Jupiter and Io, 14th March 2007

Jupiter and Io from Murrumbateman Australia

Anthony Wesley, 14th March 2007

Notes:

Images acquired with my 13.1" newtonian ( LEXX ) as a sequence of 16 bit FITS frames at 40 frames per second, 50 seconds capture in each of Red, green and Blue.
Seeing was poor at 3.30am local time (1630UTC) but then improved dramatically to sub-arcsecond by 5.30am local time (183UTC) and stayed at this level until 6am (1900UTC) when the morning dawn light was starting to interfere.
Here is an Animation (1.8Mb) of frames from 1836UTC to 1851UTC. The frames in this animation have been less processed than the still images below.

1836UTC

1840UTC

1844UTC

1847UTC

Experimental de-rotation and combining of colour images

I'm experimenting with de-rotating in software to let me combine images that come from different times - Jupiter rotates so fast that it's normally not possible to combine images like those above, the animation should make it clear that there is too much rotation to simply merge the images together.

With my software package ninox there is an experimental command-line flag "-rotate" that will attempt to automatically find and rotate the planet to allow images from different times to be merged. In theory this results in a decrease in the residual noise in the combined image.

The software finds the planet automatically, and assumes that it is aligned so that it's axis of rotation is horizontal. A positive rotation angle will rotate the planet from right to left, and a negative angle will rotate from left to right. I was careful to align the planet to have the polar axis vertical before capturing the raw data.

I used the 1836UTC image as my base, and then derotated the 1844UTC and 1847UTC images to match it. By trial and error I found that 5 degrees of rotation on the 1844UTC image and 8 degrees of rotation on the 1847UTC image was about right. I also had to apply a very small anticlockwise rotation on the resulting images in the GIMP before I could combine them as the axis of rotation was not quite vertical (jupiter was slightly tilted in the original images).

Here are the commandline options I used to generate the rotated images:

ninox -outdir=rot -cutout=off -rotate=5 j20070323-1844-rgb.bmp
ninox -outdir=rot -cutout=off -rotate=8 j20070323-1847-rgb.bmp

Notes:

-outdir=rot indicates that the output images are to be written to a subdirectory named "rot" so that they do not overwrite the originals.
-cutout=off instructs ninox to process the whole input image, instead of its default behaviour of locating and centering.
-rotate=N specifies that the planet in the image is to be rotated by N degrees (+ve is right-to-left)

Notice in the rotated images that the preceding and following limbs are not properly handled, this is due to the change in lighting that should take place as the planet rotates. At this stage I do not try and correct this, so some artifacts are visible.

Combined image

Ok, so here is the final image from recombining the original 1836UTC image with 1844UTC (rotate 5 degrees) and 1847UTC (rotate 8 degrees):

It's clear that this image has less noise than any of the individual images above, although there is an increase in the visible artifacts around the preceding and following limbs. Also Notice that Io is now a blur, as it's proper motion is not compensated for by this derotation.