Discontinuity just after sunset

[martinmCGG]

PySolar seems to produce discontinuous sun elevation/altitude around sunset (and also near sunrise): there is a ~0.6deg jump in the following test case. May be related to #115.

Minimal example showing the issue:

import pysolar.solar as solar
import datetime
import matplotlib.pyplot as plt
import numpy as np

lat, long = 37.249870621332086, -115.81458511013524 # a randomly picked position
strange_timestamp = 1565491275.92175
epsilon = 420 #3600*12 #0.00008 # <- scale
x = np.linspace(strange_timestamp - epsilon, strange_timestamp + epsilon, 100)
y = [solar.get_altitude(lat, long, datetime.datetime.fromtimestamp(t).astimezone(datetime.timezone.utc)) for t in x]
plt.scatter(x, y, label='pysolar')

# optional, for comparison (pip install pysolar suncalc)
import suncalc
y2 = [np.rad2deg(suncalc.get_position(datetime.datetime.fromtimestamp(t).astimezone(datetime.timezone.utc), long, lat)['altitude']) for t in x]
plt.scatter(x, y2, label='suncalc')
plt.legend()
plt.show()

[pingswept]

Interesting. That certainly looks like a bug. Thanks for the test case.

My guess is that some correction factor is blowing up as altitude nears zero.

I’d welcome more details from anyone who wanted to dig into the details.