Failing rectangle astrometry

40 views
Skip to first unread message

Jerry Zhang

unread,
Sep 7, 2022, 4:54:07 AMSep 7
to astrometry
 Hello,

I am trying to precisely solve a rectangle optical field (2.8' x 1.4') but always failed, although I have tried almost all the parameters and constrains, for example 

solve-field xxx.fits --ra 277.126 --dec 26.84 --radius 0.5 --overwrite --scale-high 0.1


This is the field:
Screen Shot 2022-09-07 at 09.36.27.png

And this is the correspondent PanSTARRS field, you can see the field is not crowed and positions of the stars are quite clear.
Screen Shot 2022-09-07 at 09.26.06.png


I wonder if it is due to the shape of the field?


Thanks all,

Jerry

Dustin Lang

unread,
Sep 7, 2022, 9:25:18 AMSep 7
to Jerry Zhang, astrometry
Hi,

That's quite a small field.  By default, solve-field's --scale-low is 0.1.  So you would have to set "--scale-low 0.03" for it to have any chance of working.

cheers,
dustin

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/5346b694-0c56-4858-a660-7625ad3dac3bn%40googlegroups.com.

Jerry Zhang

unread,
Sep 12, 2022, 8:11:00 AMSep 12
to astrometry
Hi Dustin, 

Thanks for the suggestions.
Finally it failed to manage to solve, no matter how I set --scale-low .  Is that just because the field is too small to have enough reference stars in the catalog? Which catalog is astrometry using? 

Jerry

Bryan

unread,
Sep 12, 2022, 8:57:15 AMSep 12
to astrometry
Jerry

I recall the default indices for a local install are 2_MASS.


Bryan

Dustin Lang

unread,
Sep 12, 2022, 9:33:52 AMSep 12
to Bryan, astrometry
For a field as tiny as that, you'll want the "5200-series (light)" linked here,
specifically the index-5200-31.fits file for your field's RA,Dec.

cheers,
dustin



Jerry Zhang

unread,
Sep 28, 2022, 9:42:25 AM (4 days ago) Sep 28
to astrometry
Hi Dustin,

I used the Gaia catalog index 5200 series but still failed.... I think in the field there are at least 10 Gaia sources according to Aladin.

Could the reason be the rectangular shape of the field?

Screen Shot 2022-09-28 at 14.40.57.png
Best,
Jerry

Dustin Lang

unread,
Sep 28, 2022, 9:49:21 AM (4 days ago) Sep 28
to Jerry Zhang, astrometry
I think you need to set the --radius if you set --ra, --dec.  A fairly large radius (1-5 degrees) should work fine.

Mark Copper

unread,
Sep 29, 2022, 5:40:45 PM (2 days ago) Sep 29
to astrometry
I have difficulty with small fields as well. Generally, I had been trying 2.6' x 2.6' ROI to keep sizes down, but even with the entire camera field at 7.3' x 4.8' I've had trouble. And that's using the 5200-series indices. Especially challenging are sparse fields and fields with a bright star. And this even when the coordinates of the field are known within a few minutes.

So it got me to wondering whether it would help to modify the "odds-to-XXX" default parameters, whether XXX was tune-up or solve or reject or stop looking. For example, the odds-to-solve parameter defaults to 10^9 (to 1?). Would it be practical to ask for lower odds? If I *think* I have a target star lined up, and I *know* I'm in the neighborhood, and yet the field doesn't solve, would I get any useful information by setting odds-to-tune to 10^4 and odds-to-solve to 10^6? Practically speaking, that is.



Dustin Lang

unread,
Sep 29, 2022, 6:25:59 PM (2 days ago) Sep 29
to Mark Copper, astrometry
Hi,

Generally, I would expect the larger 7.3x4.8 fields to be more successful -- 2.6x2.6 is getting down to where the Gaia catalog starts to run out of enough stars.  And yes, usually you need 15-20 stars to get a solution.  Reducing the --odds-to-* could potentially help with that part, and if you're using --ra, --dec, --radius (and --scale-low, --scale-high) to cut down the search space, the chance of getting a false positive match shouldn't be too bad.  I would be interested to hear if it works if you try -- I haven't messed with that much :)

If you want to post a failing image that you think should work, please go ahead and I may be able to comment further.

cheers,
dustin



Reply all
Reply to author
Forward
0 new messages