OverflowError: math range error when calculating get_radiation_direct at sunrise.

[SamuelLarkin]

Hi, I came across a bug where, at first glance, there seems to be a problem calculating the direct radiation right around sunrise and sunset.

Here's the error message.

2022-03-29 06:36:32 - collector.elevation - INFO - Going up.
Exception in thread Collector:
Traceback (most recent call last):
  File "/usr/lib/python3.9/threading.py", line 954, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.9/threading.py", line 892, in run
    self._target(*self._args, **self._kwargs)
  File "/home/balta/git/Hepa/collector/collector.py", line 537, in run
    self.update_sun_position()
  File "/home/balta/git/Hepa/collector/collector.py", line 519, in update_sun_position
    self.sun.update(
  File "/home/balta/git/Hepa/collector/collector.py", line 95, in update
    rad = radiation.get_radiation_direct(date, self.elevation)
  File "/home/balta/git/Hepa/venv/lib/python3.9/site-packages/pysolar/radiation.py", line 50, in get_radiation_direct
    return flux * math.exp(-1 * optical_depth * air_mass_ratio) * is_daytime
OverflowError: math range error

I'm using pysolar==0.10.

[pingswept]

Yeah, that looks like a bug.

Could you post the inputs that produced the error?

[SamuelLarkin]

Sorry for the wait. My logging was insufficient to reproduce the bug with the info I had.

Here's a snippet that should help you reproduce the bug.

import datetime
import pytz
from math import isfinite
from pysolar.solar import (
            get_position as get_sun_position,
            radiation,
            )
latitude  = 45.4924
longitude = -72.3150
date = datetime.datetime(2022, 4, 3, 10, 27, 15, 830398, tzinfo=datetime.timezone.utc)
azymuth, elevation = get_sun_position(latitude, longitude, date)
rad = radiation.get_radiation_direct(date, elevation)

In get_radiation_direct radiation.py:50 we have this code return flux * math.exp(-1 * optical_depth * air_mass_ratio) * is_daytime

air_mass_ratio: -961255.0150425568
altitude_deg: -5.960518136860493e-05
day: 93
flux=1159.3544752333482
is_daytime: False
optical_depth: 0.16979271842852264
when: datetime

then math.exp(-1 * optical_depth * air_mass_ratio) is yielding OverflowError: math range error.

I'm wondering why bother do that math.exp() if we then multiply it by False which is 0?

Unfortunately, I'm not knowledgeable enough in the math behind computing the sun's radiation to properly solve this.

[pingswept]

That’s trying to calculate roughly e^100000, which is around 10^43000, so that’s where the overflow is happening.

I suspect the right solution is to alter get_air_mass_ratio() so that it is restricted to a reasonable range.

Patches welcome, if anyone wants to think that through.