This document is archival.
The information here is from year 2006 and before.
Please see this page for more up-to-date information.


Consequences for Observatories of Discontinuing Leap Seconds

The tone in following descriptions is perhaps too dire, but it is important that observatories determine how long they will need to implement changes.

The impending demise of leap seconds?

There is strong evidence that an effort will be made to discontinue adding leap seconds to civil time. Although it is relatively clear that this change cannot be implemented immediately, it nevertheless appears that a discontinuation of leap seconds may be announced during 2003. It is supposed that the effective date for the change would be 5 to 10 years from now, and after that the broadcasts that have long defined civil time will probably not have leap seconds.

It is not yet clear how the new form of civil time will be realized. Those who favor the change hope to redefine UTC such that it will never have another leap second, but if they cannot modify UTC they may create a new instance of atomic time and broadcast that as civil time.

In either case the effect on many telescope pointing systems is the same, for most existing time hardware and software is designed to make use of various forms of civil time distribution. For the purposes of simplicity in the survey it will be supposed that the definition of UTC will be changed.

When will the details of the change become clearer?

The possibility of discontinuing leap seconds has been discussed for more than three years. I have a bibliographic web page which gathers together as many references as I can to the process and the parties.

In various forums much time and effort has been spent detailing the advantages, disadvantages, and rationale of discontinuing leap seconds. The principal result of these discussions has been to make it clear that there is a serious possibility that civil time will cease to have leap seconds.

The next meeting to discuss the future of civil time will be in Torino Italy in 2003 May. A decision is expected some time after that. Unconfirmed reports indicate that the most likely change would be to discontinue the (currently nearly-annual) leap seconds in favor of leap hours scheduled decades in advance. Even though such a change would impose significant costs on astronomers, there have been indications that some would prefer to discontinue leap seconds immediately.

Under these circumstances it seems more than prudent for the astronomical community to present an accounting of the costs as an argument against premature discontinuation of leap seconds. There are two important components of this account: the dollar amount required for remedies, and the time required in order to implement remedies.

The effect on existing telescope systems

DUT1 is defined as ( UT1 - UTC ). It gives the difference between sources of civil time and the diurnal rotation angle of the earth. The current practice is to insert leap seconds into civil time so as to keep the absolute value of DUT1 less than 0.9 s. The effect of discontinuing leap seconds is to cause DUT1 to grow without bound.

A DUT1 of 0.9 s is small enough that it is not relevant to many telescopes that predate the digital age. For such telescopes it is often routine to "reset the setting circles" manually after centering on a catalog object. For many such telescopes a change in UTC would require new procedures for telescope operating personnel.

A DUT1 of 0.9 s is large enough that it is important to telescopes that perform automated target acquisition. The difference translates to 13.5 arcseconds of RA on the equator. For many such telescopes a change in UTC would require new hardware and/or software.

Problems and remedies

The current radio broadcasts of civil time typically provide both UTC and DUT1. If leap seconds are discontinued then hardware which is currently able to extract DUT1 will become obsolete.

Signals from GPS satellites contain enough information to determine UT1, but few if any receivers currently provide UT1, and there are no protocols for its dissemination.

Values of DUT1 will undoubtedly remain available from the IERS over the Internet, but the current forms require text files to be retrieved at regular intervals and then interpreted.

NTP will continue to operate to disseminate UTC over the Internet. At the Torino meeting it is supposed that new forms of NTP-like protocols will be proposed. These protocols might have multiple "channels" which might make it possible to synchronize a computer to UTC, TAI, and maybe even UT1.

Counting the cost of implementing remedies

There are several components of the cost which are likely to be common to many observatories:

For the one observatory which has furnished an offhand estimate the total cost for implementing and testing remedies was given as US$100000.

Counting the time required to implement remedies

If the decision to discontinue leap seconds from civil time were effective immediately, then astronomers would have to scramble to implement interim remedies while looking for better long-term solutions.

If the decision to discontinue leap seconds were announced years in advance, then the precision timekeeping industry should be able to create new products which provide all of the necessary earth orientation information.

If the decision to discontinue leap seconds were announced decades in advance, then observatories should be able to incorporate new hardware and software into their systems as a natural part of the ongoing cycle of maintenance and upgrades.

One observatory has indicated that it would be infeasible to modify telescope systems in less than two years after the details of a plan to discontinue leap seconds is published. Note that this is not two years after the announcement of intent to discontinue leap seconds, but rather two years after details of possible hardware and software remedies become available.

A survey regarding schedule and cost

In the hope of determining how much time and effort will be required for observatories to implement remedies I have created this survey. The results will be sent to the meeting in Torino in May.
Steve Allen <sla@ucolick.org>
$Id: consequences.html,v 1.8 2013/04/08 16:48:27 sla Exp $