Continuing saga of flatfielding

[aah@nofs.navy.mil]

Chris wrote:
>What is the rate of change of flat fields?  What makes them change?  Dirt
>on the lens?  Do we know?  It seems that if you could take a good flat with 
>a flat field device known good to <<1% then if the rate of change were close to
>zero then you could simply apply the known good flat retroactively.  Trouble
>is I don't think we know if the rate of change of flat fields is near zero
>of not.
  There are hardware-related flatfield changes, such as new dust and crap
on both the CCD and the lens, different focus, etc.  Then there are problems
with the sky-derived flatfield itself: improperly subtracted stars, gradients
in the sky, etc.  I think both can change on a nightly basis, though you
can probably go through good periods where everything is stable for weeks.
Who knows for sure?  I take flats every night, and have been thankful on
several occasions when a new feature appeared between nights.
  Let me give you a straw-man system.  Use a very simple flatfield system,
such as a light box, for every site, rather than sky flats.  As long as you
mount this light box in a consistent manner, then whatever gradients it
imposes on the flat will be the same from night to night, and it can be used
to remove dust and other effects.  Then have a better flatfield device that
is mailed from site to site that is used to find the gradients in the local
light box.  This solution removes the sky flat requirement, permits a cheap
local solution for the dynamic effects, but also does proper gradient
removal.
Arne