Re: Throwing away more data ...

[Andrew Bennett <andrew.bennett@ns.sympatico.ca>]

On Sat, 15 Mar 2003 05:38:56 +0000, Tom Droege
<tdroege2@earthlink.net> wrote:

>Andrew,
>
>As I point out in TN-90, all this data was reduced using sky flats.  I now 
>understand that this produces an incorrect result due to the gradient in 
>the sky.  I am not sure that it is possible to unfold this error once made 
>in the data reduction.

I used to believe this too ... with a passion.
But I don't anymore.

As I showed in TN92, it made very little difference
if one used sky flats, lightbox flats or no flat at
all after one subtracted off a sufficiently high
order spatial polynomial correction derived entirely
from the data, after the fact and without any need to
go back and reprocess.

The order of merit was:
current sky flat best
out-of-date lightbox flat second
no flat at all worst.
From which I extrapolated the guess that a current
lightbox flat might be best of all, if we had one.
But the difference was small compared to the image-
to-image variations and minute compared to the day-
to-day variations. That's why I am currently
concentrating on automated methods of throwing out
"non-photometric" images instead of continuing to
ride my PSF fitting hobby horse.
>
>There is hope for the future.  I have 10,000 or so images to reduce once we 
>get the pipeline polished up.  By the time Andrew finishes his 
>spring/summer/fall job, I should have a nice data set to keep him occupied 
>for the winter.  I should also be able to go back and reprocess the earlier 
>data.  Then we will have a really large data set to study.

As I now claim that reprocessing will make no
perceptible difference, I say you already have a really
large data set to study! Send it along and I will have
a go at finding a long period variable in it!!
 
Andrew Bennett, Avondale Vineyard
Andrew Bennett, Avondale Vineyard