be careful running cameras in "tracking mode"

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

  Tom has written that his plan is to run the cameras in a 
"tracking mode" this season: pick one region of the sky and
follow it for several hours as it moves from east to west.

> 2) Look at where I can point the telescope.  Then look at a table of tiles
> to see which it the best tile to take.

> 3) Point the telescope at the tile and take a set of images to make a
> flat.  This is done by offsetting the telescope slightly in all directions
> to get sky flat data.

> 4) Point at the tile and follow it as long as possible taking images.  For
> tom1 this is two hours.  For tom2,3,4,5 it may be 3 or a little more
> hours. This might produce a 30 image sequence at 200 second exposure for
> tom1 and 45 images for the rest.

  I just want to mention an effect that _possibly_ might appear
when running the cameras in this mode, rather than the previous
"look at different regions of the sky all the time" operation.

  The reduction pipeline has several options for making master
flatfield frames.  One option is to use every "object" image throughout
the night, combining them via a median operation.  If you look at different
regions of the sky, then a typical pixel will usually see the blank
sky background, and only occasionally, in one or two frames, see a star.
One can easily discard the high data values caused by the star and
create a meaningful flatfield using the blank-sky values.

  If, however, the camera stares at the same patch of sky over and
over and over, then some pixels will see stars most of the time.
Even a robust median filter will not remove the pixels with high
values, and the resulting "master flat" frame will have lots of
little star-sized residuals in it.  Using this malformed flatfield
frame to correct all the raw images will add some noise to the
results.

  The pipeline _can_ be told to use only a subset of all the
frames to create the master flats.  It sounds as if Tom is well
aware of this issue, and is planning to do it.  Fine and dandy.

  Since the season is just starting, and this is a new mode of
operation, I thought I'd mention this.  Tom, if you glance at
the master flatfield frames produced by the pipeline, you'll be
able to see immediately if there are star-sized spots at any
significant level. 

  By the way, I'm currently revising bits and pieces of the pipeline
software.  No big changes, but some fixes to known problems, and
a few enhancements.  I'm testing the new code on several sets of
Mark IV data to make sure that it works properly.  I'll send out
a message when the new versions of the software are ready for use.
One of the new options will be to include "orphan" detections
(stars seen only in one passband) in the final output.  

  Tom, I'll have a set of special instructions (no big deal, just 
a few lines) describing how you might most easily merge the new stuff with
your existing code.  I know that you have figured out exactly which
parameter values make the most sense for the current data, and I
want to make it as easy as possible for you to upgrade without
having to re-edit a bunch of parameter files.

                                             Michael Richmond