Nº 7 2013 > The future of time
To abolish or not to abolish the leap second?
Global navigation satellite systems and their system times
W. Lewandowski, International Bureau of Weights and Measures
The International Bureau of Weights and Measures (Bureau International des Poids et Mesures — BIPM) is in charge of computing the international reference time-scale known as Coordinated Universal Time (UTC). It is derived from a uniform and continuous time-scale, called International Atomic Time (Temps Atomic International — TAI), by applying a correction of an integral number of seconds. UTC is the sole reference time-scale for coordinating the world’s time. It serves as the basis of legal time in many countries.
Global navigation satellite systems (GNSS) rely on precise time to enable accurate ranging measurements for positioning, which in turn requires consistent intra-system synchronization. For this purpose GNSS use the following internal reference time-scales, constructed from clock ensembles: GPS Time, GLONASS Time, Galileo System Time (GST) and BeiDou System Time.
These system times are pseudo time-scales and should be regarded as being merely internal GNSS technical parameters and not as time-scales to be used as a reference for other human activities.
Usually, system times are steered to an external stable reference time-scale. For example, GPS Time follows UTC(USNO) modulo one second via its local representation at the United States Naval Observatory. But UTC is a stepped time-scale because of its discontinuity resulting from the use of leap seconds. In particular for the purposes of safety of life services, some providers of global navigation satellite system services have preferred to adopt alternative continuous (unstepped) time-scales. This is causing difficulties for designers of global navigation satellite systems because there is no ideal way of choosing a reference epoch for numbering the seconds of alternative continuous time-scales.
The various approaches chosen by providers of global navigation satellite system services, and the relationship between these system times and UTC can be seen in Figure 1.
GPS Time is continuous and is not adjusted for leap seconds. It was set on 6 January 1980, at 00:00 UTC to have zero seconds difference from UTC. GPS Time is 19 seconds behind TAI, and — because of the leap seconds added to UTC — is now (in 2013) 16 seconds ahead of UTC.
GLONASS Time, unlike GPS Time, follows UTC seconds and thus is not a continuous time-scale. Galileo System Time (GST) is continuous and has the same initial epoch as GPS Time. In the early stages of defining the Galileo system, a preliminary decision was taken that GST would use TAI as reference. But bearing in mind that TAI is not intended for general dissemination, the designers of the Galileo system considered that setting the internal time-scale of Galileo to TAI would cause confusion. The final decision was to set to zero the second difference between GST and GPS Time.
BeiDou System Time is continuous and was set on 1 January 2006 at 00:00 UTC to have zero seconds difference from UTC. Thus BeiDou System Time is 33 seconds behind TAI, and is now (in 2013) 2 seconds ahead of UTC.
Because UTC is a stepped time-scale, the continuous internal time-scales of global navigation satellite systems become alternative time-scales for some applications. For example, the International GNSS Service (IGS) uses GPS Time for tagging some of its products.
The use of these continuous internal time-scales of global navigation satellite systems is leading to confusion among users, because the various scales differ by tens of seconds. Galileo provides an example of the potential for confusion. Some parts of the Galileo system are tagged to GST, while other parts are tagged to UTC. The greatest difficulty occurs crossing 00:00 (midnight), when for a period of 16 seconds various parts of the system refer to two different days. This may lead to major mistakes.
Although the internal time-scales of global navigation satellite systems do not need to be synchronized to the international standard UTC to meet the needs of navigation, there would be an obvious benefit in international coordination to simplify the operation of these systems and allow for their interoperability. This is reflected in the recommendations of the Consultative Committee for Time and Frequency (Recommendation S6-1999) and of the International Committee for Weights and Measures (Recommendation 1 CI -1999). It is also one of the tasks of the International Committee on Global Navigation Satellite Systems.
Today, the global navigation satellite systems represent by far the most common means of obtaining precise UTC. The GPS and GLONASS service providers disseminate corrections to their internal system times to obtain predictions of UTC as maintained at the United States Naval Observatory (UTC(USNO)) and the national time-scale of the Russian Federation (UTC(SU)), respectively. Galileo will also broadcast a physical realization of UTC, as most likely will other systems too. GPS currently broadcasts a prediction of UTC(USNO) which agrees to within a few nanoseconds with actual UTC(USNO), and UTC(USNO) agrees to within a few nanoseconds with actual UTC. This means that GPS broadcasts a prediction of UTC worldwide with an uncertainty of several nanoseconds. At present, GLONASS predictions have an uncertainty of hundreds of nanoseconds, but their accuracy is likely to be improved in the near future through appropriate calibrations.
Time to leave leap seconds behind?
Leap seconds cause difficulties to modern infrastructure, in particular to global navigation satellite systems. Also, celestial maritime navigation can now do without leap seconds, so that argument for keeping leap seconds is no longer valid.
Within ITU’s Radiocommunication Sector (ITU–R), consideration is being given to revising the definition of “Coordinated Universal Time”. ITU–R is also working on a recommendation that may lead to a new continuous time-scale.