Re: brief note on astrometry of Mark IV comatic images

[Tom Droege <droege@wwa.com>]

I have been looking at the images with "Image Scientist" (Note to Mike, I 
actually use Deep Sky as sometimes Image Scientist thinks my raw data file 
is a fits file and tries to read it.  Deep Sky does not have this problem.)

This produces a file of stars sorted first by Y and then by X.  One does 
not see duplicates, though they could be there.  I have done the usual 
checks of picking out stars of various amplitudes and looking at them on 
the image.

It seems to me that there will always be blended stars.   It is just a 
question of how we decide to handle them.  Note that there are guaranteed 
(most stars are double, I recall) to be many unresolved double stars.   I 
suppose that we will have a few cases of real double stars that fall in and 
out of our resolution.

OK, I am out of my area of experteze here, but it seems to me that we can 
either:

1) Check each measurement of a star as it is made for the possibility that 
it is close to another star by using some higher resolution catalog.

2) Wait to check until we have accumulated many measurements on a star, and 
it has popped up on a cut list as a possible variable.

Seems like 1) should be better if it does not take too much time.

As I think about it, this is an old problem, even for us.  There are the 
blended stars from the tenxcat.  We just have to do the best we can, 
studying what the best people have done in this area.   I had better stick 
to hardware.

Tom Droege



At 02:23 PM 6/1/00 -0400, you wrote:
>On Thu, 01 Jun 2000 00:47:13 +0100, Andrew Bennett 
><andrew.bennett@ns.sympatico.ca> wrote:
>*>On Sun, 28 May 2000 19:34:29 -0500, you wrote:
>*>
>*>>I find that I am getting 6000 or so stars per medium star density image=20
>*>>when I set things to detect to mag 15 or so.   This is 600+ pixels per=20
>*>>star.  On these images it is hard to find star pairs where it might be a=
>*>=20
>*>>problem to separate them.   I am waiting patiently to get an image in =
>*>the=20
>*>>milky way to look at.  It looks (by eye) pretty easy to separate stars 5=
>*>=20
>*>>pixels apart.
>*>
>*>Confusion errors are insidious. They don't show
>*>up on image to image comparisons unless you change
>*>the PSF i.e buy a bigger telescope. I just got out=20
>
>Tom, if you have a list of 6000 detections, how about sorting them by
>location and see if any of them overlap? For instance, if your software
>produces a list by pixel location, and uses an aperture of say five pixels;
>would it "find" an overlapping pair as two candidates with positions that are
>LESS than five pixels apart? (Make sure your software does not automatically
>reject such candidates.)
>
>Er, "sort" is a MS-DOS utility program. If you do this manually remember
>to use root 2 times the aperture as the distance for each axis - in
>my example that's 7 pixels or so. A QBASIC program could read a sorted
>list and look if X (or Y) changes by less than 7, and print both lines.
>
>If your list was in a database, or even a spreadsheet, I
>imagine a proper query would find them. No doubt one of our colleagues
>could take your list and do this work in less than a hour, given the
>proper info and your list.
>
>Herb Johnson
>
>Herbert R. Johnson              http://pluto.njcc.com/~hjohnson
>hjohnson@pluto.njcc.com         voice 609-771-1503, New Jersey USA
>              amateur astronomer and astro-tour guide
>            S-100 computer parts, manuals as "Dr. S-100"
>     reseller of 68K Macs & accessories for your computing pleasure