In April of 2024, the US White House directed NASA to define a workable definition and operational plan for LTC: Coordinated Lunar Time.

White House directs NASA to create time standard for the moon | Reuters

One motivation is:

for a person on the moon, an Earth-based clock would appear to lose on average 58.7 microseconds per Earth-day and come with other periodic variations that would further drift moon time from Earth time.

This leads to significant challenges for navigation, spaceship docking etc, especially when these things happen in regions which cannot connect to GPS satellites etc to get the necessary time accuracy.

The associated memo - Celestial-Time-Standardization-Policy.pdf notes that:

Each new time standard developed will include the following features:

  1. Traceability to Coordinated Universal Time (UTC);
  2. Accuracy sufficient to support precision navigation and science;
  3. Resilience to loss of contact with Earth; and
  4. Scalability to space environments beyond the Earth-Moon system

Failing to account for the discrepancy between a transmitter clock on the Earth and how it is perceived by a receiver on the Moon will result in a ranging error. Precision applications such as spacecraft docking or landing will require greater accuracy than current methods allow.

How might this standard might be defined and used, and what are the main challenges?

Would it likely involve GPS-like satellites in lunar orbit, to provide service on the far side and poles?

I'm guessing it will involve defining a new epoch. I wonder when that will be, and why?


1 Answer 1


At a technical level, there isn't any real problem. The main problem is figuring out why you want this and what properties it's supposed to have.

Requirement (1) from the memo is almost trivial. Any new time standard is obviously going to be based on time kept from atomic clocks. That all but implies the current TAI standard, which the report correctly notes is in turn related to UTC by an integer offset. Easy.

Requirement (2) also follows trivially from the use of atomic clocks, and you can solve requirement (3) by maintaining local atomic clocks. There are some details about how you re-sync clocks to deal with periodic communications loss, and how you handle speed-of-light delays with relativistic corrections and such, but as complicated as that sounds, GPS satellites have already been doing exactly that since the 1970s. It's a solved problem, though we'll return to re-syncing a bit more below.

Requirement (4) is also solved; astronomers and spaceflight mission planners already must consider this, and there are different relativistic frames that have already been defined. TDB is the one returned by Horizons, for example, and so is very commonly used. TCB is the one that makes more sense on paper, and there's also TT (which TAI and UTC can be considered as measuring) and TCG. You can freely convert among any and all of these. There's Python code to do it, even.

That is, the obvious design works: you broadcast TAI to the moon, mars, wherever, and, some undergrad-level math later, they know what time it is on Earth. Maintain that time in a local atomic clock, convert to local time any way your colony pleases.

I'm frankly not sure why the government felt the need to make a report telling us to think that up. An astronaut distributing time from the Hello Kitty wristwatch they synchronized planetside before launch is not going to work long-term, but that was never on the table. It was obviously going to be atomic clocks, and prettymuch the simplest thing you can do with clocks and a broadcast signal is just fine.

Most of the challenges raised in the report appear . . . spurious? Some issues cited above, plus a few randomly spotted in a skim:

  • "Precision applications such as spacecraft docking or landing will require greater accuracy than current methods allow." This appears to just be objectively false, as evidenced by the fact that we did both in cislunar space a decade before spaceborne atomic clocks even existed.

  • "events that appear simultaneous at the Earth (e.g., the start of a broadcast signal) are not simultaneous to an observer at the Moon." Correct but irrelevant: you only need one point to broadcast from, and you can correct for any perturbations anyway, which is how we already solve related problems on Earth managing ALGOS and TAI.

  • "for a person on the moon, an Earth-based clock would appear to lose on average 58.7 microseconds per Earth-day" You can straightup ignore this for short-duration stays (read: months or less); our astronomical error is larger. For long stays (actual centuries), you can use leap seconds to correct local time (see below). For medium stays (years, decades), the computer cores get some exercise but the humans can sit tight and watch space TV in ignorant bliss.

  • "the direct use of UTC at the Moon (i.e., without correction) as the local time scale would have cascading effects for applications that require precise metrology[ e.g. throwing off the definition of the meter]." This is correct but pointless; no one in charge of timekeeping would make such an obvious mistake as blindly using untransformed UTC to count seconds. It does surely illustrate that a transform is necessary, which I hope is the point of including such a weird observation in the report. If you like, the solution is to just rescale it (and count TAI instead, please).

The primary source of error is and will be the precision of astronomical measurements, not the accuracy of the clocks or the math. It's definitely not an ignorance of the existence of General Relativity, what calculations to run, or the need to agree on common definitions per-se.

Even so, we already have highly accurate astronomical measurements in the solar-system, and to the extent the procedure results in any error, that error is judged . . . based off of those measurements as ground truth, so there's no room to complain. As technology improves, we may get ever improving estimates, as is the current case with the e.g. TT(BIPM##) series.

In the spirit of the question, though, there are a few details left to tidy—dealing with the needs of fleshy meatbags noble human explorers.

Mars's solar day is about 40 minutes longer than Earth's, which means that if you want Mars colonists to both keep diurnal activity and avoid nasty clock conversions, you need a local time with martian hours, days, etc. You maintain it with an integer offset and a different day length from Earth; the challenging part is deciding, sociologically, what scheme you want.

The situation for the Moon is similar, but probably easier. Any early permanent human presence on the moon is likely to be subterranean subselenean, in which case you can stick with Earth-centric 24-hour time, synchronized conveniently to Houston JSC mission control's Central Time (CT); the lunar days (and nights) are too long anyway. As above, the drift due to relativity is negligible over a short duration, and for long-duration could be managed with leap seconds just like UTC currently is (and forever should be, dagnabbit).

In a middle future, the owners of the many atomic clocks on Earth and the owners of the many atomic clocks on e.g. Mars might get into an asinine political struggle about whose leap seconds should correct whose. As long as exactly one side prevails, it doesn't matter at all who, from a chronometric perspective.

For the indefinite future, the scheme still works. Unless humans outrun their light cone with warp drive (narrator: they won't), you can still synchronize the whole galaxy with a TAI time broadcast. Probably you want a couple layers of repeaters in there (like we have already on Earth) to reduce triangle error, but the error in any case travels at the speed of light, which is the same speed as the broadcast, so who cares.

Perhaps the most serious issue is whether you want to define a new relativistic frame for the surface of a given celestial body. I'm not aware of any standard ones being pre-defined for particular extraterran surfaces, but it might be worth considering. You need to do the calculation either way (transform Earth's clocks to your clocks, or equivalently vice-versa); the question is whether you name it and make a standard out of it. The only reason is to coordinate surface activity and spaceflight, as otherwise there's no reason to bother (proper time is always proper time). In a mature spaceflight scenario, we might therefore name one of the conversions as the 'Mars Time' frame (TT analog) or 'Mars Orbital Time' (TCG analog), but frankly this is just a question of naming, not science.

  • $\begingroup$ Thanks - lots of interesting detail. But atomic clocks are heavy and expensive (17 kg in the example I posted in a comment above). How do they get synced with each other, without communicating with Earth? You also ignore some operational details and the use case of tiny rovers wandering around the far side of the moon, out of touch with any Earth-based time references. As I noted in the question, I'm guessing the vision here is a way for any tiny robot on the Moon to get time and position via a new lunar GPS-like satellite constellation. $\endgroup$
    – nealmcb
    Commented Apr 4 at 21:36
  • $\begingroup$ Also note that to do time transfer to a spaceship of poorly-known location, as I understand it, you need three one-way light times. So especially at Mars, Venus, etc, they'll want a local realization of time, synchronizing multiple clocks. And thus, we'll want to define carefully how to estimate past error rates among the different realizations of time. The trick is going beyond theory to cost-effective implementations for important use cases. $\endgroup$
    – nealmcb
    Commented Apr 4 at 23:06
  • $\begingroup$ Coordinating Moon time with Earth time to ~1 s precision is relatively easy. But the plan is to implement an atomic clock network on the Moon, and converting between the Moon & Earth networks at sub-nanosecond precision is not trivial, since the clock rates vary in a complicated way. Section 2.3 of The JPL Planetary and Lunar Ephemerides DE440 and DE441 describes how to convert UTC to TDB to high precision, taking into account the varying gravitational potential of the Solar System. $\endgroup$
    – PM 2Ring
    Commented Apr 5 at 0:54
  • 1
    $\begingroup$ As I said in physics.stackexchange.com/a/770976/123208 "The recent history of precision chronology has been a litany of committees that fail to reach decisions, with poor communication and misunderstanding between various involved parties [...] For a summary of the sordid details, please see A brief history of time scales, by Steve Allen of the Lick Observatory". I'm hoping that the LTC gets done right. $\endgroup$
    – PM 2Ring
    Commented Apr 5 at 0:59
  • 1
    $\begingroup$ The math is definitely involved, but it is not magic. AstroPy, cited above, implements it, SOFA implements it, even I've implemented it. The difficulty is not the math; it's astronomy —figuring out the correct parameters to plug into the clock transform. Reaching terrestrial accuracy is almost definitionally impossible, but you can still get plenty, astronomical measurements will only improve, and they're also the method against which any solution can even be judged in the first place. $\endgroup$
    – geometrian
    Commented Apr 5 at 2:00

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.