In Sep2018 I discovered that the growing number of acm header cards has begun to approach 200. This is causing some problems with my routine for replacing cards. Here are the notes.
A0.fits = original version of 20180826T110433.5_acm_sci.fits
FINAL.fits = fixed up version
I will find the position of a star that I recogize in both the images and the dss.fits image.
Here x,y are image pixels and ra,dec are from WCS in header
% ds9 A0.fits
x,y,ra,dec = 541.6 594.8 02:48:44.548 +65:15:47.08
% ds9 FINAL.fits (before wcsf_rough)
x,y,ra,dec = 542.1 593.8 02:48:44.500 +65:15:47.00
Hence, the direct ds9 measures between the original and fixed images
are nearly the same.
What happens when I run the wcsf_rough code? I processed FINAL.fits
From the dss image star 9 (my fiducial star):
Ra,Dec(ds9) = 02:48:38.470 +65:16:11.13
From catalog: 2.810684440 65.269911100 (hrs,deg)
% sexcd 2.810684440 65.269911100 hd
02:48:38.4640 +65:16:11.6800
Ra,Dec(catalog) = 02:48:38.464 +65:16:11.68
Ra,Dec(ds9) = 02:48:38.470 +65:16:11.13 # These are in agreement
When I measure the wcsf_rough-processed image:
% ds9 FINAL.fits (After wcsf_rough)
x,y,ra,dec = 541.9 594.2 02:48:38.470 +65:16:11.61
Ra,Dec(catalog) = 02:48:38.464 +65:16:11.68
*** This appears to give the correct ra,dec now.
I run usno_look_wcs to check this:
usno_look_wcs FINAL.fits Blue 19.00 1 # Yes! This looks very good.
======================================
Just as a check, I do the same exercsie with A0.fits:
I run usno_look_wcs to check this:
usno_look_wcs A0.fits Blue 19.00 1 # No, the points do not line up. There is
some probelm with using the original
image processed with wcsf_rough
The #9 star position is:
ds9(A0 after wcsf_rough) 542.4 594.3 2:48:38.451 +65:17:35.40
*** This is different from the original image, but still not right.
The original image treated with wcsf_rough
In A0a.fits:
CTYPE1 = 'RA---TAN' / Gnomonic Projection
CRPIX1 = 541.990 / Axis 1 Reference
CRVAL1 = 42.160259247 / Ra at Frame, J2000 (deg)
CTYPE2 = 'DEC--TAN' / Gnomonic Projection
CRPIX2 = 594.500 / Axis 2 Reference
CRVAL2 = 65.293196
% sexcd 42.160259247 65.293196 dd
02:48:38.4624 +65:17:35.5056
02:48:38.470 +65:16:11.13 # Star 9 from dss
The image treated with acm_ccd_fixup:
In FINAL.fits
CTYPE1 = 'RA---TAN' / Gnomonic Projection
CRPIX1 = 541.680 / Axis 1 Reference
CRVAL1 = 42.160259247 / Ra at Frame, J2000 (deg)
CTYPE2 = 'DEC--TAN' / Gnomonic Projection
CRPIX2 = 594.010 / Axis 2 Reference
CRVAL2 = 65.269912720 / Dec at Frame, J2000 (deg)
% sexcd 42.160259247 65.269912720 dd
02:48:38.4624 +65:16:11.6858
02:48:38.470 +65:16:11.13 # Star 9 from dss
**** It appears that when wcsf_rough processed the original image (A0.fits)
the DEC value (CRVAL2) was not rewritten. In the original image, not
yet processed with acm_ccd_fix:
CTYPE1 = 'RA---TAN'
CRPIX1 = 176 / reference pixel
CRVAL1 = 42.160035
CTYPE2 = 'DEC--TAN'
CRPIX2 = 379 / reference pixel
CRVAL2 = 65.293196
**** So, this is worth resolving. I have to understand why wcsf_rough does not
properly re-write the DEC (via CRVAL2).
The code that does the rough WCS installation is:
build_rough_wcs
** When I run wcsf_rough on A0.fits and I get to the
point of: "You have run either usno_identify or ds9_crpix_set." I see:
% cat ds9_crpix_set.out
Y 540.48 595.72 02:48:38.4624 +65:16:11.6858 9 # This is the correct x,y and ra,dec
** When I run wcsf_rough on FINAL.fits and I get to the same point
% cat ds9_crpix_set.out
Y 545.06 595.92 02:48:38.4624 +65:16:11.6858 9
So, the two images are treated to same to this point.
I run the build_rough_wcs on the two different images:
/home/sco/play/wcs2/T_final FINAL.fits 02:48:38.412 +65:16:11.88 541.53816 594.83184 # This is correct
/home/sco/play/wcs2/T_a0 A0.fits 02:48:38.449 +65:17:35.38 541.95708 594.72741 # This is WRONG
So, the problem seems to be with the build_rough_wcs treatment of A0.fits.
What is causing this incorrect treatment? I will dig into build_rough_wcs.
I run build_rough_wcs in stand-alone mode and I find I get the same
header file entries to be inserted:
% cat T_a0/ascii.WCS
CTYPE1 = 'RA---TAN' / Gnomonic Projection
CTYPE2 = 'DEC--TAN' / Gnomonic Projection
CRPIX1 = 541.460 / Axis 1 Reference
CRPIX2 = 594.940 / Axis 2 Reference
CRVAL1 = 42.160259247 / Ra at Frame, J2000 (deg)
CRVAL2 = 65.269912720 / Dec at Frame, J2000 (deg)
% cat T_final/ascii.WCS
CTYPE1 = 'RA---TAN' / Gnomonic Projection
CTYPE2 = 'DEC--TAN' / Gnomonic Projection
CRPIX1 = 542.020 / Axis 1 Reference
CRPIX2 = 596.110 / Axis 2 Reference
CRVAL1 = 42.160259247 / Ra at Frame, J2000 (deg)
CRVAL2 = 65.269912720 / Dec at Frame, J2000 (deg)
The code that does the new card insertion: headrev.sh
headrev.sh A.fits ascii.WCS $finalim
So, I can run this in each T_ directory as:
headrev.sh A.fits ascii.WCS B.fits
***** I see that headrev.sh does perform differently for these two images. Why?
Fuck! It is the large number of cards now in the acm FITS header.