Re: [TASS] Flat fielding

[Andrew Bennett <andrew.bennett@NS.SYMPATICO.CA>]

Various people wrote:
>>>  What *does* worry me is that the mean magnitudes and
>>>colors in tenxcat are not asymptotically approaching
>>>the Landolt magnitudes.
>>Um. Er. They've been fed in twice and still fail to
>>emerge at the far end? Statistician needed!
>  I don't understand your statement.  Expand a bit, please!

As I understand it, "known" magnitudes are used to fix
the initial scaling.

The same set of "known" magnitudes are then used to
take out the N-S gradients inherent in the method of
flatfielding.

This should ensure that the magnitudes that emerge from
the output end of the processing pipe are well and truly
nailed down.

However, in spite of these best endeavours, the final
product shows bias errors when compared back to the
same set of "known" magnitudes.

This is the kind of thing that results from any one
of a large number of selection effects; the study of
this sort of nastiness calls for the professional skills
of a statistician.

Just as you need a statistician to comfort you that the
upward curl at the high magnitude end of the TN0056 plot
of tenxcat vs Landolt V magnitudes is perfectly normal
for such a comparison. [I'm not sure why it is so much
less obvious in R and I, though.]

Andrew Bennett, Avondale Vineyard, Nova Scotia, Canada.