On the current master (96496f0), during a polar (24-hour) night, calling sunrise()/sunset() will occasionally throw the exact opposite error:
ValueError: Sun is always above the horizon on this day, at this location.
This happens at latitudes just above to the critical latitude for polar night. As an MVE: the coordinate (72, 0) in the UTC timezone does this in the middle of winter on Jan 28, 1970:
o = LocationInfo(timezone='UTC', latitude=72, longitude=0).observer
sunrise(o, date=datetime.date(1970, 1, 28))
The error is 1º wide: that is, coordinates from 71.4º to 72.3º will fail.
I see sunrise/sunset evaluates which error to throw by looking at the angle to zenith at noon to see if it's above the horizon. It seems that what causes it is a difference between what triggers the trig domain error in hour_angle and the zenith calculation. The full stack trace is:
Traceback (most recent call last):
File "sun.py", line 771, in sunrise
return time_of_transit(
File "sun.py", line 343, in time_of_transit
hourangle = hour_angle(
File "sun.py", line 243, in hour_angle
HA = acos(h)
ValueError: math domain error
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "test.py", line 7, in <module>
sunrise(o, date=datetime.date(1970, 1, 28))
File "sun.py", line 781, in sunrise
raise ValueError(msg) from exc
ValueError: Sun is always above the horizon on this day, at this location.
I was relying on this error to identify if it is polar night or polar day, especially near the poles at solstices. If there is a more robust way to calculate that directly, then I am also all ears.
On the current master (96496f0), during a polar (24-hour) night, calling
sunrise()/sunset()will occasionally throw the exact opposite error:This happens at latitudes just above to the critical latitude for polar night. As an MVE: the coordinate (72, 0) in the UTC timezone does this in the middle of winter on Jan 28, 1970:
The error is 1º wide: that is, coordinates from 71.4º to 72.3º will fail.
I see
sunrise/sunsetevaluates which error to throw by looking at the angle to zenith at noon to see if it's above the horizon. It seems that what causes it is a difference between what triggers the trig domain error inhour_angleand the zenith calculation. The full stack trace is:I was relying on this error to identify if it is polar night or polar day, especially near the poles at solstices. If there is a more robust way to calculate that directly, then I am also all ears.