[Spice_discussion] Accuracy of ECLIPDATE frame for computing equinoxes/solstices?

[Barry Carter]

I tried using ECLIPDATE defined in:

http://naif.jpl.nasa.gov/pub/naif/VEX/kernels/fk/RSSD0002.TF

to compute the date of the 2020 equinox, but it shows me the equinox 
occurs between 0342 and 0343 ET (not UTC) on Mar 20 2020.

According to http://aa.usno.navy.mil/data/docs/EarthSeasons.php however, 
the equinox occurs at 0350 UTC, and the ET-UTC difference isn't large 
enough to explain the discrepancy.

Is there a better version of ECLIPDATE I can use and/or is there a better 
way to find equinoxes and solstices?

I know I can use ITRF93 (and IAU_EARTH outside of ITRF93's range) to find 
equinoxes, by doing this:

spkezp_c(10,et,"ITRF93","CN+S",399,v,&lt);

looking for a z value (v[2] value) of 0, and then converting from et to 
UTC, but this won't work for solstices.

I tried using minimum and maximum z values in the ITRF93 frame for 
solstices, but the results were inaccurate, since the solstice is 
technically defined as when the sun's ecliptic longitude is 90 or 270 
degrees, not when it's declination is maximal/minimal.

I can provide significantly more details (and code) on request, but I thought
I'd ask the general question first: is ECLIPDATE sufficiently accurate to
compute solstices? Is there a more updated version of ECLIPDATE I should be
using? If not, what's the best frame for computing solstices?