Re: Data reduction methodology for V-I colors

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

On Tue, 31 Aug 2004 13:44:15 -0500, Michael Koppelman
<lolife@bitstream.net> wrote:

>Well I think a normal photometrist would take alternating fields of 
>program stars and standards throughout the night and would in general 
>create one transform for the whole night. If you have non-photometric 
>conditions, you may need to divide the night into segments or 
>something.
>
>The good side of doing this is that we only use calibration stars of 
>high quality i.e. use the best 50 stars from the night (e.g. Landolt 
>and similar) instead of 50 stars from every frame of dubious quality.

The last time I tried this, I ran immediately
into problems with the statistics of small
samples ... zero Landolt stars in several
cases. To get enough calibration stars, I was
forced to use all the Tycho2 sources for the
initial calibration of each image.

This, of necessity, involves using a
different set of calibration sources for
each image.

The Tycho2 magnitudes are not at all precise
in the first place and the I magnitudes
computed from BT and VT are even worse than
the V magnitudes. The magnitudes are bad enough
that it is not at all unreasonable to label the
resulting TASS magnitudes as "relative" or
"approximate" or some such derogatory term
intended to inspire caution. In particular,
V-I is not at all reliable without further
calibration.

My current ensemble calibration code takes this
vagueness one step further, massaging all the
data in a chosen data set assuming only that the
majority of stars are not varying. This is an
iterative process and only in the first iteration
are the Tycho2 magnitudes used at all - after that,
all magnitudes are allowed to float and are not even
constrained to match Tycho2 on average. Resulting
magnitudes are clearly "relative".

In return, it is hoped, the entire chosen data set
can be tied back to reality using those good Landolt
stars, of which there should now be enough even though
there were not enough to calibrate the individual
images directly.

Eventually ...

Andrew Bennett, Avondale Vineyard