I wonder if anybody ever has assessed the uncertainty of the sun-earth distance calculations of pyephem? Since it might be used as an input to other calculations, this would be of relevance for further uncertainty analysis. Cheers, F
Yes! The accuracy is well defined.
The libastro library that PyEphem relies on uses the VSOP87 planetary theory — here it is in the PyEphem source code:
https://github.com/brandon-rhodes/pyephem/blob/master/libastro-3.7.7/vsop87.c
And here is a Wikipedia article giving more information:
https://en.wikipedia.org/wiki/VSOP_(planets)#VSOP87
The Wikipedia states that for the Earth-Moon barycenter, VSOP87 provides one arcsecond (1") of accuracy. If you are curious how far off the Earth-Moon barycenter position can be while still remaining within an arcsecond of accuracy, simply compute the circumference of a circle the size of the distance between the Earth-Moon system and the Sun (149.6 billion meters, according to Google) and divide by 360 (degrees) and then 60 (to get arcminutes) and then 60 (to get arcseconds):
149.6e9 m * 2 * pi / 360 / 60 / 60
-> 725 km
So over the thousands of years over which VSOP87 provides good numbers, the Earth-Moon distance to the Sun might be off by as much as the width of Germany.