Comments on Civil Time decision tree

From: Rob Seaman <seaman_at_noao.edu>
Date: Fri, 23 Sep 2005 16:37:04 -0700

I've appended what I call the Civil Time Decision Tree v0.5.
Reminder - the intent is to provide a scaffolding for characterizing
any possible civil time standard (whether practical or not). Please
let me know if you perceive holes or inaccuracies in the tree. You
will also undoubtedly let me know all the problems you perceive with
my editorial comments in the next few paragraphs, but the more basic
question is whether we have identified all possibilities for Civil
Timekeeping.

In any event, the LEAPSECS list has been very quiet lately while we
wait for the entertainment to be provided by the November ITU
meeting. Perhaps we can fill the time most productively with some
reasoned debate. Here are my own comments about the major nodes of
the decision tree:

I, II & III) YES - we should have a single international civil time
standard that is restricted to the immediate vicinity of the Earth
(applies to spacecraft whose operations centers and scheduling remain
on Earth).

IV) Civil time should be explicitly recognized as a flavor of Mean
Solar Time. If it can ALSO be synchronized with Atomic Time, as with
the current UTC standard, all the better. It is, however, trivial to
show that short of a cosmic realignment of the human psyche, that
Civil Time must remain synchronized to Solar Time to better (much
better) than one second per day:

     1) Assert that a divergence of Civil Time of one hour from Solar
Time is "noticeable".

     2) Assert that a noticeable divergence is unacceptable over a
period of a decade.

     3) One second per day is more than an hour's divergence after a
decade.

You may disagree with either the amplitude or the period, but
reasonable values will lead to similar or tighter divergence
constraints. In your own analysis, please distinguish secular
effects (leap seconds or hours) from periodic effects (e.g., daylight
saving - Spring forward AND Fall back).

Finally, I argue that recognizing that Civil Time must mimic Solar
Time to better than a second per day is the same thing as saying that
Civil Time IS Solar Time. That being the case, we should do the best
job we know how of ensuring Civil Time and Solar Time remain paired
from one day to the next. Another way of saying that is that time
zones (of an hour's width) in a world with leap hours are an
oxymoron. Yes, the current divergence of ~2 ms/day is small enough
to "cheat" and hide the effect for most purposes (ignoring the
gawdawful expense to astronomy), but is cheating really what the
precision timing community wants to do?

It may be annoying that Mother Earth spins irregularly, but spin she
most certainly does.

Rob Seaman
National Optical Astronomy Observatory

---
Civil Time Decision Tree v0.5:
I) Existence [Adopt an international (or intergalactic) civil time
standard?]
     A) yes
     B) no
II) Multiplicity [How many standards?]
     A) one
     B) many
For each civil time standard:
III) Locale
     A) restricted to Earth [projects or users, not necessarily
hardware]
     B) other than Earth [e.g., Martian rovers]
     B) Solar system scope
     C) truly Universal
IV) Synchronization [with underlying time standard(s)]
     A) Overloading
         1) Single
         2) Dual [e.g., UTC is synchronized to both Solar and Atomic
scales]
         3) Multiple [e.g., Earth+Mars+Atomic]
     For each fundamental standard:
     B) Phenomenology
         1) Natural
             s) Solar
                 i) apparent
                 ii) mean
                 iii) standard
                 iv) daylight saving
                 v) Sidereal
             b) Tidal
             c) Lunar
             d) Seasonal
             e) Planetary
                 i) various
         2) Physical
             a) Atomic
             b) GPS
         3) Dynamical [roughly both]
             a) various
     C) Optimization
         1) remain within tolerance (e.g., 0.9s)
         2) minimize figure of merit (e.g., DUT1)
         3) minimize uncertainty in schedule
     D) Method
         1) steps (I'm tired of the word "leap)
             a) arbitrary size
             b) fixed size
                 i) millisecond
                 ii) second
                 iii) minute
                 iv) hour
                 v) other
         2) rates ("rubber seconds")
         3) both
     E) Adjustments
         1) discontinuity
         2) time slice (e.g., UTS)
     F) Triggering
         1) manual (e.g., reset a clock by hand)
         2) automatic
     G) Scheduling
         1) adaptive as needed
         2) sliding
         3) fixed in advance
     H) Sampling
         1) continuous
         2) daily
         3) monthly
         4) annually
         5) decadal
         6) pre century
         7) greater than a century
         8) arbitrary
     I) Corrections
         1) issued as table
         2) issued as formula
         3) implicit in method
     J) Residuals [e.g., DUT1]
         1) None
         2) Fixed
         3) Monotonic
         4) Periodic
         5) Piecewise
         6) Arbitrary
     K) Notification
         1) as needed
         2) daily
         3) monthly
         4) six months
         5) annually
         6) decade
         7) fifty years
         8) none needed

Received on Fri Sep 23 2005 - 16:32:48 PDT