Today, I compared a bit longer set of Sun rise/transit/set times calculated by supernovas to USNO data and noticed another discrepancy pattern (event times are rounded to the nearest minute for convenience and converted to local time without accounting for DST):
Failed sunset calculation (nan) for 2025-01-07, then ±4-minute difference in rise or set times for certain day sequences.
Ephemeris: d440s.bsp
Elevation angle: -0.26, novas_standard_refraction
Location: lat 39.74, lon -104.99
Leap seconds, DUT1, polar_dx, polar_dy: 0.0 (which I believe is negligible for 1-minute precision results)
Before each event calculation, the frame time is set to the previous event time.
When trying to calculate civil twilight (elevation -6.0), even more failed calculations (nan) occur.
A similar shift is observed for the Moon events as well.
Today, I compared a bit longer set of Sun rise/transit/set times calculated by supernovas to USNO data and noticed another discrepancy pattern (event times are rounded to the nearest minute for convenience and converted to local time without accounting for DST):
Failed sunset calculation (nan) for 2025-01-07, then ±4-minute difference in rise or set times for certain day sequences.
Ephemeris: d440s.bsp
Elevation angle: -0.26, novas_standard_refraction
Location: lat 39.74, lon -104.99
Leap seconds, DUT1, polar_dx, polar_dy: 0.0 (which I believe is negligible for 1-minute precision results)
Before each event calculation, the frame time is set to the previous event time.
When trying to calculate civil twilight (elevation -6.0), even more failed calculations (nan) occur.
A similar shift is observed for the Moon events as well.