Data Taking

[Tom Droege <tdroege2@earthlink.net>]

Since the end of November I have been taking data with TOM1 such that it 
scans 20 degrees in declination and covers whatever crosses the 
zenith.  This has resulted in a 20 by 180 degree piece of sky for most of 
the month.  I have accumulated about 1.5 million measurements of 350,000 stars.

There is starting to be enough data to look for variables.  One problem is 
the overlap region.  Here stars are measured on one side of the CCD in one 
frame, and on the other side in the adjoining frame.  This results in two 
measurements of some stars.  These are almost always different, and when a 
search is made for variables, these stars dominate the ws statistic.  This 
even though they are a small fraction of the stars measured.  Hopefully, 
someone with data set 24 will figure out what to do about this and this 
problem will be reduced.

On one night, the RA drive was acting up.  When I reduced the data, I used 
a 7 pixel aperture instead of the normal 4 pixel one.  When I searched for 
variables, I got a lot with high statistics.  When I looked at the data, it 
was dominated by stars with one high point.  This point was always on the 
night with the large aperture and almost always in  a dense star field.  It 
is obvious that a larger aperture sometimes picks up an extra 
star.  Conclusion, blends are possible if the aperture is too large.  But 
we knew that.  I think this is an argument for always using the same 
aperture size.

I have 13 days of data.  Due to clouds and such, there are at most 8 or 9 
measurements of some stars.   This is enough to start to see interesting 
objects.  In this mode, a star is always measured in the same position of 
the CCD.  There is an indication that the statistics would be pretty good 
if I could (easily) eliminate the overlap stars.  I just sorted out the 
center of one 4 x 4 degree image to look at this.  Some of you could 
probably write 5 lines of code and sort out the centers of the areas.

OK, if someone wants to do this, the centers of the images start at 0,0 and 
step by 4 degrees in each direction.  So what I want to eliminate for the 
first square, for example, is the boundary -2 , -2 ; -2,+2 ; +2, -2 ; -2, 
-2.  0.2 degree of each side should do it, so I want to cut the 3.6 x 3.6 
degree center out of each square.  Assume we start with a collect_stars.pl 
output.

When I take a night's data and sort out all the 2 hit stars (these only 
occur in the overlap regions) and plot them, they look as I would 
expect.  i.e. just a frame of hits around each 4 x 4 degree square.  At 
first glance, they are tidy and uniform which would not be the case if 
there were errors in positioning the squares.

This just to show the practical difficulties of working on data.  There 
always seems to be something that prevents finding all but the obvious 
variables.  The process is slowly getting better as I learn what to do.

If I am lucky and actually get some data later this month, then there will 
be enough to post, I hope.

Tom Droege