Re: More from ASAS

[Tom Droege <tdroege2@earthlink.net>]

Experts please comment!

Thank you Brian for pointing out this paper.  I have been following 
Pojmanski's work with interest.  In the early days of tass Bohden Paczynski 
brought Pojmanski to visit my new house wing to discuss this work.  At that 
time I was trying to provide cameras for Pojmanski's set up.  I just could 
not do everything so I dropped supporting him with 
cameras.  Sigh!  Actually what I said was that I could not build a 
special   camera system for him, but would provide him with a complete Mark 
IV.  He wisely turned me down and is now about a year and a half ahead of me.

I would like to point out the errors shown in Figure 
3.   http://xxx.lanl.gov/abs/astro-ph/0210283 Particularly Figure 3c.  Here 
Pojmanski compares the error he sees using the center of the image Figure 
3a vs the overlap of two images,  Figure 3c.  The error is more than twice 
when one looks at two images vs the center of one image.  Pojmanski says 
this is due to "systematic effects caused by non-perfect flat-fielding".

Note that whatever problem that Pojmanski has with flat-fielding, we should 
have less. He points out that his lenses are down 40-50% in the 
corners.  Ours are down more like 10% in the corners and 5% at the 
edges.  This should mean that we are applying only 1/5 the correction.  We 
should have an easier time of the flat field.  The errors I see are 
actually larger.  OK, I think errors in the Batavia suburb of Chicago 
should be larger than Las Campanas because we have much more sky to contend 
with.  Pojmanski seems to have more sensitivity, but the error is not much 
different.

I have been doing a lot of image reduction trying different methods.  I 
continue to think that these problems are due to the reference catalog.  I 
am trying to work up an experiment that will demonstrate this 
conclusively.  Sigh!  It is a struggle.

I am now getting very nice flat fields.  At least the flat and dark 
corrected images *look* very flat when examined with the ds9 sectioning 
feature.  The images are flat to within the noise.  I don't see that one 
needs to do much better than that.

One experiment is to take images with clouds and use my cloud removal 
program with various cuts.  This results in uneven flats if I don't remove 
all the cloud effects.  If the problem is the flat field, then I would 
--expect the errors to increase as the flats are made worse by removing 
fewer clouds.  There is little effect.  Michael's pipeline gets about the 
same answers in spite of the clouds.  Where there is an effect of about 0.2 
mag peak to peak, is with position of the star in the image.  This would 
seem to point to the reference catalog.  Since the I images errors are 
always larger than the V image errors, and the I refcat is derived from a 
computation, I would expect this to be the case.

OK, I will eventually work up a tech note that will attempt to demonstrate 
this.  At the moment I am in a big muddle of data trying to make sense of 
it.  I will welcome anyone who wants to join in this effort.  I can send 
you a data set of your choice.

I now have roughly the same scanning capacity as Pojmanski.  So it is hoped 
that I can do the northern hemisphere while he does the southern.  I keep 
looking at the results of this and other surveys and comparing it with my 
results.  I keep wondering why I am not doing as will.  One difference is 
that most of the other surveys always point to specific fields.  This 
results in a star always being measured by the same CCD pixels.  I note 
that the errors I see are much smaller if I follow a star and take 
successive images.  Now the star is in roughly the same position in each 
image.  OK, it could still be flat fielding, but I think not.

One way to demonstrate this one way or the other would be to use a specific 
set of comparison stars for each star measured.  Then as the star moves 
across successive images it would still use the same set of comparison 
stars instead of a different (large) set for each position as it does 
now.  At the moment, I do not have the ability to get into the code to 
accomplish this.

Tom Droege

At 10:11 PM 10/14/02 -0700, you wrote:


>The All Sky Automated Survey. Variable Stars in the 0h - 6h
>Quarter of the Southern Hemisphere
>
>
>\Brian