Re: aperture photometry and sky variations

[Tom Droege <tdroege2@earthlink.net>]

Yep, I was wrong here.  A bright cloud just stands the star on a 
pedestal.  The aperture should still do the job.  I can see no reason for 
the scheme not to work for uniform cloud cover.  When the ensemble 
photometry is done, it should not care that some stars on one image were 
actually brighter than others on another image because they were sitting on 
clouds.

I worry that this is asking too much for the case where the clouds vary 
over the image and so have set max_skysig relatively low for my 
reductions.   If I can see obvious sky streaks from clouds in the image, I 
throw it away.

The reason is that I worry that the situation is not so simple.  If all the 
photons from the star make it to the image, plus a contribution from the 
cloud, then the scheme is OK.  However, in the likely case that some of the 
photons from the star get dispersed in the cloud, then it will be measured 
low compared to a star of the same brightness sitting in a clear patch.  I 
think the ensemble photometry will be confused if an image contains stars 
in clear patches and stars peeking through clouds.  Comment Michael?  Others?

Tom Droege

At 10:13 PM 4/4/03 -0500, Stupendous Man wrote:

>   Tom wrote:
>
> > Once one does aperture photometry, then if there is a bright star sitting
> > in a bright cloud, then one gets the measurement wrong, as the background
> > aperture gets the bright cloud, and that is wrong.
>
>   Well, it depends on the size of the bright cloud compared to the
>size of the star.  If the bright cloud is much larger than the star,
>so that the area "under" the star is about the same brightness
>as the area immediately surrounding the star, then aperture photometry
>can still do a decent job; one must simply define an annulus surrounding
>the star, measure the brightness there, and subtract it from the
>light inside the aperture.  The current Mark IV pipeline performs
>this sort of local sky subtraction.
>
>   However, if there are small bumps and holes in the background
>level, of roughly the same size as (or smaller than) the PSF,
>then one will indeed make big errors in the photometry.
>
>   The routines in the version of the pipeline Tom is using try to
>remove large-scale, gentle gradients with a simple linear or
>quadratic model fit to the entire image, _and_ also use a local
>sky annulus around each star during the aperture photometry.
>It's not foolproof by any means, but it can do a decent job in
>some situations.
>
>                                    Michael