In spaceflight keeping track of time between systems widely separated in space (e.g. people on Earth vs spacecraft in deep space) and/or time, moving at substantial relative velocities and in different gravitational potentials is a challenge.

The documentation for JPL's Horizons at https://ssd.jpl.nasa.gov/?horizons_doc includes the following discussion of timescales.

The three time systems are described as follows:


("Barycentric Dynamical Time"); typically for cartesian, osculating element, and close-approach tables. The uniform time scale and independent variable of the planetary ephemeris dynamical equations of motion.


("Terrestrial (Dynamic) Time"), called TDT prior to 1991, used for observer quantity tables. This is proper time as measured by an Earth-bound observer and is directly related to atomic time, TAI. TT periodically differs from TDB by, at most, 0.002 seconds.


is Universal Time This can mean one of two non-uniform time-scales based on the rotation of the Earth. For this program, prior to 1962, UT means UT1. After 1962, UT means UTC or "Coordinated Universal Time". Future UTC leap-seconds are not known yet, so the closest known leap-second correction is used over future time-spans.

Different time scales are managed differently, and some are updated and/or corrected relative to others from time to time (pun intended).

Question: When and for what purpose is TAI updated? (not about a one-second error)

  • 1
    $\begingroup$ @DrSheldon It seems user37920 is worried that if they add the context they will be shot down for asking multiple questions in one post. Here is the context they provided: "what [do] the [astronauts] feel and experience when TAI is updated? And what is the minimum or maximum TAI allowed to be updated?" $\endgroup$
    – called2voyage
    Commented Oct 4, 2020 at 4:32
  • 6
    $\begingroup$ I’m voting to close this question because it is off-topic. International Atomic Time (TAI) is a physics subject, and runs continuously without corrections. Universal Coordinated Time (UTC) is an astronomy subject, and corrects for the motion of the Earth. Both systems are maintained from the Earth's surface, without the involvement of satellites, astronauts, or space agencies. $\endgroup$
    – DrSheldon
    Commented Oct 4, 2020 at 5:23
  • 1
    $\begingroup$ @DrSheldon - An accurate concept of time is essential for accurately modeling the motion of a satellite orbiting the Earth or of a probe sent from the Earth to somewhere else in the solar system. Accurate timekeeping is essential for space exploration. I'll vote to reopen if this question is closed as off-topic. $\endgroup$ Commented Oct 4, 2020 at 6:15
  • 1
    $\begingroup$ @DrSheldon That this question is on-topic at physics.SE, astronomy.SE, and possibly others, does not mean that it is off-topic here. There are overlaps between different elements of the Stack Exchange network, and that is okay. $\endgroup$ Commented Oct 4, 2020 at 6:46
  • 3
    $\begingroup$ @uhoh: Completely irrelevant argument. There are questions here about food, yet that does not make every question about food on-topic. TAI (and UTC) would still exist even if rockets and satellites were never invented. The subject is off-topic. $\endgroup$
    – DrSheldon
    Commented Oct 4, 2020 at 13:40

1 Answer 1


TAI conceptually is time measured by a perfect atomic clock running exactly at the geoid. There are some issues with this concept:

  • A perfect atomic clock does not exist.
  • Older and presumably less accurate atomic clocks are regularly replaced with newer and presumably more accurate atomic clocks.
  • Few, if any, atomic clocks are at sea level.
  • Mean sea level is close to but not the same as the geoid.
  • The geoid is not perfectly known.

This means that the averaging and scaling used to combine atomic clock outputs to generate TAI needs to be updated as older, less accurate atomic clocks are replaced with newer, more accurate atomic clocks, and as knowledge of the geoid improves.

That said, errors made in the TAI timestamps of past events due to imperfect clocks remain errors forever. TAI is not updated retroactively.

  • 2
    $\begingroup$ What is the rationale for the downvote? $\endgroup$ Commented Oct 4, 2020 at 15:40
  • 2
    $\begingroup$ I wrote a similar question on a post of my own once, and somebody replied "This is Stack Exchange, there's always a down vote." Of course here in space perhaps it was a cosmic ray-induced bit flip. :-) $\endgroup$
    – uhoh
    Commented Oct 4, 2020 at 22:57
  • $\begingroup$ @DavidHammen I semi-sure TAI is never amended after the event (can't find a definitive statement on the BIPM site but Steve Allen's timescales page which is trustworthy says so). The clock weights and frequency of TAI are measured and adjusted but predict for the future. There is a post-processed version called TT(BIPM) which appears later which is a "better TAI" and changes to TAI's corrections have happened before $\endgroup$ Commented Oct 5, 2020 at 19:25
  • $\begingroup$ @astrosnapper While TAI is not amended after the fact, how the scale factors and weights by which the different atomic clocks that collectively form TAI are updated for current and future determination of TAI. $\endgroup$ Commented Oct 5, 2020 at 21:43
  • 1
    $\begingroup$ Agreed, and this is probably pedantry, but the last sentence of the answer could be read as past TAI times can be amended retroactively. This doesn't happen, even when problems become known such as the lack of gravitational redshift (time slowing with altitude/distance from the geoid) correction before 1977. Also note for OP/others: TAI is not distributed (only UTC is) and "true" TAI and UTC are only available after 10-40 days following the comparisons, weight adjustment and publication of Circular T. Time is weird... $\endgroup$ Commented Oct 5, 2020 at 22:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.