Re: Some Real Work

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

On Wed, 5 Nov 2003 13:04:27 -0600, Tom wrote:

>Let me try to state the problem I see:
>
>1) The data is as good as it can be,
Not quite, but pretty much.
> is useful,
Yes!! Anybody who doubts that should reread this 
summer's posts.
> and the survey should be
>continued.
Obviously.

>2) The data is not useful at it's present quality level.  The survey should
>be stopped until it can be improved.
Twaddle, balderdash and poppycock!!

>3) The data will never be useful based on the first year's data.  I should
>buy a junk Rolls and work on it for the rest of my life instead of wasting
>time on tass.
Sir! Junk and Rolls should not be used in the
same phrase.

I did once have the pleasure of watching an
elderly Rolls race at Snetterton. They cleared off
the XK120C's and the remains of an early BRM (it had
started out with 16 cylinders ...) and had a "race" 
for Classic Cars. The owner had thrown out his 
chauffer for the day and drove around at a
stately 80 mph for the requisite number of laps
obviously having the time of his life. Fun. But, I
think rather limited as a life's work?

> ...
>For 2) I am out of ideas. OK, the possible exception is trying local
>ensemble photometry.  I will have a go at this.  
As I may have mentioned before, I have tried this.

I gave up on the simple version where one draws a
circle round the star being analyzed and uses all
the stars inside the circle as an ensemble. The
number of images with flag-free measurements of all
the required stars turned out to be small and using
a partial ensemble, different from image to image,
did more harm than good.

My alternative procedure was to fit spatial
polynomials to the residuals over each image. This
is somewhat similar in effect to the local ensemble.
The gory details of this rather complex procedure
are in TN94. There, I claimed to get the errors
down from 0.0197 to 0.0081 mags in V-band  and from
0.0211 to 0.0060 in I-band after throwing out
half the images and almost half the calibrating
sources.

As has been discussed before, this improvement is
much greater than Tom managed to achieve by throwing
out poor images. I have not had time to try to track
down the reasons for this.

Earlier this summer, there was some discussion of the
need for much better catalog magnitudes than Tycho2 to
improve TASS bright star measurements: I was unable to
understand the arguments.

Andrew Bennett, Avondale Vineyard