Wrong time calculation for some angles

[tmpcache]

[I'm not quite sure what I am reporting here.]

Before version 0.16.6 correct times were being shown for the first three items on the list on the first screenshot.

Now, astronomical twilight is given as 'none'. And the two custom angles below it are wrong as well.
Actual sunrise and actual sunset are correct. I don't know about the rest of the items on the list.

[forrestguice]

I'll take a closer look and verify the contents of that first screenshot.

The second screenshot looks like an older version of NaturalHour. I see a bug in the twilight periods that should have been fixed in v0.2.3. The latest release v0.3.1 has some cool new features too, but I'd recommend waiting for v0.3.2 if you are running Android 12+ (and use the widget).

[tmpcache]

I updated Natural Hour to v0.3.2 but this bug is still persisting, which could mean that the bug isn't from Natural Hour.

[forrestguice]

I'm actually not sure there is a bug - after looking closely at the output of your first screenshot it actually looks right to me. I'm curious what makes you believe that its incorrect? My current thinking is that the UI is just confusing (and could be refined).

According to https://www.timeanddate.com/sun/germany/berlin those sunrise/sunset, civil, and nautical twilight times are correct. However astronomical twilight does not have a dawn or dusk time - during this time of year the sun sets less than 18 degrees below the horizon at solar midnight (at your location). I believe this phenomenon is referred to as "midnight twilight". It never becomes "night" but remains astronomical twilight all night long, which is why the app lists "none" for the rising and setting times (there is no "astronomical dawn" or "astronomical dusk" on these days).

A few degrees farther north and it becomes more pronounced - when the sun sets less than 6-7 degrees at solar midnight it is basically light outside "all night long". I think this is referred to as a "white night". Even farther north the sun doesn't set below the horizon at all, which is called a "midnight sun".

So I don't think any of the recent patches broke anything in this respect. The date of those releases just happened to coincide with the transition to "midnight twilight". I think it is potentially confusing because "rising" times indicate the start of twilight, while "setting" times indicate their end, so, for example, the time given for evening civil twilight indicates its end (and the start of nautical twilight).

I hope this clears up some confusion.

[tmpcache]

Did you just say that the sun doesn't reach the 16°-below-the-horizon point at the given time and place? I have another software that gives the times.
For example, SunTimes says that today sun will be 14° below the horizon after sunset at 00:31 but the other app says it's at 23:35.
For 16° below the horizon before sunrise, SunTimes says "none" but the other app says 3:16.

I have to get in touch with the developer of the other app and ask him about the library he uses.

[forrestguice]

Did you just say that the sun doesn't reach the 16°-below-the-horizon point at the given time and place?

Yes, according to https://www.timeanddate.com/sun/germany/berlin, today the sun reaches -14° below-the-horizon in Berlin, so there aren't any times for -16° (scroll the mouse along the altitude graph to the time of solar midnight).

Suntimes reports the minimum altitude to be -14.06°, but it also remains very close to that angle for some time (maybe an hour). The underlying library is also sometimes inaccurate (+-0.25°), so under these circumstances I think its reasonable to expect a large degree of error. I don't know what that other app is doing but its not hard to imagine reasons the outputs might differ.

I think if there is a bug though, its in the underlying library (Suntimes relies on the output of https://github.com/MenoData/Time4J). There is definitely some degree of error, and while reasonable, it becomes much more noticeable near the boundaries on the graph.

[forrestguice]

says that today sun will be 14° below the horizon after sunset at 00:31 but the other app says it's at 23:35

I think both times look reasonable, with 00:31 being a rising time and 23:35 a setting time (Suntimes reports -14° at 00:21 and 23:55 (sea level)).

Try toggling the "use elevation" setting on/off (under Settings -> General) and the app may actually report none. I think accounting for altitude is probably where the aforementioned degree-of-error originates.

[verdulo]

Similar latitude, also the Sun is always higher than -18° above horizon. When you tap on the Sun Position bar, Astronomical twilight is wrongly calculated as 1 d 21 h instead of 3 h.

[forrestguice]

@verdulo Thanks, I think your screenshots show a bug introduced while fixing #878. It should be fixed in the next patch.

@tmpcache I am going to close this issue, but don't let that deter you from opening another (or continuing to add details to this one).
I'm not sure if there is anything actionable for me to do. It is an interesting edge case; the sun approaches some angle but doesn't cross it (that a fraction of a degree in accuracy makes a huge difference).