Return to Home Page
From John Greaves:
On 6 Aug 2003, Bill J Gray wrote:
> Hi John,
> Presumably, CMC-13 r' magnitudes could be turned into
> "real" Rc magnitudes with still greater accuracy. You'd get
> the same sort of formula as before, but the .6 coefficient
> would be replaced by something a lot smaller.
[Incidentally it's coefficient 0.5967 and intercept -0.0306 for the purists: I
felt safe rounding those to .6 and -.03]
Hi Bill
Well, you'd think, wouldn't you.
Nope. Abysmal fit. It seems that, ironically, the very small distance between
r'_CMT and Rc is the problem, as the scatter is relatively immense. That is,
the errors are significantly large relative to the values. All the red minus
red plots versus colour are badly scattered... ...remember Ru-r' versus J-H
was a big black parallelogram!
Contextually, the correlation coefficent on the V fit is 0.9, on the Rc fit
it's 0.1!
On the other hand, it could be related to the source TASS Mk III Rc I'm using.
On the other, other hand, one of the original plots I sent out had r'_CMT as a
function of TASS Mk III Rc with very good fit, if you check back.
Similarly, we can get a simple noncolour corrected
Rc^ = r'_CMT - 0.2 +/- 0.118
where 0.118 is the standard deviation on Rc^-Rc
For ~13000 objects in a 0 to 1 degree declination circumsky strip, when r'_CMT
is between 9 and 14 (best linear fit of .996r'_CMT - 0.155 +/- 0.118 with
correlation coefficient on the fit of 0.99 according to the automatics, but as
you can see, simply adding 0.2 gives just as good a standard deviation whilst
retaining a 0.00 mean and median).
I don't think that's too bad for a noncolour corrected term, probably best to
avoid objects that are very red though, if possible.
For VSOers, I've got to mention that any individual magnitude from any of these
astrometric surveys has the problem that it could simply be wrong due to being
a blend, and therefore artificially bright, as this is not too great concern
for the catalogue compilers. Using many comparators and averaging can get
around this, but this is not standard VSO practice. Simply try and check an
image of the star via aladin or similar first, in that case, or even upon your
own images if a CCDer.
Looking again at the Rc/r' plot, the problem may be that there is a dogleg
around J-Ks 0.6, but the scatter is so large, I can't tell for certain.
Or maybe the wider r' takes in an important absorption line or two that Rc
misses, screwing up the 'comparing continuum fluxes' assumption for two such
overlapping passbands. With V and r'_CMT we are comparing the difference
between two totally separate fluxes. With Rc and r'_CMT it's more of a
relative flux thing. Maybe that's enough reason in itself for errors being
more significant in one than the other.
SUMMARY on Rc
if quoting only to one decimal, Rc=r'_CMT-0.2 (+/-0.1) to mag 14 or so.
> I had even lower expectations for turning UCAC-2 instrumental
> magnitudes into "good" V or R magnitudes, and am not too
> surprised that you've had less luck there... &*!@ it; it would
> have been nice, at least, to have good photometry over the
> full range of UCAC-2.
I'd increasdingly rather take Norbert Zacharias' point of "don't use it", but I
can see how it's tempting to folk.
I can't get any fit to solve the arcing of UCACn when compared to any other red
magnitude (even compared to USNO A/B red)!!! Whether using increasing
polynomials, power law or exponential or whatever, there seems to be no
improvement over using a linear fit. So, folk more capable could have a bash
at cleaning that systematic trend if they wished by finding the proper
nonlinear fit, leaving only the inherent scatter to angst over, but I
personally put that in the 'more trouble than it's worth' category.
Incidentally, noone has picked up on the point of how this systematic relative
nonlinearity in UCACn red mags versus other red mags (it's there in both 1 and
2, so is probably instrumental) reflects on the wider issue of reported
unfiltered CCD mags, especially if they've been somehow "transformed" to be
'like' filtered red mags.
Cheers
John
Return to the Daisaku Nogami
vsnet-adm@kusastro.kyoto-u.ac.jp
