To better understand the UTC(NIST) time scale, it is helpful to look at its similarities and differences with respect to UTC. The UTC time scale is maintained and distributed by the Bureau International des Poids et Mesures (BIPM), or in English, the International Bureau of Weights and Measures, which is located near Paris, France in an area considered to be international territory. Therefore, the BIPM maintains neutrality as an intergovernmental organization that is not controlled by any country. Its mandate is to provide a single, coherent system of measurements based on an International System of Units (SI).
The SI is built upon seven base units of measurement. One of the base units is the second, which serves as the standard unit for time interval and the foundation for world timekeeping. Since 1967, the SI second has been an “atomic” second. It’s defined as “The duration of 9,192,631,770 periods of the radiation corresponding to the transition between two hyperfine levels of the ground state of the cesium-133 atom”, and its duration is precisely determined by cesium atomic clocks. This leads us to the first similarities shared by the UTC and UTC(NIST) time scales, both are based on the SI definition of the second and both depend upon atomic clocks to keep time.
Another similarity between UTC and UTC(NIST) is that both keep time not with just one atomic clock, but rather by using an “ensemble” of atomic clocks. You can think of an ensemble as a group of items that is viewed collectively rather than individually. The time produced by the UTC and UTC(NIST) time scales ensembles is obtained by averaging the time produced by individual clocks. A weighted average, rather than a simple arithmetic average, is used. This means that some clocks – specifically the clocks with the best frequency stability – are given more “weight” and contribute more to the average than others.
Some key differences between the two time scales relate to the respective size of the two ensembles and to where the clocks are located. UTC is a global time scale with a clock ensemble much larger than the NIST ensemble. The UTC ensemble is spread around the world; with more than 400 clocks residing in more than 80 laboratories in more than 60 nations. In contrast, UTC(NIST) is a local time scale, with an ensemble typically consisting of 10 to 15 clocks, all of which reside in the NIST laboratories in Boulder, Colorado. Each NIST clock is also a member of the UTC ensemble, so UTC(NIST) can be thought of as a subset of UTC.
The most important distinction between UTC and UTC(NIST), however, is that UTC is a virtual or “paper” time scale that does not produce any physical signals, and that UTC(NIST) does produce physical signals. The signals produced by UTC(NIST) are in the form of electrical pulses or sine waves, and these signals are distributed to users and then used to synchronize or calibrate other clocks.
The concept of a virtual time scale can be difficult to grasp, but perhaps becomes easier when you consider that none of the clocks in the UTC ensemble reside at the BIPM and thus the BIPM doesn’t measure any physical clock signals. Instead, the BIPM calculates UTC from measurement data that is sent to them by the more than 80 laboratories mentioned earlier. Several different methods are available for measuring and collecting clock data to send to the BIPM. For example, NIST submits its data to the BIPM using the satellite technique described here.
Once a month, after all clock measurements have been collected and after all calculations are complete, the BIPM publishes the results in an on-line document known as Circular T. This document lists the time difference between UTC and each of the local time scales that are subsets of UTC, such as UTC(NIST). In Circular T parlance, the local time scales are known as UTC(k), where k is a variable that designates the name of the local time scale’s laboratory, and where values for UTC – UTC(k), are reported at 5-day intervals. The publication of these values is extremely important for ensuring that UTC is indeed “coordinated” around the world, because it verifies the accuracy of each UTC(k) time scale with respect to UTC. By doing so it allows each local UTC(k) time scale to establish metrological traceability to both UTC and to the SI second.
To make data available more frequently than once per month, the BIPM also publishes an unofficial time scale called Rapid UTC, or UTCr. Rapid UTC is published once per week, with values of UTCr – UTC(k) reported at 1-day intervals. Currently (November 2020), NIST and more than 60 other laboratories participate in UTCr.
Although the values computed by UTC and UTCr are not known until after the fact, UTC(NIST) is continuously adjusted using forward prediction techniques that anticipate where UTC will be when the Circular T is published. As a result, UTC(NIST) is an excellent approximation of UTC that has the advantage of producing physical signals in real time, and that typically differs from UTC by less than three billionths of a second (3 ns). You can view recent UTC – UTC(NIST) time differences. You can also visit the BIPM Circular T archive.