presently we don't know whether the transit will
# begin at 09:09 or at 10:09 in the proposed UTC scale, and hence it is not
# possible to create a long list of events with the instants expressed in
# UTC.

Very True! Therefore avoid UTC in those situations and use TT instead. Or
possibly UT, if a good approximation to TT-UT is available for that time
(which excludes events far into the future and too far into the past).
But TT is the preferred choice of time scale for events far into the future.

Since UTC is a civil time, it can be changed by political decisions
any time in the future. Even if we now succeed in keeping UTC
unchanged and continue using leap seconds, future generations can,
after all of us here are dead, get annoyed at the more frequent leap
second and decide to change UTC, in 100 years or so. Therefore one
should never ever compute the time for astronomical events in the far
far future in UTC, simply because we don't know what "UTC" will be
around year 2500 or 3000. We have absolutely no control over what
decisions future generations will make.

(In an astronomy book I have, written in 1930, there's a very good
example of unawareness of possible future political changes: there
the author used Oppolzer's "Canon der Finsternisse" and sorted out
and listed those solar eclipses which are total somewhere in Europe,
from the year 1700 to the year 2135, when Oppolzer's Canon ends.
Now, the problem here is his definition of Europe - he included the
European colonies in North Africa, like Algeria and Tunisia, in his
"Europe", and includes several eclipses which are total there but not
in what we today consider to be Europe. Probably he was completely
unaware that these colonies would cease being colonies long before
the year 2130....)


# (6) Finally, for sundials, too, the situation would be complicated.
# Presently, to convert true solar time (as given by a sundial) to "official"
# time, we have to take into consideration: the longitude difference with
# Greenwich, the equation of time, and the fact that we use or not the
# "summer" time. But if the US proposal is accepted, a further correction
# would be needed: the difference between UT and UTC, a difference that is
# now negligible, but that will gradually increase over the years if the US
# proposal is accepted.

That wouldn't really mean significantly more trouble. If this proposal
is accepted, the difference UT-UTC can be regarded as constant for sundial
use (i.e. an accuracy of one to two minutes) during a major part of a
human lifetime. The longitude difference to your time zone is another
constant. Add those constants and learn the sum of them as your new
"constant adjustment" to your standard time. After some 30-40 years,
adjust the value of the constant slightly.


# Finally, I don't understand why the ITU and the people of GPS insist to
# suppress the leap seconds. Are they really unable, notwithstanding the
# modern technique of the 21th century, to handle this "problem"? Should
# astronomy suffer because those guys cannot handle the leap seconds easily?

Jean Meeus has obviously never worked in a large software project... :-)

The "problem" with leap seconds is their unpredictability: once every
seven years, or every three years, or every year, or twice a year, or
in the future even more often than that, a leap second is inserted,
with only a few months notice. The unpredictability of those leap
seconds mean that they must be handled manually (in those cases where
they must be handled). This was quite feasible in the past, when
computers were huge "dinosaurs" managed by a team of men in white
coats. But since then computers has shrunk -- and multiplied.
Today's computers are far smaller, far more powerful, and far more
common that the "dinosaurs" of the past. Most people in the
industrialized world carry a few "computers" with them, probably
without thinking of them as computers: their wris****ch, their mobile
phone, their PDA, their mp3 player. A modern car has dozens of
computers. A train has hundreds, and an airplane .... probably many
more than that.

With such a vast number of computers running all the time, and our
increased dependency on them, it is a top priority to make as
much as possible automatic. Timekeeping is obvious here, it should
be automatic.

There's a common time measure in the software world (yep, another
time scale!) which is called "UNIX time": it is the number of seconds
since 1 Jan 1970, 0h UT. Nowadays UT here means UTC -- but will this
second count include the leap seconds or not? Usually the "UNIX
time" disregards the leap seconds - but then there may be a "hiccup"
in some systems whenever there's a leap second. Some systems may
have tried to ac**** for the leap second -- but then their "UNIX
time" will be inconsistent with the "UNIX time" of other systems,
and this can cause trouble whenever time critical data is to be
exchanged between systems.

So that's basically the reason why the software world wants to get
rid of the leap seconds: it's not that they're impossible to account
for of course. But it's because it's a source of bugs.

-------------------------------------------------------------------------





3 - The difference of UT1 from UTC should not exceed 1 hour.

The width of a time zone is one hour. The idea of leap hours to be
issued every half millennium or so is equivalent to no civil
time-of-day at all.

Then what about going back and forth between standard time and daylight
savings time? That too is a jump in time of one hour.



--
----------------------------------------------------------------
Paul Schlyter, Grev Turegatan 40, SE-114 38 Stockholm, SWEDEN
e-mail: pausch at stockholm dot bostream dot se
WWW: http://stjarnhimlen.se/

Paul Schlyter wrote:

In article ,
Jonathan Thornburg -- remove -animal to reply wrote:

In sci.astro.research Rob Seaman wrote:

On July 5th, the head of the Earth Orientation Center of the IERS,
Daniel Gambis, made the first official pronouncement that the next leap
second may be the last leap second.

[[...]]

Steve Allen of Lick Observatory provides an excellent page of UTC and
leap second resources:

http://www.ucolick.org/~sla/leapsecs

I also found the following article to be interesting:

R. A. Nelson et al,
"The Leap Second: Its History and Possible Future"
Metrologia volume 38 (2001), pages 509-529
http://www.iop.org/EJ/article/0026-1...6/6/me1606.pdf

That article is accessible only from "computers" (i.e. IP addresses)
registered as belonging to an institute subscribing to that journal.

[[Mod. note -- Alas true. However, typing
"Nelson leap second Metrologia" into scholar.google.com
quickly found what seems to be the identical article at
http://www.cl.cam.ac.uk/~mgk25/time/...leapsecond.pdf
which (I hope) is free for all. Ich liebe google!
-- jt]]

Yes. It is! Thank you very much for posting the open access link.
It is a very nice up-to-date reference article.

Regards,
Martin Brown