UTC issues with Epoch.rise_set() and Epoch.utc2local()

[sta-c0000]

Hi,
Just started looking at pymeeus, and it looks great, but perhaps I'm misunderstanding:

The documentation states that "Epoch internally stores time as Terrestrial Time (TT), not UTC.", "When a UTC time is provided, the parameter utc=True must be given. […] it is converted to (and stored as) Terrestrial Time (TT). If utc is not provided, it is supposed that the input data is already in TT scale."

Yet the Epoch.rise_set()'s documentation notes: "The results are given in UTC time.", and the example provided calls get_full_date() without the utc=True parameter.
Perhaps jrise and jsett should be set with the utc=True param and the example changed for consistency?

Also, it appears that Epoch.utc2local() currently can't work if the difference between current local time and current UTC time cross the 24 hour boundary (e.g. 23:00 and 3:00) because it's doing (localhour - utchour).
Also utc2local() does not work for dates other than today, where the current timezone offset may change (DST).

BTW, instead of using utc2local, I'm doing something like this:

from datetime import *
# from zoneinfo import ZoneInfo # optional for Python 3.9+

from pymeeus.Angle import Angle
from pymeeus.Epoch import Epoch
from pymeeus.Sun import Sun

def epoch_datetime(e):
    edt = e.get_full_date(utc=True) # ! not when rise_set? !
    s = edt[5]
    return datetime(*edt[:5], int(s), int(s%1 * 10E5), tzinfo=timezone.utc)

longitude, latitude, altitude = Angle(45.423646), Angle(-75.698592), 90
e = Epoch(2021,7,1)
dtrise, dtset = map(epoch_datetime, e.rise_set(longitude, latitude, altitude))
dtls = dtset.astimezone() # UTC to local timezone
# dtls = dtset.astimezone(ZoneInfo('EST5EDT')) # Specific timezone, Python 3.9+
print(f'Sunset = UTC: {dtset}, Local: {dtls}')
dtwinter = epoch_datetime(Sun.get_equinox_solstice(2021, 'winter'))
dtlw = dtwinter.astimezone() # UTC to local timezone
# dtlw = dtwinter.astimezone(ZoneInfo('EST5EDT')) # Specific timezone, Python 3.9+
print(f'Winter = UTC: {dtwinter}, Local: {dtlw}')

Output:

Sunset = UTC: 2021-07-02 00:57:11.147205+00:00, Local: 2021-07-01 20:57:11.147205-04:00
Winter = UTC: 2021-12-21 15:59:17.402847+00:00, Local: 2021-12-21 10:59:17.402847-05:00

Notice sunset local time is EDT (-4) and winter local time is EST (-5) as expected.
(Sunset appears incorrect here due to TT vs UTC confusion.)

Thank you!

[architest]

Dear Sir, Thank you for your e-mail. I'll look into it but it may take some time for me to get back to you. Best regards, Dagoberto Salazar

…

On Fri, Jul 2, 2021 at 5:38 PM sta-c0000 ***@***.***> wrote: Hi, Just started looking at pymeeus, and it looks great, but perhaps I'm misunderstanding: The documentation states that *"Epoch internally stores time as Terrestrial Time (TT), not UTC.", "When a UTC time is provided, the parameter utc=True must be given. […] it is converted to (and stored as) Terrestrial Time (TT). If utc is not provided, it is supposed that the input data is already in TT scale."* Yet the Epoch.rise_set()'s documentation notes: *"The results are given in UTC time."*, and the example provided calls get_full_date() without the utc=True parameter. Perhaps jrise and jsett should be set with the utc=True param and the example changed for consistency? Also, it appears that Epoch.utc2local() currently can't work if the difference between current local time and current UTC time cross the 24 hour boundary (e.g. 23:00 and 3:00) because it's doing (localhour - utchour). Also utc2local() does not work for dates other than today, where the current timezone offset may change (DST). BTW, instead of using utc2local, I'm doing something like this: from datetime import * # from zoneinfo import ZoneInfo # optional for Python 3.9+ from pymeeus.Angle import Angle from pymeeus.Epoch import Epoch from pymeeus.Sun import Sun def epoch_datetime(e): edt = e.get_full_date(utc=True) # ! not when rise_set? ! s = edt[5] return datetime(*edt[:5], int(s), int(s%1 * 10E5), tzinfo=timezone.utc) longitude, latitude, altitude = Angle(45.423646), Angle(-75.698592), 90 e = Epoch(2021,7,1) dtrise, dtset = map(epoch_datetime, e.rise_set(longitude, latitude, altitude)) dtls = dtset.astimezone() # UTC to local timezone # dtls = dtset.astimezone(ZoneInfo('EST5EDT')) # Specific timezone, Python 3.9+ print(f'Sunset = UTC: {dtset}, Local: {dtls}') dtwinter = epoch_datetime(Sun.get_equinox_solstice(2021, 'winter')) dtlw = dtwinter.astimezone() # UTC to local timezone # dtlw = dtwinter.astimezone(ZoneInfo('EST5EDT')) # Specific timezone, Python 3.9+ print(f'Winter = UTC: {dtwinter}, Local: {dtlw}') Output: Sunset = UTC: 2021-07-02 00:57:11.147205+00:00, Local: 2021-07-01 20:57:11.147205-04:00 Winter = UTC: 2021-12-21 15:59:17.402847+00:00, Local: 2021-12-21 10:59:17.402847-05:00 Notice sunset local time is EDT (-4) and winter local time is EST (-5) as expected. *(Sunset appears incorrect here due to TT vs UTC confusion.)* Thank you! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#14>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFQIZWIZCTY7JFYCQR7RW2DTVXMNZANCNFSM47W6J25A> .

[sta-c0000]

Thank you for your response, of course there is no rush.

To be more clear about why Epoch.utc2local() does not work, here are some examples:

Let's say now is July 7 2021 at 20:00 in New York:
Epoch.utc2local() will return: 72000.0, which is +20 hours!
Because now=2021-07-07T20:00:00-04:00 and utcnow=2021-07-08T00:00:00+00:00

Let's say now is November 7 2021 at 19:00 in New York:
Epoch.utc2local() will return: 68400.0, which is +19 hours!
Because now=2021-11-07T19:00:00-05:00 and utcnow=2021-11-08T00:00:00+00:00

...of course these problems persist until the next day (when past 24 hours).

Also, utc2local() is not Epoch date and DST aware, so:
Let's say now is July 7 2021 at 12:00 in New York:

e = Epoch(2021,11,7,12) # targeting a different date than now
e.utc2local() # will return -14400.0 instead of the correct -18000.0.

I personally don't see the Epoch.utc2local() function as useful even if it worked, because Python already handles timezones correctly using it's ubiquitous datetime objects. Perhaps adding an Epoch function that returns a datetime object (with tzinfo=datetime.timezone.utc) would be more useful?

On an unrelated note, I can't seem to get correct Epoch.rise_set() results for the location above regardless of TT/UTC, but that might be another issue...

Thanks again.

[janbredenbeek]

Hi,

The Epoch.rise_set() method is not based on Meeus's algorithms and only gives approximate results, since it uses only a single value of the Sun's position (which varies during the day). A more accurate result can be obtained by calculating the Sun's positions of the previous, current and next day and calling the Coordinates.times_rise_transit_set() method to find the rising and setting times:

ra, delta, r = Sun.apparent_rightascension_declination_coarse(epoch) # 0.01 degree is sufficient for this
epsilon = true_obliquity(epoch)
ra1, delta1, r1 = Sun.apparent_rightascension_declination_coarse(epoch - 1.0)
ra2, delta2, r2 = Sun.apparent_rightascension_declination_coarse(epoch + 1.0)
# This method expects negative Eastern longitudes!
rising, transit, setting = times_rise_transit_set(-longitude, latitude,
                                                  ra1, delta1,
                                                  ra, delta,
                                                  ra2, delta2,
                                                  Angle(0, -50, 0), #Atmospheric refraction for Sun
                                                  epoch.tt2ut(year, month),
                                                  Angle(epoch.apparent_sidereal_time(epsilon,nutation_longitude(epoch)) * DAY2HOURS, 
                                                  ra = True))
rising = Epoch(epoch + rising / 24.0)
transit = Epoch(epoch + transit / 24.0)
setting = Epoch(epoch + setting / 24.0)
y, m, d, h, mi, s = rising.get_full_date()
print("Rising time: {}:{:02}:{:02}".format(h, mi, round(s)))
y, m, d, h, mi, s = transit.get_full_date()
print("Transit time: {}:{:02}:{:02}".format(h, mi, round(s)))
y, m, d, h, mi, s = setting.get_full_date()
print("Setting time: {}:{:02}:{:02}".format(h, mi, round(s)))

Since the times_rise_transit_set() method returns times in UTC there is no need to set utc=True in get_full_date().
I've also noticed that get_full_date() with local=True actually subtracts the time difference obtained with utc2local() where it really should add it!

Therefore, it's best to write your own method to convert UTC to local time (based on a given date and local DST rules).