refcat problems

[Tom Droege <tdroege2@earthlink.net>]

I have looked at the difference between what the pipeline measures for the 
refcat star magnitudes and the magnitudes listed in refcat for 
DS-24-3.  This has been done frame by frame.  Some of the stars are 
measured 2 magnitudes off from their catalog values.  It is obvious that 
each maverick star will affect the measurement depending on whether or not 
it is used for a particular frame.

As might be expected, the V catalog has a tighter error grouping than the I 
catalog.

This is not dominated by noise - i.e. the 2 mag error stars still have a 
tight error grouping.  It is a real offset.  There is a clear bias in the 
symmetry of the groupings.  Lots to show when I can post a TN with plots.

I am working on cleaning up the refcat.  As I have suggested earlier, I 
think this may be the dominant source of the measurement error.

The present plan is to remove the outliers from the refcat, then process 
DS-24-3 and again look at the differences between the refcat and the 
measured values.  Possibly loop through this process a few times.  There 
are plenty of reference stars.  Removing stars above 1 sigma only gets rid 
of about 25% of the reference stars.

Experts can point out how this will bias the result.  Obviously it 
will.  However, I don't see what else to do without a reliable all sky 
photometry catalog.  This should generate better relative photometric 
measurements particularly in the case where stars appear in different 
positions in the frame.  Possibly later the result can be calibrated from 
some future photometric catalog.

The assumption is that there is lots of junk in the tycho 2 
catalog.  Undetected variable stars that have moved on to much different 
values and the like.   It is hoped that selecting refcat stars that agree 
closely with the tass measurements will result in a more stable photometric 
measurement.  We shall see.

Tom Droege