Return to Home Page
The following message is from John Greaves: === Some little while ago I checked ucac1 red magnitude against USNO B1.0 R. Although there was serious scatter from the latter, the fit seemed somewhat non- linear, the assumption being that this was a natural consequence of the former being unfiltered. This seemed somewhat interesting on three counts: i) within the context of variable comparator sequences, as some people sometimes use ucac1 red adjusted to "V" via a colour term from USNO Ax.0 blue minus red. ii) unfiltered CCD photometry of variable stars, especially larger amplitude stuff, that is if any CCD owners bother with said iii) some have wondered on utilisation of ucac2 red magnitudes in tandem with 2MASS colours for calibration purposes and/or colour conversions of one kind or another, and similar ucac2 now extends to hemisphere's northern: ad.usno.navy.mil/ucac CMC13 ranges from -3 to +30 degrees declination: http://vsnet.ast.cam.ac.uk/~dwe/SRF/cmc13.html TASS Mk III tenxcat has been available via the CDS for some time (see the readme there for references and urls): http://cdsarc.u-strasbg.fr/cats/II/230/ They all have something in common... ...they all contain a red magnitude of one sort or another. Details of said magnitudes, along with the _intended_ remits of the respective catalogues, can be found via the above urls, but in short ucac2 - wide passband lying between Johnson V and Cousins R - and astrometric catalogue in no way intended to be photometric r'_CMCT - SDSS r' filter used on drift scanned CCD, but calibrated on the Vega system (via Tycho2) rather than some obscure halo subdwarf F star or 4 as in the SDSS, although no internal colour information exists - an astrometric catalogue with a more serious than usual approach to the "bonus photometry" produced TASS Mk III - Rc filter used on drift scanned CCD, calibration via Tycho - a multicolour photometric survey Although rigorous analysis makes use of primary standards and transformation coefficients and the like, if you use big enough samples then simple scatter plots can be quite informative. So, simple scatter plots of red magnitudes are attached. Although the sample sizes are in the thousands or tens of thousands, this is still relatively small compared to the tens of millions CMC13 and ucac2 each contain. Now, the UCACn readmes state that the red magnitudes are in no way meant to be photometric, and although they may well be within 0.1 of a mag of each other within an observing zone, variations in observing seeing can lead to this being as high as 0.3 between such zones. First I'm going to match UCAC2 red against CMC13 r'_CMCT [The fist intent was to plot ra and dec differences out of interest for these two, but along the way I seem to have forgot this and rounded the decimal degrees to four places at one point for another reason, and as these two catalogues appeared to mostly have exactly the same positions to four places for the first few hundred I scan viewed, it appears I'll have to start again from scratch for that]. The strip of sky used is from 0 to +1 degrees declination and 0 to roughly 140 degrees RA, a region selected to allow overlap between all three catalogues. ucaccmcr.gif shows the plot of REDucac2 as y axis versus r'_CMCT as x axis for 64000 objects. ucdiff.gif shows the differences, REDucac2 minus r'_CMCT as y axis against r'_CMCT as axis uccolour.gif utilises the inclusion of 2MASS photometry in the UCAC2 catalogue to attempt an assessment of any colour dependencies, such that the y axis is again REDucac2 minus r'_CMCT and the x axis is 2MASS J-H colour for the same objects (J-Ks was not used as Ks can be in excess due to circumstellar matter). Their is evident nonlinearity. However the situation isn't alinear. The situation gets worse at fainter magnitudes. It seems from the J-H plot that all colours and all differences are somewhat evenly distributed, with no clear trend (though admittedly I'm not entirely sure what that big splob in the colour diagram denotes). Which is the nonlinear? Let's try a kind of bootstrap. There's no massive standard set of stars to test against, so let's have a look at the filtered CMC13 r'_CMCT stuff in comparison to the filtered TASS Mk III Rc stuff, and see if they can self qualify each other. There's ~14000 common objects in that sample. The TASS data has a faint limit of 14 in comparison to a faint limit approaching 17 for the other two catalogues, so there are less stars to match anyway (approximately a third of a milliion stars in tenxcat, not all having Rc data). tasscmcr.gif has a linear fit with a high correlation coefficient of 0.98 (though some would prefer 0.99 or even higher for certainty). tcdiff.gif also suggests linearity between the two datasets, the plot showing Rc minus r'_CMCT as y axis versus r'_CMCT as x axis and having a mean of -0.20, a median of -0.19 (no skew) and a standard deviation of 0.15. tccolour.gif shows Rc minus r'_CMCT versus V-Rc from the TASS Mk III data. There appears to be colour dependency in that the redder the object the more likely it is for Rc to be brighter than r'CMCT at any particular magnitude. UCAC2 may well be the photometrically nonlinear then. Time for some Rc-ing about. Only ~8400 common UCAC2 and TASS Mk III Rc objects were gleaned between 0 and +1 degrees declination and 0 and 180 degrees RA. tassucacr.gif shows Rc as x axis and REDucac2 as y axis. It looks fairly linear. tudiff.gif shows things aren't that linear though, with a plot of REDucac2 minus Rc as y axis versus Rc as x axis. Mean REDucac2 minus Rc is 0.15 whilst median is 0.13, a slightest of skews, with standard deviation of 0.18. A linear fit of 0.98 correlation coefficient for REDucac2 versus Rc can be obtained for this lesser magnitude range (8 to 14) : REDucac2 = 0.96 Rc + 0.64 Inspection shows that both these plots recapitulate the equivalent cmc13/ucac2 plots of above for magnitudes down to 14 in those plots, where the differences equally appear not to be too great. tucolour.gif shows Rc minus REDucac2 as y axis against TASS Mk III V-Rc as x axis, and that it can be compared to tccolour.gif, and again the colour dependency is recapitulated in that plot for objects of magnitude 14 or brighter. As REDucac2 and r'_CMCT, though both relatively broad passbands, do not overlap greatly, there seems to be something of a slight colour term in the TASS Mk III V-Rc, although I could readily have mistaken it and the problem could lie elsewhere, or there be no problem at all. The relationship appears linear, but whether it strictly needs to be flat, given the differing passbands, I do not know. It is interesting though that REDucac2 lies between V and Rc, so they are not completely unconnected. END BIT As the UCAC2 readme strongly suggests, UCAC2 red magnitudes are not meant for photometric use, but cross identification double check work. They either shouldn't be used or at least should be used only strictly between magnitudes 9 and 14 as last resorts. The CMC13, and the final CMC due next year, which will span -15 to +50 declination, seems to have an interesting bonus of possibly useful photometry, at least for ensemble reductions (individual stars would have to be checked via images to ensure they are not blends unresolved in CMC13). Though in no way a standard, nor on a common system, and unchecked here at the faint end, it may have some use in reducing error ranges during calibrations and/or scatter between observers. Again, I'm not suggesting it as a standard in anyway, but for those more photometrically expert than I, there may be potential uses. However, it seems likely that SDSS r' type filters would be needed in such a case, as evidently unfiltered CCDs cause problems, and although nonlinear solutions are just as applicable as linear ones, the increasing increase in scatter at the faint end is worrying. PERSONAL VIEW Those hoping to, or already, using UCAC photometry, don't. If you want to experiment with 2MASS colours and red magnitudes or similar, play with CMC13 instead. Hopefully it will be on vizier soon, too. However, it has a narrower declination range than UCAC2, as will the final offering. [Interestingly it seems far denser in the ranges where they do overlap (current UCAC2 ~48 million stars from -90 to +40-50 degrees declination, current CMC13 ~ 36 million stars from -3.2 to +30.2 degrees declination]. Cheers John John Greaves PS I don't really know where I was intending to go with this one. Just look at the graphs, they say it all really, whatever it is that is to be said ;^) Hopefully the 'grown ups' may have something to say about the ucac2 cmc13 relative astrometry, and even the cmc drift scan in general, as it seems an impressive unsung "little" thing that's sneaked out quietly.
Return to the Daisaku
vsnet-adm@kusastro.kyoto-u.ac.jp
