Re: [TASS] photometric vs. non-photometric

[Chris Albertson <chrisja@jps.net>]

Andrew Bennett wrote:
>
> TN056: 26,000 close pairs vs 346,000 V-band measurements.
> Less than 10% even before you purged them. I don't think
> this can produce the drastic scatter observed.

The close pairs that made it into the database were all
detected.  What I think may happen it that close pairs
in the sky show up in the FITS image as an oblong blob
that is larger than the PSF of a single isolated star.
The detection algorithm in STAR would reject these
blobs.  On good nights, on one of the cameras the
stars resolve.  This effect, I think is independent of
a star's brightness.
>
> Consider a chunk of sky covered on average 25 times.
> Assume the coverage completely random.
> Then 99.9% of the time, one would expect the actual
> coverage for a particular star to lie between 10 and
> 40. Most of the time between 20 and 30. The actual
> distribution is quite different. Remember, this is
> brightish stars - probability of detection about unity.
>
> Of course, this could simply be the result of the
> coverage being non-random. But as nobody seems to
> know where their cameras were pointing, this is
> not easy to test ...

You can find out exactly where each camera was pointing
without going back to the raw data.  It can be derived
from the database with one query.  Each data point in
the observation table has stored with it a "frame number"
If you ask for the min and max RA and the min and max DEC
grouped by frames number, ordered by frame number you will
get a list of all the frames numbers and the field of
view of each.  frame numbers relate 1:1 with FITS images.
If you try this Michael's computer will thrash on the
disk maybe for a couple hours then send you a few
thousand lines of text.  If you limit the query to
only a few degrees of RA you can speed things up.

With this you can tell _exactly_ how many times a
specific chunk of sky was looked at by each
of the TASS cameras.  You don't have to use a statistical
approximation.  So for any star in the database that
is seen 2 or 100 times you can know exactly which cameras
detected it and which cameras looked but did not see it
all without going back to the data files.  If you want I
can help you get the SQL syntax correct to do the queries
>
> Andrew Bennett, Avondale Vineyard, Nova Scotia, Canada.

--
   --Chris Albertson             home: chrisja@jps.net
     Redondo Beach, California   work: calbertson@logicon.com