Sun geocentric true coordinates

[bsics]

Hi, I implemented VSOP87 and ELP2000-82B to calculate sun and moon geocentric positions, and wanted to validate my implementations against JPL ephemerides. Out of convenience, I used skyfield. Everything was perfect for the Moon, but with the Sun positions I had consistently about 20 arcsec difference in the true coordinates between VSOP87 and skyfield. It was bad, because the estimated VSOP87 accuracy is ~2 arcsec

Here is the difference I get for 700+ positions over the span of 7 years.

Interestingly enough, the difference is gone when I remove the aberation (ligth time) correction in my code:

For that I went straight to Horizons. I found that the Horizons web application results are coherent with my VSOP87 code and not with skyfield.

For exemple, skyfield gives:

>>> planets = load('de440.bsp')
    ts = load.timescale()
    y, mo, d = (2012, 9, 27.084351851851853)
    t = ts.utc(y, mo, d)
    psun = planets['Sun']
    pearth = planets['earth']
    pos_sun = pearth.at(t).observe(psun)
    lat_sun, lon_sun, distance_sun = pos_sun.frame_latlon(ecliptic_frame)
>>>    lon_sun.degrees
184.38185171462558
>>>    lat_sun.degrees
0.0001164730838003623

My VSOP87 code gives:

>>>lon_vsop
184.37626251244774

And Horizons gives:

Date_________JDTT     R.A.___(ICRF)___DEC  R.A._(a-appar)_DEC.     ObsEcLon    ObsEcLat
***************************************************************************************
$$SOE
2456197.585129456     183.85449  -1.66898  184.01641  -1.73901  184.3761625   0.0001113
$$EOE
***************************************************************************************

I use skyfield v1.35 installed from anaconda, the conda_forge channel. The same conclusions with ecliptic or equatorial coordiantes.

[brandon-rhodes]

If you want Skyfield to apply aberration, you need to call .apparent() on the astrometric position returned by .observe(). Could you try that and see if it changes the result?

(Oh, also: I've edited your question so that it uses GitHub's Markdown formatting on your code; double-check to make sure you like how I applied their formatting. Thanks!)

[bsics]

That is great. This solves the systematic discrepancies I observed. For the date I used above:

>>>    lon_sun.degrees
184.37617577276538

Thank you!

[brandon-rhodes]

I'm glad it agrees more closely now! Here's the documentation in case you at any point want to read more:

https://rhodesmill.org/skyfield/positions.html#barycentric-astrometric-apparent