Welcome to iraf.net Friday, March 29 2024 @ 11:00 AM GMT

Buglog #344

  • Thursday, August 01 1996 @ 03:00 PM GMT
  • Contributed by:
  • Views: 1,196
MODULE: imcombine, combine
SYSTEM: V2.10.2-V2.10.4
NUMBER:	344
MODULE:	imcombine, combine
SYSTEM:	V2.10.2-V2.10.4
DATE:	Thu Aug  1 09:00:22 MST 1996
FROM:	valdes

BUG:	When there are exactly three images, mclip=yes, and scales, zero
	levels, or weights are computed and are not all the same there is a
	bug that will confuse the scales and weights for about 1/6 of the
	pixels.  This becomes more significant as the scales/zeros/weights
	differ more from each other.  This affects computation of the
	weighted average (combine=average) and the sigma clipping
	algorithms (ccdclip, crrej, sigclip, avsigclip, and pclip).
	In addition, even if the scales/zeros are all the same but the
	gains and read noise values are different for the ccdclip or
	crrej algorithms the gains and read noise values will be misapplied
	in 1/6 of the cases.

	The bug is that when mclip=yes the pixels are sorted and the
	idenification of which image each pixel came from is kept.  When
	there are exactly three images an explict sort is done but in one
	of the six possible orders of the 3 pixels the identification
	mapping to the image is wrong for two of the pixels.  Thus the
	wrong scale, zero, weight, gain, or read noise will be used.  If
	they are equal there is no problem but if they are not equal the
	weighted average and the scaling of the sigmas in the clipping
	rejection algorithms will be wrong.  If the values are close there
	is still little error.

	The effects of this are probably negligible in most cases
	unless the scales or CCD parameters of the images are grossly
	different.  The only workaround is to set mclip=no.

STATUS:	Fixed for the next release expected to be V2.11.

Buglog #344 | 0 comments | Create New Account

The following comments are owned by whomever posted them. This site is not responsible for what they say.



Privacy Policy
Terms of Use

User Functions

Login