[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Photometric solutions: Disk 16 vs. Disk 17




  I've modified the pipeline to handle the data on Disk 17.
There were three major issues:

      a. the convention for image file names changed.  The old file
         names looked like this:

                  H3R1816.755

         and the new file names look like this:

                  hvra1992810.fits

         I know that the first character encodes the observatory site
         ('h' means Tom's house), the second character encodes the
         filter name ('v'), but I don't know what "ra1992810" means.
         Could someone let me know?

      b. the FITS header keywords and values changed

      c. there weren't enough stars in the subset of Tycho-1 to
         match some of the fields

  In order to fix item c., I made a new subset of Tycho-1, in which
I relaxed the criterion on stdev of photometric measurements from 
0.05 mag to 0.07 mag.  That increased the number of stars over the entire
sky from 257,000 to 382,000; the number of stars in each Mark IV field
increased from 30-50 to 45-70.  Some of the fields with only 30 stars
were failing to match up properly, but with 45 or so catalog stars,
they all matched up.  

  I haven't done any examination in depth of the results, but I thought
I'd show you a comparison of the photometric solutions for entire 
Disk 16 versus the first disk in Data Set 17.  Remember

          Disk 16       taken Sep 28, 2000
                        sets of 9 images of each field
                        all exposures 50 seconds

          Disk 17       taken Mar 25, 2000
                        1 image of each field, at spacings of 0.7 degree
                        all exposures 100 seconds

  So, here are the photometric solutions for each night.  Recall that
the solution relates raw, instrumental magnitudes to calibrated magnitudes
via an equation like: 

        calibrated mag  =   raw mag  +  a   +   b * (raw color)

The parameters we solve for are "a", the zero point in each passband,
and "b", the color term in each passband.  

            passband   #stars     a     +/-       b    +/-         RMS
----------------------------------------------------------------------------
Disk Set 16    V         749    -8.108 0.008    0.047 0.009       0.096
               I         749    -7.930 0.007   -0.110 0.007       0.154

Disk Set 17    V        1084    -8.782 0.016    0.035 0.013       0.079
               I         779    -8.287 0.015   -0.100 0.012       0.118


  What does the values mean?

        1. the color terms are the same on each night, within the 
           uncertainties.  This is a good sign.

        2. the zero points are NOT the same.  The values for Disk Set 17
           indicate that stars of some given magnitude (say, 8.00) appeared 
           _fainter_ on Disk 17 than on Disk 16.  That is, a star which
           was really V=8.00 gave rise to fewer "counts" on Disk Set 17
           than on Disk Set 16.  

           This might be due to clouds (more clouds on March 25, 2001),
           or to dust on the optics (most dust on March 25, 2001), or
           changes in the sensitivity of the chips (I hope not!),
           or even a big systematic error in the exposure times 
           (but that's really unlikely -- they'd have to be longer
           than requested by a factor of 1.8 or so).
 
  My guess is that the night of March 25, 2001, had thin clouds,
and thus less starlight hit the CCDs than on the night of Sep 28, 2000.
I also wouldn't be surprised if there were more dust on the lenses
at the later date.

  More analysis as I can find the time.  I need to run the data for
two more disks from Set 17 through the modified pipeline, of course.

  From the first disk "a" in Set 17 alone, I find 34 (V,I) pairs of
images, and succeed in calibrating the magnitudes of 21,182 stars.
Of these, 13,978 were measured at least 3 times each.  Looking at
repeated measurements of stars with at least 3 observations, I see
that the standard deviations are larger than expected for the bright
stars: there's a very large scatter in the stdev, from 0.01 to 0.05 mag
in the V-band, from 0.01 to 0.03 mag in the I-band.  Hmmm.  

  On first glance, it appears to me that clouds might account for
some of this increased scatter, but I'll need to do some additional
work to verify the guess.

  As I work on this data, I am realizing more and more that it would
be extremely useful to have a set of diagnostic values printed out
somewhere (probably into a log file(s)), showing quantities on an image-
by-image basis.  For example, given a list like this:

          image number          sky value           FWHM (pix)
         --------------------------------------------------------
             1                    1802               3.4
             2                    1815               3.3
             3                    1799               3.4
             4                    1830               3.5
             5                    2102               4.0
             6                    2250               4.2
             7                    2883               4.1

one could very easily and quickly conclude "clouds moved in after image 3".

  I guess I'll need to build some routines to create such log files
for the pipeline.

  It would also be helpful if someone just happens to have a routine 
which reads FITS images, and creates GIF or JPEG thumbnails from them
using "threshold" and "scaling" information supplied by the user.
The ImageMagik program "convert" _almost_ does it all, but doesn't
allow me to set the contrast myself.  Any ideas?

                                         Michael Richmond