NavList:

   

Reply

For example, let's simulate a computation of the 2023 equinoxes and
solstices with the tools of 20 years ago. Then compare results with the
tools of today.

To simulate the old computation use the JPL DE406 ephemeris (1997), the
IAU 2000 precession model (actually a correction to the 1976 model), and
2000B nutation, which is a simplified model but accurate to 1 mas (milli
arc second) from 1950 to 2050.

For the modern computation use the JPL DE441 ephemeris (2020), and the
current IAU precession / nutation models 2006 / 2006A. The latter is the
complete model with 1300 terms and includes the small adjustments to
make it compatible with 2006 precession.

Geocentric apparent longitudes of the Sun at the times given by my site
for 2023:

359°59'59.98" old
359°59'59.98" new

90°00'00.01"
90°00'00.01"

180°00'00.02"
180°00'00.02"

270°00'00.01"
270°00'00.00"

where 0.04" of longitude is about 1 second of time. Clearly predictions
from "20 years ago" are very accurate. However, the phenomena are
independent of Earth rotation and therefore long term predictions must
use a time scale which ticks SI seconds with no step adjustments. You
could use TT, TAI, or even GPS time, but not UTC or UT1.

I think irregularities in the latter two scales are behind some
discrepancies in equinox and solstice tables. They may have been
computed years ago with estimated Earth rotation, but Earth did not
cooperate. A comparison between my table and Wikipedia shows two
discrepancies (rounded in the wrong direction) in 2023.

http://sofajpl.com/solstice/
https://en.wikipedia.org/wiki/Solstice

I won't say Wikipedia is wrong, but one of us is a little weak on being
right. It practical terms it doesn't matter, except that some people,
like our original poster, may wonder why astronomy cannot be more precise.

--
Paul Hirose
sofajpl.com

   

Reply