Re: TN-107

[Tass Mailing List <tass@mail.alembic.net>]

Date: Wed, 24 Oct 2007 20:48:55 -0700 (PDT)
From: Adam Kraus <alk@phobos.caltech.edu>
To: Tass Mailing List <tass@mail.alembic.net>
Subject: Re: TN-107


I explored some similar ideas in characterizing the data for a variability
survey of mine, the MOTESS-GNAT survey. I found, as you have, that more
crowded stars tend to have worse photometric accuracy. I attribute this to
changes in seeing - as the seeing changes, stars will spread more or less
of their light into the photometric apertures for other nearby stars
(making their measurements more uncertain than stars which have no other
nearby neighbors to contaminate the aperture).

Characterizing this rigorously would probably require a fair amount of
work, but I found that I could achieve a decent improvement in my
uncertainty estimate with a very basic toy model. To very briefly
summarize, I assume a basic PSF for each nearby star, and then for each
science target, I sum the total expected flux from all other stars in that
aperture. The observed standard deviation is then a surprisingly linear
function of the total flux contribution from all neighbors.

(I assume that it's linear because in most cases, the big effect comes
from having a faint science target next to a bright neighbor. The noise in
that case is almost entirely dominated by changes in the contaminating
flux. If you double the amount of contaminating flux, you'll double the
typical variations in that amount of flux, and therefore in the noise.)

I go into more detail in Section 3.4 of the survey paper:
http://adsabs.harvard.edu/abs/2007AJ....134.1488K or
http://www.astro.caltech.edu/~alk/MG1.pdf

Adam Kraus


On Tue, 23 Oct 2007, Tass Mailing List wrote:

>
> Date: Mon, 22 Oct 2007 15:55:24 -0500
> From: Thomas F. Droege <droege@fastmail.fm>
> To: tass@tass-survey.org
> Subject: TN-107
>
> Hello Everyone,
>
> For years I have been trying to find something, *anything* that would
> improve the scatter of the tass data.  I have looked at everything that
> I could think of.  Does the bad data look better than the "good" data?
> No, when I, for example, take clear nights and compare them with hazy
> nights.  Many other things have been examined with no success.  Finally
> I have found something that seems to work.  You will find it in TN-107:
>
> http://stupendous.rit.edu/tass/technotes/tn0107.html
>
> Andrew are you here?
>
> There is lots more to be done.  I will work first on "how close is
> close".  It is slow going since I am not very clever when under
> sedation.  At least I figure that is as good an excuse as any.
>
> Tom Droege
> --
>  Thomas F. Droege
>  droege@fastmail.fm
>
>