[LEAPSECS] PTTI leap second discussion
The following is to be published in the Proceedings of PTTI-2000. I
understand it is almost ready to go to press. As I have indicated, the
paper by Chadsey and McCarthy won't be available electronically.
LEAP SECOND PANEL DISCUSSION
RONALD BEARD (NRL): Good morning. As many of you are aware, the ITU is the
international organization that regulates many of the things that relate
across
the different countries and organizations that require coordination. Such
things as radio spectrum, telecommunication standards, and many of these
things
that are required for nations, companies, organizations, and systems to work
together. You may also be aware that time and frequency broadcast services
in
the nations are also based on recommendations of the ITU-R. Within the
ITU-R,
Study Group VII, Science Services has established a working party, 7A, which
looks at and recommends time and frequency services in relationships to time
that are broadcasted by these services and their relationships to different
time
scales.
Many of the things that Dennis has talked about earlier are based on
ITU-R recommendations on time and frequency broadcasts and their
relationship
to time scales. The tolerance between the UT and UTC are based primarily on
recommendation ITU-R, TF460-5, which was established first in 1970 and
modified
in '74, '82, '86, and '97---we do a lot of modifications sometimes, to the
0.9
second difference between UT1 and UTC. These recommendations, opinions, and
other things that the ITU establishes become ultimately established in legal
facts in the various member states. These also regulate international
treaties
and other things relating to frequency spectrum and services used
internationally. Consequently, UTC is in a number of other recommendations
relating to the use of UTC frequency; the use of the term ``UTC,'' what does
it
mean, and how do you establish that internationally; how do you compare
these
international time scales and other time scale notations; and a lot of other
things that are looked at recommended by the ITU-R.
The way the procedures work within the ITU-R is they first establish a
question
to change, regulate, modify, or add some type of recommendation that may
affect
a number of these different treaties and things that are related to the
ITU-R.
Once these are studied and the questions are answered, these may end up in
recommendations. These types of recommendations were first studied, agreed
to,
and looked at by many people before they were established. The
recommendations
and changes to them are also based on different studies that may last a
number
of years and examined by a number of different people.
>From last year's result of many of these questions that are being asked that
Dennis just went through, a new question was established on the future of
the
UTC time scale. If you were to consider going through a different time
scale,
that might affect time and frequency coordination and broadcast worldwide.
These need to be accommodated in these types of recommendations. So a
question
was established last year as to what are the requirements if we change these
recommendations. Whom does it impact? Does it make any difference? What
are
the present and future requirements for this tolerance? Do some of these
changes make sense in the terms of the different systems, organizations, and
legal bodies that may use them? Does the current procedure essentially
satisfy
everybody or should an alternative approach be adopted? These kind of
general
questions were asked last year.
After this occurred or around that time, the issue also raised in the
consulting
committee for time and frequency and other scientific bodies resulted in the
Director of the BIPM writing a letter to the Secretary General of the ITU,
who
presides over the ITU-T and the ITU-R. The letter said that if these types
of
changes were to be contemplated, the ITU would need to take them up,
actually
incorporate them, and put them into effect. So as a result of this, the
issue
seemed to be a little bit more urgent than the normal method of studying the
question for several years and ultimately coming to a conclusion. So this
year,
a special repertoire working group, by correspondence, on the UTC question
was
established. These are members: I am fortunate enough to be the chairman
of
this group which is going to work this issue. We have adopted a plan of
action
of how to attack this. Since we are not exactly sure how much of an impact
this
is going to have and how much reaction we are going to get from the
telecommunications, radio industry, and the scientific communities, this is
a
preliminary plan to at least assess the situation, see what might be
necessary
in order to fully address this issue, and establish a new recommendation or
lack
of recommendation.
The first step is that the Director of ITU-R will be sending a letter to the
member states and section members within the ITU saying that this group has
been
established, what the basic plan of action is going to be, and what we are
going
to be doing. This is the usual formal step for the ITU to establish these
types
of groups and study plans. We had hoped that would have been out by the end
of
this month, but it is still not out yet, so it is still in the first step.
Once
that is done, then the participating organizations of sector members or
member
states would be then be identifying points of contact who would work with
this
special group to address the issues and assess the impact, or lack of
impact, or
changes that might be necessary to coordinate. We would then provide
additional
material on the question, distribute that, and begin the process of actually
doing the study.
We're now in the fourth step, and the first step really hasn't occurred yet
of
introducing this question and our intentions to study this problem and
address
it with the various member states and organizations involved at this
meeting.
So we have accomplished one of our tasks already. During the time after the
letter comes out and this discussion here, we will be collecting statements,
comments, and studies that may have occurred already from the various
agencies
to incorporate those into a database or a library of information to base the
studies on.
The next step would be to conduct a coordination meeting with the various
member
states or organizations and participants at the EFTF that will occur in
France
in March. Hopefully, we will be able to get enough information by that time
to
at least try to size the magnitude of the problem. It could be that nobody
cares and nothing happens. Or it could be that everybody cares and we are
covered up in information. We really don't know yet. So we would be able
to
assess the impact agencies that want to participate in the study and in the
study group. Then we can discuss and try to coordinate what really should
be
what I anticipate to be the final plan of action to be formulated at that
time.
Exactly what we need to do, do we need to do certain studies? Do we need to
do
simulations? Do we need to coordinate with different agencies? Or exactly
what needs to be done? We would hope to compile the results of that at that
time and report back to the next working party 7A meeting, which will occur
in
May in Geneva next year, so that we may either finalize the issue or
formulate
a new plan of action and try to determine how long it's going to take to
resolve
this issue and come to closure on it.
So basically that is what we are working to right now. The working method
for
this is still being established as well. We would hope to establish a
contact
point at the ITU for e-mail or providing of material through their e-mail
server
that they have there. That has not been established yet. So at this time,
I
would offer up---I hesitate to do this---my e-mail address at the ITU. Any
material sent to me by participants who would be participating in this, or
representatives of member states, sector members, or scientific
organizations
could contact me. We would like to limit participation in the study group
itself to representatives of agencies so that we can try to maintain a
reasonably sized body. They could act as the focal point for their
respective
organizations and provide that to the study group.
Once the e-mail reflector is actually established, and hopefully that will
occur
within the next month or so, you may either send e-mails to me directly and
I
will distribute it to the rest of the working group, or later on you can
subscribe to the e-mail. You can get involved and see what is going on in
the
formulation of the plan or submit inputs that you have representing your
organization. So basically, that is what we are up to. Thank you.
DENNIS McCARTHY: Thanks, Ron. The next member of the panel that
we have here is Steve Malys from the National Imagery and Mapping Agency.
And he's here to point out some of the concerns about making any changes in
the definition of UTC. So Steve is here to argue for the status quo---just
to
more or less leave it alone.
STEVEN MALYS (NIMA): Thank you, Dennis. As we heard from Harold earlier,
on
the
surface of the earth for a ship or any other navigator, 1 second of time is
about a quarter mile. But if you think about a satellite in low-earth
orbit,
1 second of time corresponds to about 7 kilometers of movement. So it's
obviously very important that we account for 1 second very accurately. At
the
GPS altitude, which is higher than low-earth orbit, at that altitude a GPS
satellite moves about 3 kilometers in 1 second. So leap seconds become
very important, and procedures have been developed over the years to take
leap
seconds into account when they do occur. Keep in mind that over the last
few
decades we saw the evolution of time. Satellite systems have become
operational
over those last few decades and they are no longer experiments. GPS is
obviously a very good example of operational system that we wouldn't want to
have an interruption of service for any reason. There are other operation
systems, particularly in the DoD, that the DoD depends upon very heavily. A
good example of another satellite system would be something like the Defense
Support Program, which looks at the earth for missile launches. You don't
want
to have an interruption of service to a system like that.
Another example is the operation that takes place in Cheyenne Mountain out
in
Colorado, the space surveillance operation that keeps track of more than
8,000
objects in orbit. It's a very operational system. There are strict
procedures
setup to handle things like leap seconds, and they have evolved over the
years.
Certainly, mistakes were made in the past and procedures have been refined.
It
is my experience that things have improved significantly. People have come
to
understand much better how to handle leap seconds to prevent the problem.
We
know, of course, that the Russian GLONASS systems has some difficulty
dealing
with that kind of operational system. It is my experience that within the
U.S.
DoD, we have become much better at dealing with leap seconds when they do
occur.
So if there were changes to be proposed, and I'm just thinking of one of the
cases that Dennis pointed out, if we were to allow the tolerance between UT1
and
UTC to grow more than 0.9 seconds, what would happen? What would we
actually
have to do within the U.S.---DoD and other U.S. Government systems---to
accommodate that? Keep in mind, for all of these satellites that are in
orbit,
we're doing orbit determination for these satellites in an inertial
reference
frame. We integrate the equations of motion in the inertial reference
frame,
but our tracking stations typically are on the ground. So our tracking
stations are rotating with the earth, which means we need very precise
knowledge
of UT1. We routinely account for the difference between UT1 and UTC. That
parameter is predicted on a routine basis. That is a very important piece
of
that transformation from the earth-fixed reference frame---it is the
inertial
reference frame.
So no matter what happens to time scales in the future, we need to account
for
the earth rotation rate because the tracking stations are most likely going
to
remain on the ground---at least, some of them will. The earth rotation is
not
going to go away from the perspective of doing orbit determination. So if
we
were to change current procedures, systems like GPS and other DoD systems
would
require some modification to the software. Remember these are operational
systems, so a change of UT1 minus UTC graded on 1 second would mean there
would have to be some effort initiated within each of these systems by
somebody
in the government who runs them. Most satellite systems are operated within
the
U.S. by the government. Of course, there are commercial systems evolving
now.
You may have heard of the commercial imaging systems like ICONOS, which is
operational. There are other commercial systems as well. I'm mostly
referring
to government-operated systems here. But there would have to be some
initiation of software changes, documentation changes, other changes to the
procedures that train personnel, and a significant amount of testing. It
wasn't
that long ago that Y2K was the big deal, and there was a lot of testing that
went on. Even with all the testing that went on, there were still minor
problems that occurred. But it's that type of thing that costs a lot of
money
to do, and there are strict interfaces set up among different satellite
systems
within the DoD. Those interfaces would have to be carefully looked at and
analyzed.
Those are some of things that would have to start off if we were to make a
change to the current procedure. Keep in mind that the original definition
UT1
minus UTC, as we know it today, cannot exceed this 0.9 seconds. Well, many
software systems were designed with that piece of information in mind.
There
are lots of software packages that treat that as a tolerance. The software
will
not allow UT1 minus UTC to be bigger than 1 second or it declares that there
is a problem of some kind. There is a lot of range checking on parameters
that
go across different interfaces. Well, today these interfaces have this
limit
imposed in it. So it is similar to the problem that Dennis mentioned where
many systems that broadcast UTC parameters have a limit of 1 second. This
is another manifestation of that same kind of problem. These interfaces
would
have to be changed to accommodate something greater than 1 second.
So really, this is really a practical argument for keeping things the same.
If we wanted to make changes, it is going to cost money. Like any good
government institution, it is tough to change something once it gets
institutionalized and operational. There would be money involved to the
U.S.
taxpayers, and other countries that run systems like this would have to
allocate
resources to make changes. Of course, experimental systems or systems that
are just being designed now would be easier to change. That is mostly
the operational system that I am talking about here today.
The typical procedure to change an operational system is to first obtain a
rough
order of magnitude from the contractors who are working with these systems.
People who are involved in those systems will present a request for change
to
some configuration management board. If it is approved, there would have to
be
funds identified to go and change it. Believe it or not, even for a little
thing like changing a tolerance from 0.9 seconds to something greater, you
need
to go through this whole process for an operational system. We would do all
of
that for no identifiable benefit. It would just be another way to handle
the
difference between earth rotation rates and some atomic time scale. There
would
be no improvement in accuracy that I have been able to identify. It would
be
just be another way of handling a procedure different from what we do today.
That would be difficult to sell. If you are going to argue for making a
change
to one of these systems, you have to explain to the people who manage the
funds
for those system why they should do this. That would not be an easy
argument to
make because there is no identifiable benefit to any of these operational
systems. It's just a different way of handling it. That is really the
practical side that I am here to talk about today.
McCARTHY: Thanks, Steve. That is a real concern, one of the big issues.
Now I would like to open this for discussion. Wlodek, did you have
something
that you would like to say?
WLODZIMIERZ LEWANDOWSKI (BIPM): What I would like to say about this issue
is
that the BIPM does not have an official position on this issue. We are just
taking calculations, computing UTC. We do not have a specific point of view
or
expertise to express ourselves on the issue because we do not have touch
with
the users and so on.
But I would like to make some comments about the possible use of TAI,
because during CCTF, there was some discussion in Europe on this issue and
some exchanges on this matter. After the last CCTF, about 2 years ago, our
director wrote a letter to the operators of Global Navigation Systems
informing
them that if they have any troubles with UTC because of leap seconds, the
suggested alternative is TAI. That is fine. If they use TAI for internal
time
scale
systems, I argue with this.
But what worries me is that this letter is maybe not expressing as clearly
as it
should that the use should be limited only to such internal applications. I
also saw in another document of ITU somewhere that TAI was suggested as an
alternative to UTC for much broader applications. That worries me because
this
means, already having UTC, a legal time scale, we are suggesting
introduction
into civil life another time scale, TAI, which does not have legal meaning.
But
because of this discussion about leap seconds, some people begin to
say---and
this happened at the last CCTF---we in fact have a time scale that does not
have
leap seconds, so let's use it. But this is a problem. Because if people
begin
to use TAI for civil applications, which will be apparent and visible to the
public, that will be a problem because we will be going to two time scales
and
that will lead to mistakes and possible disasters even. Because now the
difference between two time scales is 32 seconds.
For example, on BIPM's Web site, we can see making two time scales, UTC and
TAI.
That worries me a little bit, because many people go into this Web site and
they
ask what the matter is and what time is it. What should I put my watch on,
UTC
or TAI? Which is the right time? In fact, the right time is not UT1, it is
UTC. We should not make mistakes with TAI time. TAI is something else. In
fact, TAI, I should say, is a UTC system time. It is an internal time scale
to
generate the final, official, legal time scale for the world which is UTC.
So
what I would like to point out, in summary, we should not go too quickly to
TAI
as an alternative because we would go into some big trouble. When I spoke
about
this issue with some people, they said that GPS time does not have leap
seconds
and people are getting GPS time. In fact, what I know of this, according to
my
experiments, GPS users don't use GPS time. GPS time is an internal time
scale
to the system. GPS users are using UTC(USNO) as broadcast by GPS so they
use
the right second.
McCARTHY: We often have that question about GPS time. People often think
they're using GPS when they are not really using what is strictly defined as
GPS
time. They are using UTC, but they call it GPS time. Demetrios, if you
could
just briefly say something about URSI. Demetrios conducted a survey of an
URSI
group to give its opinions. If you could just say briefly what the results
of
that survey were.
DEMETRIOS MATSAKIS (USNO): You probably know this too. A lot of times,
people
turn their GPS receiver the wrong way and we get a phone call on why their
receiver is 13 seconds off. That's an easy problem to fix. We did a
survey---I
actually talked about this a little bit at the last PTTI when the survey was
in progress. It was under the auspices of URSI, but I tried to send it
everywhere. I asked people to distribute it around, and I got several
hundred
replies. The committee prepared a final report, which we sent to everybody
who
sent us a reply. If any of you want it, we can send it to you. We set up a
chat group to talk about the problem after the report went out. The
computer
got wiped out by a virus, we believe, about 2 weeks ago, and unfortunately,
it
wasn't adequately backed up. We recovered most of what we had in that, and
I want to give it to Ron Beard. I'll give him all of the comments that
people
made about the report and after the report.
BEARD: If you look at the international community and the people
who might be involved in this, it would get extremely large and complicated.
MATSAKIS: I made some notes and but did not come prepared to talk.
Typically,
the majority was against the change. Most of the people I got were
complainers
who didn't like the change. That's typical: when you say you want to do
something, those who are in favor don't say much about it. When you look at
why
they were opposed to it, most of the people were opposed to it for reasons
that
were not related to money or anything practical. Steve was one of two
people
who came in with a practical objection. They both were the same, having to
do
with the expense of going over codes for large, expensive systems.
But the greater tone was a lot of very strong, sometimes emotional, people
who
said, ``don't mess up our clocks.''
But there was nothing religious that I got, which was a bit of a surprise.
I
didn't go out of my way to contact religious leaders. I got one opposition
from
Saudi Arabia, and I asked him why; and the answer was he was concerned about
amateur astronomers. And that was one class of user, amateur astronomers
who
cannot get the number of leap seconds or want to know how they can
point their telescopes.
That was along with the other people who were giving reasons why you just
don't
want to have time going off. There was a problem with the NIST WWVB. When
they
broadcast the difference between UT1 and UTC, they only have a fixed area in
their format. So that will eventually fail, and quite quickly. Any user
who is getting that correction off of WWVB series will run into troubles.
They
don't know how many people, if any, are using that system, but they may find
out
if this thing happens.
So those are all the comments I can make just now concerning the notes I
made on
this.
THOMAS CLARK (NASA Goddard Space Flight Center): A couple of things that I
want
to make comments on relate to Dennis's comments earlier. First of all, just
so
you all are aware, of course, UT1 is an astronomical definition of time.
The
current arbiter of astronomical time is VLBI observation techniques. That
comes
through programs in this country from NASA, namely my program and Dennis's
at
USNO.
One of the things in talking about the parabolic type of tidal model that
was
not included in that, which is one of the things that concerns me, is that
in
addition to the predictable tidal terms there are a series of essentially
random-walk phenomena. The recent El Nino that happened, the transition
from
El Nino to La Nina caused a sundial error of about 5 milliseconds. It
happened to be about 3.5 milliseconds on one side and then 1.5 milliseconds
as
it recovered into the other. So the peak-to-peak range was about 5
milliseconds. So there was this random-walk curve of 5 milliseconds due to
one
discrete, albeit couple-year-long weather events on relatively short times.
Those happened and it essentially has to be treated as a chaotic, stochastic
noise process in terms of the clocks.
In addition, Dennis alluded to the decadal scale variations. Most of that
comes
from essentially climatological variations in our atmosphere and, more
important
than that, climatological variations in terms of the circulation of fluids
inside the earth. Those have to be considered in all of this. They are not
really predictive quantities, at least at our current level of knowledge.
They
have to be treated as a random-walk term. So I just wanted to make sure
that
people realize that it is not just the soli/lunar tidal drag of the earth
that
cause these effects.
I tend to come down on the side of let's not make changes based on the ``it
ain't broke, don't fix it'' model. The current scheme keeps the attention
of
the populace. In many cases, the population, the human experience lives by
astronomical events. I'm surprised at Demetrios's comments that he didn't
get
more in the way of religious types of input. I think it could be that URSI
did
not really solicit the opinions of religious and civil communities. They
were
soliciting scientific communities. Certainly, several faiths have events
that
are scheduled by either solar or lunar events. Many of the fundamentalists
that
have those beliefs are also potential hostile enemies of more
technologically
advanced who don't want to have the feeling that Americocentric ideas are
being
crammed down their throat. So I would offer the caution that if changes are
made, it could be viewed politically/religiously as being a very negative
thing.
I think that does have to be factored into all of this. We really need to
think
about it.
Ron Beard talked in terms of the ITU events. A couple of us were talking
back
here, and we are a little surprised that the ITU views this as a crisis
event
that is putting it into a fast-track status. I'm not sure that it is a
crisis
event. The civilization is certainly living with the current situation.
Until
the year 2600, when we hit a definition problem that the tidal effects make
it
so that it drags us into a 6-month refresh interval, not being adequate to
maintain the current definition, there really is not a serious problem with
the
status quo. I hardly believe that making a decision that doesn't have an
effect
until approximately 2600 puts us into a crisis condition. So I guess I
argue
for maintaining the status quo.
McCARTHY: I would just like to offer one thing there. The decadal
variations
are what you saw in that simulation, so that's where that comes in. But
it's
not just 2600, because those decadal variations could force us to go to more
than every 6 months insertions within the next 100 years. The crisis
comment
still stands.
BEARD: A crisis in the ITU---you are not familiar with the ITU time scale,
obviously. ITU time and whatnot are very much governed by a lot of
bureaucratic
procedures. Putting it on the fast track means it will happen before the
next
decade happens, more than likely. So I did not mean to say that we were
going
to general quarters to address this by next month, certainly not. The issue
was
significant enough---perhaps I should have said---to put special focus on
it.
Perhaps shorten some of the time. But help focus the study and the
highlighting
of it to assess the full impact. As Steve pointed out, there is a
significant
impact on various section members, on costs of doing changes. Certainly,
the
status quo minimizes these types of status change cost.
However, some of the other issues are---let me say, many systems are using
internal time scales rather than the official time scales in order to avoid
leap
seconds. GPS time, I think, is a classic example. Many other systems are
doing
this so that they can have a continuous time scale and do a lot of automatic
processing that a discontinuous time scale does not permit. So if you look
at
the other systems that are coming on, the relationships and trying to bring
all
of these systems onto a common time scale have a significant problem by
having a discontinuous time scale. One of the more significant decisions
that
is going to be made in the next year or so is the Galileo time scale. Its
relationship to the other satellite systems, GPS, GLONASS, possibly other
telecommunication systems on the ground that all these things need to be
seeing
this with, what time will they use? Will they use another internal time or
will
they use international standards? So that is part of the reason.
JUDAH LEVINE (NIST): I have just a few comments. First of all, David Mills
and
I are presenting a paper on the idea of simultaneously transmitting UTC and
TAI
over the Internet with the idea of addressing one of the solutions that you
proposed of making TAI more available. That way you have both UTC and TAI
sort
of simultaneously. That doesn't take a position on the question, it kind of
provides a solution that is available today without having to wait for the
ITU
to go into crisis mode. I think that is the first issue.
The second issue is that I have written to Demetrios about the finite
resolution
of our time services which would be broken if UT1 minus UTC were allowed to
become bigger than a second. I don't think that is a big issue. We could
redefine the time service transmissions to have a different resolution. I
don't
see that as a real issue.
When I was involved with Demetrios's questionnaire---I live in an
astrophysical
and astronomical institute, and so, of course, I had a long line of folks
out
the door discussing the astronomical and astrophysical religious fervor of
``you
mustn't change the time'' and so on. But one of the things that emerged
from
that discussion is that there already is, of course, an annual term,
because,
when you talk about mean solar time, that is not a physical time. That's an
average over a year. There is an annual variation in the time, which is
like 15
minutes. Those folks have managed to cope with that 15-minute time without
any
difficulty at all. Of course, the whole leap second effect is a
perturbation on
this 15 minutes, and will remain a perturbation on this 15 minutes for quite
some time to come. So I don't think it is really such a big issue as has
been
made, in a sense that you don't have to deal with this 15-minute question.
I guess the final comment I have is that I have been involved on and off in
the
definition of the Jewish religious calendar, which is locked to sunrise and
sunset as defined locally. And we just use the tables from the Naval
Observatory; there is no issue. We just define sunset, we look up what time
it
is, we print it on the calendar; and that is the end of it. It is just not
a
question.
WILLIAM KLEPCZYNSKI (Innovative Solutions International): I have a comment
to
make about UTC and GPS time. A lot of papers I've seen at some of the
conferences, especially the ION conferences, are now referring to UTC(GPS).
But
really, they are referring to GPS time. So GPS time is neither UTC nor TAI,
or
even close to it because it,s off by about 12, 13 seconds or so. So that is
the
problem that Wlodzimierz was referring to when the difference between TAI
and
UTC is a whole integral number of seconds. The difference between GPS time
and
UTC is also still a whole number of seconds, and it has to be really kept in
mind by the users.
THOMAS CELANO (Timing Solutions Corporation): I'd like to address a point
that
Steve made and I believe something that was missing in the previous talk. I
think cost to the user community is an issue that is going to drive a lot of
this. But I think that you missed a point in that there is a cost
associated
with how we do it now. You made a point of all the systems know how to deal
with it now. But we install timing systems in a lot of places that
literally
shut down when the leap second happens. There is a cost associated with
that
that needs to be taken into account and if we're going to consider making
the
change. Because that cost would go away if we could become operationally
continuous over these intervals. I don't think that we should use the
changes
in RFC that are going to be required to do all this stuff, but I don't think
we
should use the process of change as a reason not to do it. I think we need
to
be able to recoup the operational cost that we're spending now in testing
and
the loss of time during the steps in a more continuous fashion.
One thing that I think was missing in all of the options that you guys
provided
for how we deal with this is the cost associated with each one. I think
Steve
made that point very well. Different ideas are going to have different
implications cost-wise, and one thing that produces cost is predictability.
You
had a couple options that had predictability, and if you have that, it
really
simplifies a lot of things operationally and it does reduce cost.
MALYS: One comment to that that I would like to add. Speaking of shutting
down, one little anecdote I wanted to share: About 15 years ago, I was
doing
orbit determination on the Navy's Transit satellites. There was a gentleman
working with me who did the prediction of earth orientation, including UT1.
Well, every time there was a prediction of leap second, he would go on
vacation
because it was too stressful for him to handle. So I took over his job, and
I
owe him a thanks for introducing me to leap seconds.
HUGO FRUEHAUF (Zyfer, Inc.): I'm dressed in black to represent all the
religions of the world today. The three major high-profile religions that
deal
with time. First of all, in Judaism, as Judah has already mentioned, it is
a
matter of sunset, and that's taken care of, as he mentioned. In
Catholicism, I
know of no particular issues with respect to time, so I think that part of
it
is okay. In Islam, we're dealing with pointing to Mecca, and that can
certainly
be done without the leap second consideration. So there you have it, no
mystery.
CHADSEY: One issue about the leap second and the timing for the religious
community was from the people we've been able to talk to and get the
information
from. Most of them base it on the tables produced by USNO, or you can look
them
up in several books. Those are general tables, and it changes by a minute
for
about every 9 miles that you move in position, so there is a little bit of
leeway there.
The people who are very orthodox believers say ``Well, we need to worry
about
the refraction of the sun around the earth, and what about the mountains and
things like this?'' So a lot of those folks, through their religious
upbringing
and their training, have come to realize, ``well, let's adjust it by 4 or 5
minutes,'' whatever their leaders have instructed them on. So they can
account
for these small variations of not only their location, but also the
scientific
fact of refraction of the sun and things like that.
So they're handling it and it is a minor problem for them. The major
problem is
more for the scientific and the communications industry. The costs of it
are
going to be ridiculous whichever way we go.
ROBERT NELSON (Satellite Engineering Research Corporation): The principal
difficulty with the leap second is the operational problem that it presents
to
complex timing equipment. So, therefore, I would speak to eliminating the
need
for the leap second and continuing a time scale such as UTC, which means
continuity with the present civil time scale. The difference between UTC
and
UT1 can be applied mathematically by those people who are best equipped to
understand it, who are the celestial navigators.
I think history can provide us with a guide. What we are facing today with
the
atomic clock technology is that we have a paradigm shift. In the 14th
century,
when mechanical clocks first became possible with the invention of the
escapement, they were used to ring church bells. It introduced to the
public
perception of time, for the first time, the notion of an equal hour rather
than
an unequal hour.
In the early part of the 19th century, astronomical ephemeredes were
constructed
with apparent time as the argument, instead of mean solar time. But when
pendulum clocks advanced to the state that they could reliably provide a
direct
measure of mean solar time, then the equation of time, as Judah alluded to,
with
the maximum difference of 16 minutes between apparent and mean, was used in
reverse. Instead of being used to determine mean solar time from apparent
time,
which was directly measured by the altitude of a star or the sun, it was
being
used to determine apparent time from the measured mean solar time as given
directly by a clock. So I think the time has come in the 21st century, in
modern society, to break the tie with the sun all together. After all, in a
given time zone, the clock reading can be off from apparent time by as much
as
half an hour. We use the difference between Daylight Time and Standard Time
regularly, which is a difference of a whole hour.
One of the options mentioned was the possible use of TAI. I think that was
already addressed by the fact that it is different from UTC by 32 seconds.
So
if we went to TAI, we would have to change our clocks by 32 seconds. It is
much
like the calendar had to be changed by 10 days in 1582 when the Gregorian
Calendar was adopted.
So I would propose then that instead, UTC be maintained continuously without
leap seconds and that, if necessary, a new time scale, which one could call
``UT1C,'' could be provided, much as UTC is used today to provide the means
of
celestial navigation. Those people will need a direct measure of UT1. The
difference between these two could be provided, for example, by coded
signals,
much as D-UT1 is provided now to give the difference to the nearest tenth of
a
second.
Received on Fri Mar 30 2001 - 12:17:56 PST
This archive was generated by hypermail 2.3.0
: Sat Sep 04 2010 - 09:44:54 PDT