Re: V image fainter than I

[Stupendous Man <richmond@stupendous.cis.rit.edu>]

  The current pipeline does detect objects which appear in the I-band
image alone; they are listed with their proper (RA, Dec) positions,
but a purely instrumental magnitude, in the .ast files.  I do not
know if these files are stored anywhere, or if they are discarded
as intermediate products.

  However, the penultimate step in the pipeline, the "collate"
stage, is where they are dropped.  The list of objects found in
V-band is compared to those found in I-band, and only those detected
in both bands are sent to the final step of photometric calibration.

  A number of people very reasonably request that the pipeline
be modified to keep objects which are detected only in I-band.
Could you please tell me what to do with the instrumental
measurements of the stars?

     - the final step, photometric calibration, does two things
              to the instrumental magnitudes: it shifts them
              to match the standard magnitude system, and it
              also uses the (V-I) color to correct for
              a color term in the detectors.

     - the orphan I-band measurements will have no V-band magnitude.
              If one performs a shift on them, without a
              color correction, then they will be systematically
              off in magnitude; this simple procedure would treat them
              like stars of color (V-I) = 0, which are relatively
              hot, blue stars.

     - moreover, these true (V-I) color of these stars might very
              well change if the stars are variable.  If one does
              not correct for that changing color, one gets 
              incorrect I-band magnitudes, on top of the problem
              mentioned earlier.

  The easiest thing would be to assign I-band orphans some very red
color -- say, (V-I) = 3 or 4 -- and use that to make the color
corrections.  One could them send these stars through the pipeline
with those detected in both bands.  One would have to flag the
orphans, but there is a facility for flags.  I'm not sure how
all the software in the pipeline and, more important, in the
database stages, would handle a non-existent V-band measurement.
I could assign V=99.0, the standard value I use to mean "no real value",
but I'm not sure if that would play well with the database.

                                       Michael Richmond