Re: Photometry

[Arne Henden <aah@nofs.navy.mil>]

Tom Droege wrote:
> This brings up the question as to whether I should also save the .pht, 
> .coo, and .ast files.  They are relatively small.  Still, they can be 
> reconstructed perfectly.  For study, one can just start with a test set 
> of images, run then through once, and then play using the .coo, .pht, 
> and .ast files as needed.  There are also the .clt files.
> 
These files are all small compared with the original image, and compress
easily.  However, they are not often used, and it might be easier to
delete them and recreate them if necessary later.

> I figure the following probabilities:
> 
> 1) By far the most probable.  We want to look at a particular star and 
> see if there is anything funny in the image.  For this purpose the files 
> I am saving are the best.  You don't have to process them again to get a 
> good looking image.  They look as nice as we know how to make them.  One 
> can get pretty close to a star position.  I know what I have been doing 
> when I am looking up "flasher" events.  I get out the image and position 
> the cursor on the coordinates.  If I see *anything* near it (like a 
> satellite track) I assume that it is the cause of the problem and 
> discard the event as a possible flasher.
> 
My point is that you cannot with certainty look at a processed image and
tell whether an artifact was present in the raw image, or placed there
during the calibration process.  Certainly satellite tracks, asteroids,
etc. can be examined with either raw or processed images, so you are correct
in that regard.  It is the unusual event which may require the raw image.
Whether you want to ignore the unusual event or not is your decision.


> 2) A possibility.  In the future we have a better reference catalog for 
> photometry.  We want to reprocess the images to see if this improves the 
> error.  For this the .ast, .coo, .pht, and .clt files save time but we 
> can always regenerate them exactly from the corrected .fits files.
> 
As Michael R. indicated, new astrometry or photometry reference catalogs
can be used without returning to the actual image, once you have done
the initial step of star finding/extraction.  However, if an error is
detected, then you may want to go back and reprocess your images.  That
might be just a different starfind/extract algorithm; it might be an
additional correction beyond just flatfielding; or it may mean new flats
entirely.  Only the latter step would require the original, raw images.
But if these are not available, you have automatically prevented any
improvement.

> 3) Zero possibility.  In the future someone with a time machine goes 
> back and gets us better flats for a particular date and we want to use 
> them to produce a better corrected .fits image.  I do not see any 
> possibility of this happening.  I don't know where we could get better 
> darks or flats in the future than the ones we have at the time the data 
> is taken.  I am open to a specific example where there is anything we 
> might do in the future where the corrected .fits files would not be 
> satisfactory.
> 
Again, an example here would be finding that your flats for a given night
were incorrect.  Perhaps new dust fell on the CCD.  Perhaps the flats
themselves were corrupted.  Sometimes you don't notice these things until
days or weeks later.  You may be able to use flats from nights before
or after to "save" the night in question.  Simple reprocessing of a
bad night is trivial if you have the original raw images;
it is far more difficult if you have to take processed images and
back-out the processing steps.
Arne