Re: Leap-seconds, the epsilon perspective
On Wed, Jan 29, 2003 at 12:33:48AM -0800, Steve Allen wrote: What for? Why should we (the people of the Earth) care about mean solar days? For some purposes, apparent solar time is important, but most of the time it's civil time that counts. Why should that be tied to mean solar days? Partly because eventually the scheme of turning UTC into a constant offset form of TAI requires a leap hour lest our descendants find themselves having their midday meal at 18:00 local civil time. This sort of pushing the problem off onto our descendants implies that we really don't have the right solution, just one that salves some needs now. But civil time already has leap hours (positive and negative) jumping around most parts of the globe twice a year. Furthermore, about half of the year I have my midday meal at about 20:00 UTC, and the other half I have it at 19:00 UTC. And when I travel, my midday meal time often shifts to some other UTC time. Oh, sure, the locals refer to the time as noon or 12:00, but that is not what my watch (which is set to UTC) says. It is also very unusual for it to be what my sundial would say (if I happened to bring it with me). Trying to tie civil time to mean solar noon is merely a historical convention. While global civil time standards have been been fairly consistent in using this technique (even if there were sometimes competing definitions for which meridian to use), it is not a fundamental requirement of civil time. There are very few locations on earth which use unadulterated UTC as their time base (not even Great Britain, ignoring the distinction between GMT and UTC, because about half the year they are on BST.) The requirements of civil time are very forgiving --- if local solar noon is within a couple of hours (sometimes more) of clock noon, that appears to be good enough for civil use (based on observation of current practice), and offset-from-time-standard is controlled by political whim, which means that current mechanisms are more than amply capable to make any leap hour adjustments when such adjustments are felt to be necessary. Can't we declare civil time a non-issue for now? I think both UTC and TAI are significantly more-than-adequate for its meager requirements (by precision time metrics). I know that plain TAI is not acceptable to the astronomical community, who is an important consumer of precision time, and that UTC (because it is an approximation of UT1) is acceptable, but that's about it. Is there another community of time consumers for which UTC is better than TAI? (Okay, astronavigators, but unless one of them speaks up about special needs I'll pretend like they are a subgroup of the astronomical community.) Is there a more useful time reference than UT1 for the astronomical community? It seems like no one who is able to answer these questions of mine ever feels like they are worthy of an answer (or maybe I have just been too subtle in my asking), but I think they are vitally important to the question of whether (and if so how) our time standard should be changed. Yes, I am an ignorant fool, but I can learn. Enlighten me. --Ken Pizzini
FITS and the crafting of standards
On Tue 2003-01-28T16:31:03 -0700, Rob Seaman hath writ: oscillatory modes.) Just one more example (among many) is my long time participation in the FITS standards process. FITS is astronomy's universal data format, whose metadata standards rely explicitly on UTC. For the sake of further seeding the discussion of standard this explicit reliance deserves exposition. The FITS standard was re-crafted hurriedly in 1997 because an astute fellow noted that the original definition from 1980 was so short-sighted that it would break in Y2K. The resulting standards document contains the following: The value of the DATE keyword shall always be expressed in UTC when in this [the new Y2K-clean] format, for all data sets created on earth. The purpose of the DATE keyword is to indicate the creation date/time of the FITS file. The trailing clause about earth is an explicitly inserted safety valve due to my concerns that the required timescale should be accessible to the writer of a FITS file. The presumption was that UTC will be readily available on earth. For FITS file authors on spacecraft or elsewhere the standard tacitly admits that UTC may be neither available nor relevant, and that the timescale which might be used in such a case is beyond the scope of the document. In actual practice this requirement is not always met because the creation date of the file is usually determined by a call to the operating system. The system clock may not know absolute time even as well as my wristwatch does. Even if the system has access to NTP, the letter of the standard is going to be violated near the insertion of leap seconds. Very few systems have access to any actual form of UTC. On the other hand, the loss to posterity of not knowing when a file was created to within a second is rarely important, so science does not suffer much from this violation of the standard. Regarding the DATE-OBS keyword the standard says: When the [new Y2K-clean] format with a four-digit year is used, the default interpretations for time shall be UTC for dates beginning 1972-01-01 and UT before. Other date and time scales are permissible. The DATE-OBS keyword is intended to record the date/time of the observation. The language used here is much looser. The standard gives the default interpretation, but does not give any means for determining whether or not this default is in use. This was intended to serve as a guide that FITS file authors should use a timescale which has been most available throughout the history of most astronomical observations. This was presumed to be UT (and not necessarily UTC), but any other time scale is acceptable if the mission requirements demand it. The FITS standard has an appendix about time scales that goes into further details about suggested practices and relativistic gotchas. The content of the appendix is not binding. During the drafting there were voices requesting that the new standard should require the use of TAI rather than UT because (as noted above with NTP) TAI is less ambiguous than UTC. This was struck down for several reasons. One is that there are many astronomical observations from before the advent of TAI (and there are ongoing efforts to digitize old emulsion). Another reason is that TAI is simply not as available as UT, and the standard cannot place such a steep requirement on all the astronomical data acquisition systems in the world. Another reason is that many sorts of observations are reduced as if the observation occurred from a point other than the earth's surface, and TAI does not make sense for observers at different relative velocities and different depths in gravitational potentials. The point of all this is that the specification of UTC in FITS is for pragmatic purposes and for guiding usage toward the most precise meaning that is possible for a document that holds no power over pre-existing implementations. Several aspects were left intentionally vague in the anticipation that social trends in time scale usage would evolve. Hopefully the time and frequency community who initiated this forum are attempting to find a similarly sensitive way to handle the leapseconds issues. -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 [EMAIL PROTECTED] Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E49 89 0E FE 26 B4 14 93
Re: FITS and the crafting of standards
Can I muddy the waters with some facts/evidence I have collected recently? (if your answer is no - then hit 'delete' now ;-) First where do I fit in this debate. I am the systems manager for the Astronomy group at St.Andrews University. We run a small observatory with 4 telescopes, a couple of Meade LX200s, a 1980s built scope on a ~1900 Grub Parsons mount (where setting is done with large metal rings with painted scales - too new for Victorian engraved brass, too old for digital encoders, very nice to use) and a 1m class Schmidt-cass built on-site in the 1960s. When this discussion first came up I was about to re-re-write my code for the digital setting circles that have been added to the 1m telescope, and this needs a time feed - UTC does nicely but if it going to change I want to know. Over the past year or two I have given a talk entitled What time is it? to local Astronomical Societies. This is a quick romp from sundials giving 12 'hours' sunrise to sunset to TAI, UTC, TDT, TDB, LST, etc. with a bit about the moon slowing us down. As part of this I do a check on the accuracy of the timekeeping of the audience. The full range for the audience is usually plus/minus 20secs with ~25% in the +/-5 secs range. I had one gentleman who was happily 2 mins slow - good enough for catching an Edinburgh bus, he claimed! Thus we have a very rough measure of the accuracy of the interested man-in-the-street timekeeping. Yesterday morning over a cup of coffee I floated the question of leapseconds and their abolition passed a couple of friends in the University IT Services dept. One quickly decided that leapseconds were the obvious solution, then realised we have them, and wondered what the problem was. The other, thought for a bit, then decided that decoupling time from from the rotation of the earth was a bit more philosophical than technical and was the sort of world cultural heritage thing that should not be tinkered with. A slight worry I have is what the popular media would make of it if they decided that scientists were going to mess about with time. Yet more anti-science in the media :( Sorry nothing posative in this - but then users of clock-on-the-wall time are always going to be a problem. Roger Roger Stapleton [EMAIL PROTECTED] University of St.Andrews, School of Physics Astronomy North Haugh, St.Andrews, Fife. KY16 9SS Phone 01334-463141 Fax 01334-463104
Re: Leap-seconds, the epsilon perspective
Steve Allen scripsit: Which is more important... for civil time to be counted in SI seconds? for civil time to track the rotation of earth smoothly? IMHO the former. Mark's alternative resembles the civil time solution adopted by the martian colonists in Kim Stanley Robinson's Red/Green/Blue Mars trilogy, where the clocks tick SI seconds, but every day at midnight they stop for 39.5 minutes of slip time to let the planet catch up. I always thought that was silly. How do you keep experiments running (not necessarily precision-time ones, just ordinary ones) while your clock is saying midnight, midnight, midnight, midnight, And what about Martian time zones? If we have decided that SI seconds are to be used, which is more important... for civil time to keep noon from drifting? for civil time to increase constantly? The later. Non-monotonic civil time would be a disaster, and we are very lucky in the current regime that rotation is slowing and not speeding up. Partly because eventually the scheme of turning UTC into a constant offset form of TAI requires a leap hour lest our descendants find themselves having their midday meal at 18:00 local civil time. This sort of pushing the problem off onto our descendants implies that we really don't have the right solution, just one that salves some needs now. I think it's utopian to suppose that a right solution exists; we are trying to reconcile the fundamentally irreconcilable. Eventually (a very long time from now) we *will* have to abandon the 86400 seconds = 1 day assumption. (I utterly reject any attempt to redefine the SI second.) I suspect this would account for 99.9% of the world's clocks, including the clocks inside most computers, VCRs and microwave ovens; on your wrist; or next to your bed. Hmm. How reasonable is it to expect this to change in future? If the future is a world of ubiquitous networking where Bill Gates can make every wristwatch and refrigerator magnet into a .NET client, then we should very much expect this to change. Not what I meant. I meant, how reasonable is it to expect to find 10^-8 reliable clocks in ordinary hands in the future? If every clock is indeed networked (a very unlikely future, I'd say), then of course the time scale can be arbitrarily futzed with and all clocks will stay in sync. -- John Cowanhttp://www.ccil.org/~cowan [EMAIL PROTECTED] Please leave your values| Check your assumptions. In fact, at the front desk. | check your assumptions at the door. --sign in Paris hotel |--Cordelia Vorkosigan
Re: What problems do leap seconds *really* create?
John Cowan wrote on 2003-01-29 17:56 UTC: The problem is that they are not announced much in advance, and one needs to keep a list of them back to 1972 which grows quadratically in size. Is this a real problem? Who really needs to maintain a full list of leap seconds and for what application exactly? Who needs to know about a leap second more than half a year in advance but has no access to a time signal broadcasting service (the better ones of which all carry leap second announcement information today)? For pretty much any leapsecond-aware time-critical application that I can think of, it seems more than sufficient to know: - the nearest leap second to now - TAI-UTC now - UT1-UTC now This information is trivial to broadcast in a fixed-width data format. (For the nitpicker: The number of bits to represent TAI-UTC admittendly grows logarithmically as be move away from 1950. We know we can live with that, as O(log(t)) is equivalent to O(1) for engineering purposes.) Markus -- Markus Kuhn, Computer Lab, Univ of Cambridge, GB http://www.cl.cam.ac.uk/~mgk25/ | __oo_O..O_oo__
Re: What problems do leap seconds *really* create?
William Thompson scripsit: Any application which seeks to calculate the difference in time between two events recorded in UTC time needs to know if there are any leap seconds between the start and stop time. For example, suppose you were studying solar flares, and analyzing some data taken in 1998, and you saw a burst of hard X-rays at 23:59:53 UT on Dec 31, followed by a rise in EUV emission at 00:00:10 UT the next day. You'd think that the delay time between the two would be 17 seconds, but it's really 18 seconds because of the leap second introduced that day. Thanks for the example. Of course it is not astronomy-specific: the same thing applies if you are calculating how long somebody spoke for in field linguistics, or the amount of time it takes a moving part to stop moving in engineering. What we are dealing with here is time-zone independent civil time. That's a vital difference for the scientific analysis of the data. Indeed. And yes, part of that software package includes a list of all leapseconds added since 1 Jan 1972. Currently, my software doesn't handle TAI/UTC conversions between 1958 and 1972, when UTC seconds had varying lengths. Modern Unix time packages (both GNU and ADO) assume that TAI-UTC was 10 from the epoch until 1972-06-30T23:59:60 UTC. Or to put it another way, the epoch was at 1970-01-01T00:00:10 TAI. When did the TAI timescale first come into existence? One answer seems to be that TAI was born on 1958-01-01T00:00:00 UT2, which was also 1958-01-01T00:00:00 TAI. But OTOH the definition of the SI second changed in 1967 and again in 1997. What did these changes do to the uniformity of TAI? I found the following interesting statement at http://www.maa.mhn.de/Scholar/times.html : # The need for leap seconds is not caused by the secular slowdown # of Earth's rotation (which is less than 2 milliseconds per century) # but by irregular variations in this rotation and by the fact that the # definition of the SI-second is fixed on the duration of the year 1900 # which was shorter than average. -- Not to perambulate || John Cowan [EMAIL PROTECTED] the corridors || http://www.reutershealth.com during the hours of repose || http://www.ccil.org/~cowan in the boots of ascension. \\ Sign in Austrian ski-resort hotel
Re: What problems do leap seconds *really* create?
Those of us with a day job may be having a hard time keeping up with the messages as they arrive fast and furious :-) # The need for leap seconds is not caused by the secular slowdown # of Earth's rotation (which is less than 2 milliseconds per century) # but by irregular variations in this rotation and by the fact that the # definition of the SI-second is fixed on the duration of the year 1900 # which was shorter than average. Basically we don't have leap seconds because the Earth's rotation is slowing down (by transfering angular momentum to the Moon). Rather, we have leap seconds because the Earth has *already* slowed down since 1900. See the rather consistent slope of about 7 seconds per decade on the plot of UT1-UTC (with leap seconds removed) versus date: ftp://gemini.tuc.noao.edu/pub/seaman/leap/noleap.pdf The current dynamical effects are the subtle wiggles imposed on this trend. This is actually a fairly useful bias since it guarantees (short of asteroid impact or armageddon) that there will be no negative leap seconds. Again, please note how consistent the divergence is between atomic and Earth timescales over decade long periods. Where precisely is the urgency to adopt a quick fix? Ken Pizzini says: I realize that the astronomical community has evolved to a consensus that UT1 (approximated by UTC) is a highly useful way to mark time, Rather we've evolved a consensus that different problems require different systems of time - not surprising, since we invented most of them. with the additional feature that it is usable as a civil time standard, It isn't just usable - it is preferable to many alternatives. but there is so much of that evolution which is based on historical accident rather than purely technical requirements Historical accident makes it sound like the practice of timekeeping was some afterthought to events. The reality is that timekeeping has often been central to other, bloodier, battles - from Augustus Caesar appropriating an extra day from February into his month - to Harrison's chronometer that was instrumental to the building of a later empire. Is there nobody on this list who was present at the birth of UTC? It is a good, solid, pragmatic standard. The ability to issue leap seconds monthly clearly represents a recognition that these would be needed to preserve the standard over hundreds or thousands of years. that I find it hard to believe that there would be no possible way to improve upon it, even after non-astronomical constraints are factored in, if only it were possible to start anew with a clean slate. The astronomical community isn't afraid to discuss a successor to UTC. The mere presence here of several vocal members of that community should demonstrate this. There is always room for improvement - although the elegant simplicity of UTC will be hard to rival. What we object to (if my friends don't mind my saying so) is not the idea of *improving* UTC. What we object to is this current process which appears to be an attempt to discard the standard entirely - and to do so with minimal consent. Please! Let's talk about ways to improve UTC and civil timekeeping. And let's take the appropriate amount of time to reach a decision - say - 40 or 50 years. In the mean time, let's pay attention to the real question, which is how to build an infrastructure that will dramatically improve the dissemination of all time signals. NTP is a lovely mechanism. Astronomers are likely among its most passionate users. The limitations of NTP or of any other general purpose mechanism for disseminating time signals should not limit the definition of the standards behind those signals. Rather, NTP, WWV, GPS - and heavens to Betsy, GLONASS - should be built in a fashion that can handle a general parametrized time standard. This should include a static offset as used by GPS, as well as frequently or infrequently introduced time jumps of variable sizes in either direction as required by daylight saving or leap seconds. These systems should perhaps be parametrized to support epsilon schemes as described by Calabretta or by Kuhn's UTS. And a general purpose time distribution mechanism should support differing rates as required by sidereal time, for example. Similarly, personnel associated with various projects are expected to know enough engineering to build incredibly complicated radio equipment and other devices. Why aren't these same projects expected to handle time issues with similar professionalism? If they need unsegmented time - they should use some variation of TAI - and if they don't require an Earth fixed time scale - they shouldn't use UTC. And if they do use UTC, well then, they should figure out how to handle leap seconds. Leap seconds are discussed very prominently in a very short document. Rob Seaman National Optical Astronomy Observatory
Re: What problems do leap seconds *really* create?
On Wed 2003-01-29T15:43:24 -0700, Rob Seaman hath writ: Please! Let's talk about ways to improve UTC and civil timekeeping. And let's take the appropriate amount of time to reach a decision - say - 40 or 50 years. In the mean time, let's pay attention to the real question, which is how to build an infrastructure that will dramatically improve the dissemination of all time signals. Alas, we do not have 40 years, we have less than 35. The end of 32-bit Unix time is 2038-01-19T03:14:07. Well before that the Unixes of the future must have decided on the proper way to implement the algorithms for handling 64-bit time_t when receiving inputs from the NTP(s) of the future. Although there are many technical aspects to the situation, for practical purposes any change of UTC is legislation. Effective legislation seeks the common good while remaining congruent with the will of the people. When that will is split because of pre-existing practices and notions of what is good, the process inevitably becomes political. If this change is going to affect civil time, then, politically speaking, the 32-bit end of time is a looming deadline that should serve to motivate an answer about the fate of UTC within the next 10 years. In the mean time we may learn that a fifth fundamental force has implications for the spacetime metric that invalidate all the current time scales in use by astrophysics. As before, the response to that kind of paradigm shift in physical thinking would trigger the creation of yet more time scales to be used by astronomers. The old time scales would remain, unmodified, and less used. On Mon 2003-01-27T17:32:19 -0500, John Cowan hath writ: I would have no problem with deciding now to change UTC, effective in 2033. Over that interval all the observatories in the world could assuredly handle any change. But in the context of the history of time scales, changing the character of UTC would still be the wrong thing to do. -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 [EMAIL PROTECTED] Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E49 89 0E FE 26 B4 14 93
Re: What problems do leap seconds *really* create?
On Wed 2003/01/29 15:43:24 PDT, Rob Seaman wrote in a message to: [EMAIL PROTECTED] Basically we don't have leap seconds because the Earth's rotation is slowing down (by transfering angular momentum to the Moon). Rather, we have leap seconds because the Earth has *already* slowed down since 1900. See the rather consistent slope of about 7 seconds per decade on the plot of UT1-UTC (with leap seconds removed) versus date: ftp://gemini.tuc.noao.edu/pub/seaman/leap/noleap.pdf The current dynamical effects are the subtle wiggles imposed on this trend. This is actually a fairly useful bias since it guarantees (short of asteroid impact or armageddon) that there will be no negative leap seconds. Again, please note how consistent the divergence is between atomic and Earth timescales over decade long periods. Where precisely is the urgency to adopt a quick fix? I agree with you that there is plenty of time to make an informed decision, that nothing need be done on a timescale of decades, and also that the process to date appears, at least to some of us, to have bordered on Machiavellian, though I'm sure it was not. It's also true that the leap-seconds we have now do come from the slowdown since 1900. However, your consistent slope of about 7 seconds per decade obscures the basic point about the long term future of UTC. Your graph shows a linear approximation to what is actually a parabola. The graph in the GPS World article shows the long term trend much better. Though its fit to a parabola is not much more convincing - we do know that it is a parabola because the physics of the Earth-Moon dynamical system tells us so. The wiggles are caused by the motion of dense material in the Earth's mantle which cause the Earth's moment of inertia to vary unpredictably, thus causing it to spin up as well as down. They are what leap-seconds were originally designed to handle, not the predictable, long-term secular deceleration of the Earth's rotation. Thus, what is 7 seconds per decade now will become 35 seconds per decade in another two hundred years time. When you add up all those so-many-seconds per decade over the next 20 decades the cumulative error runs to many minutes - already nearly 60s by 2050 according to the GPS World article, not 35s by a linear extrapolation, and more than 140s by 2100, double the linear-extrapolated value. The cumulative error grows quadratically; that is the problem. Mark Calabretta ATNF