A New Record

["Thomas Droege" <tdroege2@earthlink.net>]

Last night produced 1,928,955 measurement pairs.  I think this is a new
record.  One to be broken soon, I think.

Some of you are saying that tass emphasizes quantity over quality.

This is absolutely not the case.  

I spent most of the last several years setting up quality control for this
experiment.  There are very carefully determined limits built into the
software which optimizes the goals of the project.  This has all been
extensively discussed on this list.  It should not be new to any of you.  

Having said this, you should all remember that this is a *quantity*
experiment.  We are trying to get hundreds of measurements on each star.
While we have set our limits where we can achieve this goal in a reasonable
time. There is nothing obvious that can be done to improve the data that I
have not already done.  

Some of you say things like "only take data on photometric nights."  Well,
it  is not obvious that great nights produce any better data than good
nights.  I have studied this. There are other things that are more
important.  I observe that the tass quality level is similar to that
achieved by Pojmanski at a super location in Chile.  How can this be?  

First we are both using front side chips.  The gate which covers part of
the pixels has a different transparency from the rest of the pixel.  The
result is that the sensitivity depends on where the star happens to be in
the pixel.  By necessity if you want to cover the whole sky in a lifetime
with a finite number of telescopes one needs relatively big pixels.  This
means that most of the photons from a star land on a very few pixels. 
Possibly only one. So depending on where the photons hit a pixel they are
measured with a different sensitivity.  Defocusing does not help much, it
just loses faint stars.  This contributes a base line error of order -.03
to 0.04 mag for the brighter stars.  For the fainter stars we are limited
by photon statistics.   Roughly the 1 sigma error is of order 0.04 mag for
the brighter stars up to mag 10 or so.  Then it increases exponentially to
where it is about 0.1 at mag 12.  Arne, if you are reading this it may be
that you will not be able to take any better measurements with your Mark IV
at NOFS than I can take here in Batavia.  We shall see.  You might want to
buy a couple of back side chips to do the calibration survey.  

Next there are the facts of life about photons.  I have seen comments like
"ASAS data is no good above mag 12".  This is just not a sensible thing to
say about TASS and ASAS data.  We are both limited by photon statistics at
the fainter levels.  One just has to accept that the error bars get larger
and larger as the signal sinks into the noise.  Whether or not the data is
"good" depends on what you are trying to do with it.  There are strategies
that one can make to improve this.  Take longer exposures to improve the
fainter stars and let the bright ones saturate.  Mostly this needs to be
done with the optics, and that is hard.  One needs Schmidt optics to avoid
long focal lengths and then there is no room for the CCD at a reasonable
size.  One quickly needs hundreds of telescopes to cover the same sky as
TASS or ASAS.  We (tass) have already done something pretty clever in our
optical design.   Our lenses are much better for this than the camera
lenses used by ASAS and other similar surveys.  One more thing that helps
us to compete with his great location.  We greatly improved the performance
of our lenses by limiting their bandwidth.  It took me a while but I
finally found someone who agreed with me on this and so Elliot Burke was
able to design better and cheaper lenses for tass.  

Some of you seem to think you see "bad" nights in the tass data.  I also
"think" I see bad nights.  But I am yet to prove it.  I have done many
hours of computation to try to select data that might be better.  

Here is an example of one of the things I did. Let us suppose that there
are bad images. Bad images should result in measurements away from the mean
for fixed stars. Most stars are fixed, so most stars in the data base are
fixed. If we look at all the stars in the data base and select points away
from the mean for each star, then if the above theory is true  more of
these points should be from "bad" images than from "good" images. Now take
all the bad points and sort them by image. Compute the percentage of "bad"
points in each image and use this to make a cut on all the data. That is
throw out all the images that have a high percentage of bad star
measurement by the above definition. Now look at the quality of the result. 

I have done this and no cut I can find improves the quality significantly
as measured by a mag, sigma mag plot.  Some of you may have a better test. 
I would like to know a better test for the "quality" of a big data set like
this.  

With as many measurements as tass has taken so far, there will be many with
errors that look bad.  A back of the envelope computation says 3 million or
so 3 sigma off the mean.  Even a hundred or so 5 sigma off the mean.  Since
most of the stars are faint, that means that we should see millions of mag
12-13 stars that are 0.3 to 0.45 mag away from the mean.  This is not bad
data, this is just the facts of statistics.  We are making measurements
where the number of electrons sitting in the pixel well matters.   

I am very much aware of this as I try to find variable stars in the data. 
Some stars are "lucky" and get more than their share of bad points.  These
show up in any analysis.  When you look at a lot of data you find a lot of
noise and the noise can sometimes look meaningful.  The only solution is
more data.   ;^)  And thus we are back to the subject of this message.  I
am working on it.  

Anyone using data should understand the errors associated with the data. 
Pojmanski has discussed the errors in the ASAS data in his papers.  We will
discuss ours when we get the paper done.  Meanwhile, there is plenty of
information in the tech notes to give you an understanding of our errors.  

Tom Droege