Subject: non-repeatable patterns in GRIM bias level

From: James Rhoads

Submitted: Thu, 16 May 1996 18:33:08 -0400

Message number: 14 (previous: 13, next: 15 up: Index)

While reducing some December 1995 GRIM data, I noticed that the bias
pattern shows non-repeatable structure in the y direction, with typical
levels a few 10s of ADU and spatial scales of about 10-20 pixels.
This is most easily seen by subtracting two dark current frames of equal
exposure time.  The result shows clearly visible horizontal stripes
perhaps 1/3 of the time.  These are best examined by projecting the
difference frame along the x axis to get the mean of the 256 columns.
The pattern amplitude is usually largest near the bottom of each
quadrant, but exceptions occur.
Similar patterns can sometimes be seen in the differences of sky frames.

The differences of sky frames, he pattern appears to me to repeat, i.e.,
the structure in rows 129-256 is the same as that in rows 1-128.

I confirmed that the effect was still present in late April.
The potential effect on photometry of faint, diffuse objects is
substantial, although less than the old quandrant-to-quadrant
fluctuations in the bias level.  

To "destripe" my data, I have written a fortran program that 
(in its simplest form) subtracts from rows j and j+128 the median pixel value
in rows j and j+128, for all j in  0<j<129.  Pixels within some circular
region of the frame may be excluded from the medians (to avoid
subtracting out diffuse sources), and a model image may be subtracted
from each data frame before de-striping (though I would not recommend
this under most cuircumstances).  The code uses the IRAF Imfort image
i/o routines, and can be compiled using the IRAF "fc" task.
This method seems to reduce the problem substantially, but of course it
will not work well for objects larger than the GRIM field of view.

Subtraction of two dark frames also often reveals a low-level diagonal
striping in the lower right hand quadrant.  This looks to me like it is
nicely periodic and could probably be removed in the Fourier domain, but
so far I have not tried to do so.
					-James Rhoads
APO APO APO APO APO  Apache Point Observatory 3.5m  APO APO APO
APO  This is message 14 in the apo35-grim archive. You can find
APO  the archive on
APO  To join/leave the list, send mail to
APO  To post a message, mail it to