pyephem equinox producing unexpected sunrise/set azimuth

Question

This is less of a python/pyephem question and more of an orbital mechanics question but here goes:

Pyephem reports that the vernal equinox will occur Thu 2014-03-20 12:57 EST. This is the day when sunrise occurs closest to true east (90º) and sunset closest to true (270º), pretty much worldwide.

Calculating time and azimuth sunrise/set on that day for lon -78:30, lat 35:30 elevation 124m, using a horizon of '-0:34' and pressure of 0 (to get numbers as close to the US Naval Observatory) I get (adjusted for localtime, EST):

          Date    Sunrise    SlrNoon     Sunset    sunrise az     sunset az           
Thu 2014-03-20   07:19:57   13:20:55   19:22:27    89:55:28.0   270:19:08.5

With solar noon occurring slightly after the equinox, I would expect sunset to be closer to 270:00:00.0 than sunrise be closer to 90:00:00.0 but the opposite is true. Sunrise is about 4.5 arminutes off true east and sunset is about 19 arcminutes off true west. Why?

Answer 1

At the equinox I would expect the Sun to be at roughly due-east and due-west when it intersects the true horizon itself, rather than when it reaches a depth of 34 arcminutes below the horizon on each side of the sky. To reach that point of depression below the horizon, the center of the Sun will have to travel farther — which means in both cases that you are asking about a point father north and farther below the horizon than the actual points that are due east and due west.

Since we do not have access to your code itself, could you try re-running it with horizon set to zero and see if you get points closer to east and west?