Add a function to calculate the solar eclipse amount for an observer

[ayshih]

Go go solar eclipse!

Checking with other resources indicates that we can use our coordinates framework to calculate the start/end of total solar eclipses to within a few seconds.

To do:

• Verify accuracy
• Unit tests
• Gallery example
• Changelog entry
• Add a docstring note recommending the use of a JPL ephemeris
• Add an option to get_body_heliographic_stonyhurst() to silence the info message for the light-travel-time correction?

[ayshih]

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.dates import DateFormatter

import astropy.units as u
from astropy.coordinates import EarthLocation
from astropy.time import Time

from sunpy.coordinates import sun

obstime = Time('2024-04-08 18:43') + np.arange(-100, 100) * u.min
observer = EarthLocation.from_geodetic(-96.808891*u.deg, 32.779167*u.deg).get_itrs(obstime)
amount = sun.eclipse_amount(observer)

plt.figure()
plt.plot(obstime.datetime64, amount)
plt.gca().xaxis.set_major_formatter(DateFormatter('%H:%M'))
plt.title('From Dallas on 2024 April 8')
plt.ylabel('Eclipse percentage')
plt.xlabel('UTC')
plt.grid()
plt.show()

[wtbarnes]

Want to make this an example gallery entry? 😁

[ayshih]

It's already on the to-do list. =). Appropriately spiffed up, of course

[ayshih]

Accuracy looks excellent once I remember to use the JPL ephemeris rather than Astropy's built-in ephemeris

[Cadair]

Somewhat off on a tangent, but can we pass EarthLocation as an observer? i.e. skip this line. If not why not? 😆

[ayshih]

Because EarthLocation doesn't include obstime?

[Cadair]

Drives the PR car off down another street

Couldn't we still do the conversion though? Wont most transforms pull the obstime from the frame if it's missing on the observer?

[ayshih]

I think you're not looking at how the observer variable is later used. Despite its name, it's not being used as the value for an observer frame attribute. The function eclipse_amount() takes a single argument, which is the coordinate of the observer including the obstime. If the function were to accept EarthLocation for the observer location, the function would need to additionally require the obstime to be supplied as a second argument.

[Cadair]

Should we not put this in constants somewhere rather than in here?

[ehsteve]

And also define it better. It's not a radius but a fraction of Earth's radius.

[ayshih]

It's of course a dimensionless fraction, but it's also the value of the lunar radius in units of Earth radii.

[ayshih]

I don't know where to stick such constants, since we don't really have a home for non-Sun constants. I've tacked on a TODO comment, and someone in the future can figure it out.

[ehsteve]

I tested this code and it worked great. This is a cool example and this capability could be used for an update to NumFocus. To make it more relevant, how about choosing a point where both 2023 and 2024 eclipses pass and show both the total eclipse and annular eclipse curves on the same plot?

[ayshih]

To make it more relevant, how about choosing a point where both 2023 and 2024 eclipses pass and show both the total eclipse and annular eclipse curves on the same plot?

Okay, I've picked a location on the outskirts of San Antonio that is on both eclipse paths.

[dstansby]

Looks good, I've left a few comments 👍

[dstansby]

Should fix the test...

[Cadair]

A couple of very minor tweaks.