Plate solving in Astrometrica --- help requested.

[Xilman]

I have attached a FITS image with a superbly accurate WCS in it, built by a local installation of astronometry.net. The image has also been transformed so that North is Up and East is left as shown in the ds9 viewer. When loaded into Astrometrica, with default settings (PA 0 degrees, no horizontal or vertical flip in the CCD tab), the image appears mirror-imaged compared to that in ds9. Using the "Flip horizontall" in the Images tab brings it into concordance with ds9.

The WCS in the image contains the centre of the image to within one arcsecond, the image scale to within 0.01% and the rotation to better than an arc minute. Despite this I have been totally unable to persuade Astrometrica (Astrometry->Data reduction...) to solve the image for itself. I have tried using the Gaia and the UCAC4 catalogues and I have tried manual searching for matches, all to no avail.

How can Astrometrica be persuaded to use the WCS in the FITS? If it can't, how do I persuade it to run its own plate solver successfully?

I can't be the only person running into this situation surely.

Thanks for any helpful advice.

Paul

new_V.fits

[Xilman]

I have found a solution to "How can Astrometrica be persuaded to use the WCS in the FITS?" on the Astrometrica mailing list. It can't and it is by design. Raab doesn't trust people to use an in-built WCS in case the image was taken at a site other than that reported to Astrometrica. A request was made for this facility to be made an optional extra.

The more I fight Astrometrica, the more I am tempted to re-write it from first principles.

[Paul M]

I hadn't heard of Astrometrica until you posted your question so I looked it up. It's probably more complex than I need and for most of my solving and photometry needs I currently use ASTAP, I think I've mentioned it to you elsewhere. Did you have a look at it?

http://www.hnsky.org/astap.htm

 

[han59]

I experimented with astrometrica some time.  The solving in astrometrica is very tricky to get working and not very forgiving. I have struggled with it too. Once you have it working, the automation for measuring and reporting of minor planets is superb.

Han

[Xilman]

15 hours ago, han59 said:

I experimented with astrometrica some time.  The solving in astrometrica is very tricky to get working and not very forgiving. I have struggled with it too. Once you have it working, the automation for measuring and reporting of minor planets is superb.

Han

Any chance of some assistance? I won't be upset if you decline.

I can put a typical plate-solved image and technical details about the telescope and camera somewhere accessible to you.

Thanks,

Paul

[han59]

I experimented with a trial license twice which are now expired. Only during the second trial, I succeeded in getting it running.  What I remember you have to set the solving parameter very close to the correct values. It will ignore the WCS header.

No problem to try is once more on my laptop (different computer), but can you share a better image then "new-V" ?   To identify and measure a minor planet position you wil be faster with ASTAP which will use the existing WCS solution in the header.

Han

 

[han59]

I can find only one minor planet in the image new-V with a magnitude 20.1. The limiting magnitude of the image is about 17 so it is not visible. The image is oversampled so the limiting magnitude is relative low.

 

[han59]

Okay did try image new-V in Astrometrica. I assume the focal length is 2600 mm, pixel size 7.356 micrometer. But failed to do the Data reduction.  What helps with the initial position is to write the following in the header:

RA      = '18 46 13.8'   
DEC     = '-10 11 50'

 

Han

 

 

[han59]

Okay I have it now running. The pixel size should be BEFORE binning. So FL=2600mm, pixel size 3.678 micrometer

This took some time. I loaded first an image of the Coathanger before it came clear. Also the flipping setting Astrometrica is confusing.  For an image North up, East left, the setting in Astrometrica for vertical flip should be checked.

What is also important it reads RA and DEC keywords. This is the position of the MOUNT.  The WCS solution position is written to keywords CRVAL1, CRVAL2.  Copy the position of CRVAL1, CRVAL2 keyword to keywords RA, DEC. Then they will be read in Astrometrica or enter the position manually in Astrometrica

 

Same asteroid as in ASTAP, T9364 seams equal to 299364:

 

Some important settings:

 

Not sure if below are the best settings, but it worked for me:

[han59]

Okay, I did some other tests. Astrometrica writes the image center and rotation to a solved FITS file using keywords OBJCTRA and OBJCTDEC and more or less obsolete rotation keywords CROTA1 and CROTA2. Unfortunately only the position is read back by either OBJCTRA, OBJTDEC or the RA, DEC keywords but not the rotation angle.  Awkward.

The Astrometrica solver is very limited and can't process any large position offset of maybe 10 arcminutes and no rotation angle. Easiest way to process would be to solve the image in nova.astrometry.net or ASTAP and then to copy/paste the exact position into Astrometrica and if solving fails enter the rotation NEGATIVE as indicated below.

The flip vertical/horizontal settings are also essential for solving. See previous posting

It is pity the standard WCS solution keywords  (CRVAL1, CRVAL2, CD1_1, ......,CD2_2) are not read and used as a starting point for solving. That would be a tiny mod in Astrometrica.

Han

 

 

[Xilman]

Han59, thank you very much for taking so much time and effort to help me with this pig of a program.

Now that a worked-through example is available I can work through analyzing a much more crowded field and stand a chance of getting as plate solution.

I will, of course, take careful notes in my observing record book of your settings above and any other tweaks which may turn up. It is a shame that the FITS header has to be edited but so be it.

For this particular project i am not trying to find asteroids (though lessons learned will undoubtedly be useful when I am) but rather to do PSF-fitting photometry on gravitational lensing candidates in the hope of finding exoplanets. The great majority of the candidates are located in star fields so dense that simple aperture photometry can't find enough clear sky around them.

BTW, the reason the stars are very over-sampled is that the seeing was not very good that night, and the sky transparency very poor, because of the calima that we have been suffering for far too long now. Under normal circumstances the seeing is 2 arcsec and the pixel size is 0.6 arcsec which, IMO, is about right for photometry so that the starlight is averaged over several pixels of possibly (probably) varying sensitivity. The limiting magnitude was monitored, more or less, while the image was being stacked from subs and should be comfortably below the expected magnitude of the (possibly) lensed star.

Thanks again,

Paul

[han59]

Hello Paul,

I'm playing a little more with Astrometrica. What is very strange is that when I solve your image in ASTAP (subsample force on 2) and then feed it to Astrometrica, the solution is read correctly and data reduction works straight out of the box. No rotation has to be entered. So the format of keywords CRVAL1, CRVAL2 data  in new_V image is a little different or for an other unknown reason only readable by Astrometrica if written by ASTAP. If I modify keywords OBJCTRA/OBJCTDEC or RA/DEC keywords in an ASTAP solved file nothing changes but when I modify CRVAL1/2 Astrometrica adapts. So this proves that reading keywords CRVAL1/2 is available and has preference in Astometrica!

Quote

 

SOFTNAME= 'SWarp   '

CRVAL1  =   2.815575885946E+02 / World coordinate on this axis 

CRVAL2  =  -1.019726821377E+01 / World coordinate on this axis 

 

ASTAP:

CRVAL1  =  2.815575679693E+002 / World coordinate on this axis

CRVAL2  = -1.019724485224E+001 / World coordinate on this axis 

 

 

ASTAP write the solution to the FITS header both in the new convention (CRVAL1/2, CD1_1, CD1_2....)  as the old convention (CRVAL1/2, CDELT1/2, CROTA1/2. The old convention is partly readable by Astrometrica.

 

So to improve your work flow, it could help to find out how to make the FITS files readable by Astrometrica. Have a look what happens with the file header if you use in Astrometrica in settings "Auto-Save FITS with WCS". If your desperate, I could help with a conversion routine to update existing headers to the old convention.

Han

 

 

[Xilman]

Thanks again, Han. I can now solve that FITS immediately without issue after very carefully typing in all your settings.

 

Now to try a much more crowded field ...

 

Added in edit: that works too!  I'm a happy bunny now.

Edited July 21, 2021 by Xilman