Sun Declination Error at Equinoxes Too Large?

While checking accuracy for historical applications is very welcome, please note that Stellarium uses VSOP87 (IIRC based on some JPL DE2xx) and not JPL DE4xx (?) like today's Horizons, and should only be compared to other systems based on VSOP87 in order to assess correct implementation or systematic drifts. (Else you just compare VSOP87/DE2xx against DE4xx). I am pretty sure however that there is some quirk concerning ecliptical obliquity that becomes apparent for dates far from now, not yet located in code, and therefore still awaiting a fix.
G.

Georg -
I certainly agree with both your points. My question was academic - I saw the "Precision - Stellarium Wiki" reference to VSOP87 and the expected accuracies, so I was curious about the equinox timing error growth. Also I thought you might be interested in the comparison.

I'd like to say I've been using Stellarium for a few years - not for maximum-accuracy ephemeris data, but for the great visual interface with the cosmos it provides. I thoroughly enjoy the visual realism and plethora of observer-related controls for practical features that allow me to explore the skies over thousands of years. I've done some really cool simulations and animations. A recent example was setting myself as observer up on Stereo A, and reproducing the view in a recent time-lapsed video of ISON, Encke, Earth and Mercury (and an unidentified asteroid!). I loaded in a vast data base of asteroids which allowed me to eventually identify the interloper as asteroid Kleopatra. There is no way I could have done that with any public ephemeris I know of. Stellarium's broad functionality and high accuracy allowed me to solve that puzzle.

So, a long-term, super-accurate ephemeris calculator Stellarium may not be, but it is pretty close! However, the real benefit it brings to the public far, far outweighs the 12-arc-minute equinox timing error over 2000 years that I was asking about!

Keep up the great work! You all have created a really great product for the majority of people.
Thanks again, and Best Regards,
Jon

Important note: precession is hard coded at the moment and has const value :(

Hello Alexander,

Maybe you can clarify something for me.

I thought past and future J2000 coordinates are precessed. Interestingly, when I compare the J2000 coordinates over the same 4000-year interval and at the same TD, the maximum difference between Horizons and Stellarium is < 1 second in RA and 8 arc seconds in Declination. E.g. Horizons J2000 coordinates for the Sun on -2000-01-01 00:00 TD are identical to Stellarium's J2000 coordinates at the same TD, but Stellarium's RA/Dec of-date are off.

I don't see how Stellarium very accurately predicts precessed J2000 coordinates while the coordinates-of-date are off.

Do the predictions for J2000 coordinates use a different algorithm than the predicted coordinates-of-date?

~Jon

As far as I can recall, Stellarium uses J2000 internally, which is then
corrected to coordinates-of-date to be displayed. Though I need to check
exactly whether, when and where the correction(s) are made.

Also, please clarify what you mean by "J2000 coordinates are precessed". As
far as I know, "J2000" is a shorthand for a specific coordinate system -
geocentric equatorial using the equinox of J2000.0 as an anchor point, and
the fixing of the equinox to that date is done precisely because of
precession.

Bogdan - Thank you very much for your response and question. I have one remaining question.

Yes, you are correct that the Epoch 2000 coordinate system is a reference / anchor system, and, as in the past, likely to transition to J2050 system. I've always known this but I think I made the wrong assumption about the steps for the ephemeris calculation.

*********
After some thought (and calculation), I realized that accurate J2000 coordinate predictions (past and future) don't need prior application of the precession algorithm. I fooled myself into thinking that a J2000 coordinate could not be calculated WITHOUT precession. It's clear to me now (I hope :) that all detailed ephemeris position calculations can be done in initially the reference coordinates, and then lastly transformed to the precessed coordinate system.

Is this perspective correct?

It would certainly explain my confusion about how J2000 coordinates are very accurate while the coordinates-of-date are much less accurate.
*********

On a second note, I've verified my earlier declination error magnitudes AGREE WELL with those expected from Stellarium's precession error (contrary to my original thought). I used Stellarium's NCP J2000 coordinates, HORIZONS J2000 Sun's coordinates and some basic spherical trigonometry to account for the majority of the previously-mentioned Stellarium declination-of-date error.

BOTTOM LINE:
I think I've answered my own question. Alexander and Georg proposed the likely answer(s) to the declination drift error.

This question was expired because it remained in the 'Open' state without activity for the last 15 days.