by alter-ego » Sun Nov 25, 2018 7:09 am
Chris Peterson wrote: ↑Sat Nov 24, 2018 2:40 pm
alter-ego wrote: ↑Fri Nov 23, 2018 5:08 pm
APOD Robot wrote: ↑Fri Nov 23, 2018 5:05 am
...
Still, as part of the annual
Leonid meteor shower, the meteor trail does point back to the
shower's radiant. The constellation Leo is high above the horizon and off the top left of the frame.
It is a very nice picture, but I'm curious why the great-circle extension of the meteor trail back to Leo ends up ~15° south of the radiant?
I'm assuming this is a single image so the radiant is correctly positioned, and because the trail is well defined with enough stars visible, the back-tracking error is small.
How did you determine this error? As someone who writes software to identify radiants in images, I can say that geometrical projection effects require using a spline or some other complex curve to backtrack the path. As the FOV gets larger, it gets trickier to identify the radiant, especially visually.
That's exactly right. Except for determining RA/Dec coordinates for a few points within the meteor trail, I made no attempt to graphically extrapolate the trail to the expected radiant position (~45°away) using this or any images. Also, I couldn't estimate a radiant position error with a single meteor. However, I did estimate an
in-passing proximity error.
I assumed the accepted practice that meteor trails are ideally straight as viewed projected onto the celestial sphere (ignoring in-flight deviation), and I determined the nominal separation between the positions in the trail is ~12.5° (about 60% of the trail length). So over this test length I did not see any measureable curvature wrt the nearby stars which means the position coordinates have negligible inherent error from image distortion within the test-length range (also any in-flight deviation not a concern). I determined reasonable RA/Dec estimates for these few trail locations using stars <3° from the position of interest. Using pairs of points, I calculated the few great circles containing the trail
and intersecting the Leonid radiant RA circle. Given a conservatively large point-position error = ±½°, the "best-fit" great circle intersects the Leonid radiant RA circle at Dec = 6°± ~1°. That puts the meteor effectively passing about 16° south of the nominal Leonid radiate at { RA = 10h 5m, Dec = 22°18' }. I base my estimated 1° uncertainty on point location accuracy. Coordinate error due to local image distortion is negligible. Regardless of what little distortion exists within the 12.5°x 3° image region, the 16° proximity error is well dominated by meteor direction.
[quote="Chris Peterson" post_id=287587 time=1543070415 user_id=117706]
[quote=alter-ego post_id=287577 time=1542992918 user_id=125299]
[quote="APOD Robot" post_id=287567 time=1542949551 user_id=128559]
...
Still, as part of the annual [url=https://solarsystem.nasa.gov/asteroids-comets-and-meteors/meteors-and-meteorites/leonids/in-depth/]Leonid meteor shower[/url], the meteor trail does point back to the [url=https://apod.nasa.gov/apod/ap061118.html]shower's radiant[/url]. The constellation Leo is high above the horizon and off the top left of the frame. [/quote]
It is a very nice picture, but I'm curious why the great-circle extension of the meteor trail back to Leo ends up ~15° south of the radiant?
I'm assuming this is a single image so the radiant is correctly positioned, and because the trail is well defined with enough stars visible, the back-tracking error is small.
[/quote]
How did you determine this error? As someone who writes software to identify radiants in images, I can say that geometrical projection effects require using a spline or some other complex curve to backtrack the path. As the FOV gets larger, it gets trickier to identify the radiant, especially visually.
[/quote]
That's exactly right. Except for determining RA/Dec coordinates for a few points within the meteor trail, I made no attempt to graphically extrapolate the trail to the expected radiant position (~45°away) using this or any images. Also, I couldn't estimate a radiant position error with a single meteor. However, I did estimate an [i]in-passing[/i] proximity error.
I assumed the accepted practice that meteor trails are ideally straight as viewed projected onto the celestial sphere (ignoring in-flight deviation), and I determined the nominal separation between the positions in the trail is ~12.5° (about 60% of the trail length). So over this test length I did not see any measureable curvature wrt the nearby stars which means the position coordinates have negligible inherent error from image distortion within the test-length range (also any in-flight deviation not a concern). I determined reasonable RA/Dec estimates for these few trail locations using stars <3° from the position of interest. Using pairs of points, I calculated the few great circles containing the trail [i]and[/i] intersecting the Leonid radiant RA circle. Given a conservatively large point-position error = ±½°, the "best-fit" great circle intersects the Leonid radiant RA circle at Dec = 6°± ~1°. That puts the meteor effectively passing about 16° south of the nominal Leonid radiate at { RA = 10h 5m, Dec = 22°18' }. I base my estimated 1° uncertainty on point location accuracy. Coordinate error due to local image distortion is negligible. Regardless of what little distortion exists within the 12.5°x 3° image region, the 16° proximity error is well dominated by meteor direction.