On 2025-01-07 02:53, Hal Murray via time-nuts wrote:
For truly long haul (as in solar system scale) time distribution,
there
are all kinds of interesting things going on.
What does "time distribution" mean when relativity is significant?
It depends on the desired timing accuracy and what you want to do.
Einstein synchronization (two way time transfer) works more or less the
same anywhere in the solar system, except that some of the clock
discrepancies will be caused by general relativity. That implies that if
you are doing 1 way time distribution (as, e.g., GPS does), you will at
some level need a relativistic time model.
For the Earth-Moon system, the lunar surface time (LCT) is blue-shifted
with respect to the Earth's surface. Based on the analysis of 4 separate
groups (including my own), the mean (LTC-TAI) rate is 6.48288 x 10-10
with a standard deviation about that mean of 1.90112 x 10-13 (i.e., that
appears to be ~ the current model error). This blue shift (i. e., LTC
running faster than TAI) corresponds to a clock rate difference of
56.012 μsec / day with a standard deviation of 0.016 μsec / day. (The
April 2 OSTP announcement with a claim of 58.7 µseconds per Earth-day
appears to neglect the 2.7116 µsec / day from the lunar surface redshift
due to lunar surface gravity)
Time-varying frequency / time shifts are dominated by the effects of the
lunar orbit eccentricity and solar perturbations in the lunar orbit. If
the lunar blue shift mean rate is removed, the RMS variation of LCT is
~398 nanoseconds, corresponding to a fractional frequency variation of
9.2752 × 10^−13, ∼ 5% of the mean lunar orbital redshift. General
relativistic 3-body effects come in at about 1 x 10^-13, and will in due
course be fun to look for.
So, if you only care about milliseconds, none of this matters. If you
care about microseconds, you should remove the rate, and have an
adjusted lunar time that roughly matches TAI. If you care about
nanoseconds, you need a relativistic model for lunar time. When we get
good optical clocks (with ffs ~ 10^-18 or better) on the Moon, we'll
have to model the frequency changes caused by the lunar solid body tides
(which cause radial motions of the surface with amplitudes of ~ 25 cm).
Regards
Marshall Eubanks
On 2025-01-07 02:53, Hal Murray via time-nuts wrote:
> Jim Lux said:
>> For truly long haul (as in solar system scale) time distribution,
>> there
>> are all kinds of interesting things going on.
>
> What does "time distribution" mean when relativity is significant?
It depends on the desired timing accuracy and what you want to do.
Einstein synchronization (two way time transfer) works more or less the
same anywhere in the solar system, except that some of the clock
discrepancies will be caused by general relativity. That implies that if
you are doing 1 way time distribution (as, e.g., GPS does), you will at
some level need a relativistic time model.
For the Earth-Moon system, the lunar surface time (LCT) is blue-shifted
with respect to the Earth's surface. Based on the analysis of 4 separate
groups (including my own), the mean (LTC-TAI) rate is 6.48288 x 10-10
with a standard deviation about that mean of 1.90112 x 10-13 (i.e., that
appears to be ~ the current model error). This blue shift (i. e., LTC
running faster than TAI) corresponds to a clock rate difference of
56.012 μsec / day with a standard deviation of 0.016 μsec / day. (The
April 2 OSTP announcement with a claim of 58.7 µseconds per Earth-day
appears to neglect the 2.7116 µsec / day from the lunar surface redshift
due to lunar surface gravity)
Time-varying frequency / time shifts are dominated by the effects of the
lunar orbit eccentricity and solar perturbations in the lunar orbit. If
the lunar blue shift mean rate is removed, the RMS variation of LCT is
~398 nanoseconds, corresponding to a fractional frequency variation of
9.2752 × 10^−13, ∼ 5% of the mean lunar orbital redshift. General
relativistic 3-body effects come in at about 1 x 10^-13, and will in due
course be fun to look for.
So, if you only care about milliseconds, none of this matters. If you
care about microseconds, you should remove the rate, and have an
adjusted lunar time that roughly matches TAI. If you care about
nanoseconds, you need a relativistic model for lunar time. When we get
good optical clocks (with ffs ~ 10^-18 or better) on the Moon, we'll
have to model the frequency changes caused by the lunar solid body tides
(which cause radial motions of the surface with amplitudes of ~ 25 cm).
Regards
Marshall Eubanks
