[vsnet-alert 10356] Re: OT 020207+200715

bydra at Safe-mail.net bydra at Safe-mail.net
Mon Jul 21 21:58:19 JST 2008

Brian Skiff commented :-

"The significant proper motion (~0".1/year) and 2MASS J-K color 
of 0.85 both indicate that Kato-san's suggested star is an M dwarf, consistent with the flare-star interpretation."

And indeed the star at the position Kato Taichi noted sits in the red and near infrared colour space populated by red dwarfs, with the proper motion being a nice secondary piece of evidence.

I've been playing with colour diagrams for some long time now, and one formal useful resource I stumbled across a while back is Bessell and Brett 1988


where they took standard photometric stars with not only data in many Johnson passbands, but also of known spectral types, and although I can't remember offhand whether they allowed for interstellar extinction, the combination of widely available quality magnitude data from near UV to mid IR as well as spectra for these stars suggests they are mostly naked eye objects, so except for the very luminous likely near enough to us for extinction to be negligible.  Although their purpose was for deriving transformation coefficients between various passbands and colour indices, the plots in that paper also serve as useful simple astrophysical reference calibration points as the work places some known stellar types in strict colour space locations.

I've been extrapolating, or is it interpolating?, this stuff in terms of red or more preferably r' magnitudes instead of V magnitudes for some time now.  A couple of times I've made some passing use of it and or passing remark to it (eg


make some use of it as a selection criterion)

but likely it'll never get to print or anything, so here's a pocket rundown of how to use it in practice sometimes for variables.

Two graphs are included on this small image


in both instances the X axis is J-K and the Y axis is r'-J, and note the scales are the same.

The top graph is from matching the GCVS iii.dat file of fairly recent vintage (crossmatch recently updated) with CMC14 r' magnitudes at a strict 1 arcsec crossmatching level (viz VizieR batch inquiry) so that only mostly good matches occur.  The J-K data in CMC14 is from 2MASS, so K is in fact 2MASS Ks.

Analogous to plots in Bessell and Brett 1988 a general near linear trend can be seen, but with doglegs of different linearity due to red stars that have TiO bands (see B&B 1988 again).  The greater spread in the figure here is a combination of these objects being variable, with CMC14 r' and J-Ks being of different epochs, whilst interstellar and occasionally circumstellar extinction also broadens the plotted datapoints' extent, with likely r'-J more affected than J-Ks.

At around J-Ks 0.8 to 0.9 a spur due to the red dwarfs can be seen, again analogous to plots in B&B 1988.

Although the J-Ks of 0.85 for the candidate of this optical transient is suitable for a red dwarf, lying about midrange in J-Ks for said objects, it is not uniquely so, and as such does not have to be a red dwarf (although, as Brian says, the proper motion is a nice piece of extra evidence pointing in that direction).

The the object is in CMC14 and although somewhat faint for taht survey the magnitude is still with acceptable limits and likley good to 0.2 or so fo a mangitude, so the derived r'-J of 4.33 puts this object pretty much at the tip of the red dwarf spur in the graph.

Although not much use for blue to yellow or even orange stars, as there is much overlap, and stars of those colours do not need to be variable in anyway, plus intrinsic and extrinsic variables can readily overlap in these colour spaces, these plots can give clues to stars being red giants, red dwarfs, blue or yellow objects with infrared excesses (such as Be stars, YSOs, some RCB and RV Tau with circumsteallar matter, amongst others) which all lie below and to the right of the main trend relative to where stars of their r'-J colour normally lie.

Using the GCVS as a linkage in such a way allows variables of varying types to be used to check where in colour space their members can lie.

And equally, but conversely, the variability types in the GCVS can be assessed for feasibility.

For instance, if the BY Dra variables (red dwarfs of spectral type K and M) are selected from the cross matched subset and overplotted on the main data as red crosses or illustrated in some similar way (not shown here) it can be seen that there are a fairly large number of BY Dra stars in the GCVS that lie in the main blue to yellow trend of the plot, and therefore cannot be BY Dra variables (as "negative" interstellar extincion is not known of).  Further examination reveals these GCVS BY Dra objects appear to be mostly in Taurus with some in parts of Perseus and some in parts of Ophiuchus suggesting they may be rotational variables, ie spotty stars, of the T Tau class, amongst even other objects, and likely to have originally come from specific papers.  Some may even be constant, I haven't checked.  A small project for anyone interested.

Of course, all this would be nicer on a firmer footing, and about a year ago this became available with the release of the first SDSS II Segue data. (sdss.org)

A database table including spectral parameters for over a hundred thousand stars, possibly more, became available and interrogable (via CASJOBS) whilst six months ago Data release 1 of UKIDSS (surveys.roe.ac.uk/wsa/) became public thus enabling a more a deeper cross matching than available via matching SDSS with 2MASS (as the bright end of the former only just overlaps with the faint end of the latter for nonred stars).

Thus a bootstrap calibration could be made using SDSS r' magnitude from SDSS, SDSS derived spectra (not necessarily the most rigorous spectra going due to the moderate dispersions, hence why the exercise is a bit of a bootstrap one, as none of the datasets contain standards) from SDSS I and II and SEGUE, and, J and K infrared apermag3 values from UKIDSS.

This latter survey isn't complete yet, and K is not the same as 2MASS Ks, but UKIDSS J-K is fairly equatable to 2MASS J-Ks, with some inherent offset due to some passbands not quite being the same, some likely more scatter as final photometric reduction is not included in UKIDSS dr1, and generally there looks to be a bit of an offset anyway at the current time between even the two surveys' J mags.

Nevertheless, using UKIDSS J and K magnitudes from the LAS (Large Area Survey) of UKIDSS data release 1 crossmatched at the one arcsec level with the SDSS spparams table gives the graph at the bottom of this image


where x axis is UKIDSS J-K and y axis is r'-J, with r' from SDSS this time.

Few if any red giants appear in this dataset (which due to using LAS, which overlaps SDSS, is mostly for stars from high Galactic Latitudes), and the main extent of the data mostly ends at J-K  approaching 1 (most of the stuff redward of this are objects identified as QSOs and similar), but one important difference exists.

The colour coding used in the plot is based on the SDSS spparam derived spectral classes.

The light blue dots are spectral type A, the pale yellow dots in a black circle are spectral type F, the yellow dots are G stars, the orange dots K stars and the red dots M stars according to SDSS template fitting.  Evidently this isn't exact, as many of these objects overlap at the edges of their range, which can be due to template fitting problems, too low a dispersion, photometric inaccuracy, and various other reasons, but the general trend is there.

And many of the spectral type M objects again appear in a red spike, albeit slightly leftward in UKIDSS J-K relative to 2MASS J-Ks, but very similar in extent in terms of r'-J as in the top graph (both graphs are same scaled).  All analogous to Bessell and Brett 1988 V-J versus J-K plots and similar, well, rotated a bit sometimes.

Further uses of the various SDSS catalogues can be made by say matching the spectral typed stars against the Variable Star Candidates' Catalog from Strip82.  Then the very few stars logged in the former as say A0 that are also templated as candidate BHB (blue horizontal branch) and estimated as having low metallicity (relatively negative [Fe/H] values) or highish SDSS measured redial velocities (over to well over -100 km/s, for example, although other values are not precluded) and flagged as variable in the strip82 variable suspects catalogue, have high circumstantial evidence of being RR Lyr variables.  Unfortunately, however, few if any of this already small number of objects are also brighter than r' about 15.

As usual, this one ended up longer than intended.  Attempt has been made to make it followable by the unfamiliar, whilst trying not to generalise too much such that those particularly familiar with such stuff feel the need to stomp on it with no doubt deserved pertinent corrections and caveats.

Their is not universal utility from this colour space, and as usual things like CVs happily overlap other types, including QSOs, which becomes an increasing consideration when mags of 15 or fainter are involved, but various red, and also various red excess but "blue" objects, give good pointers to true variability class for new or not well known objects.

And for the main GCVS to CMC14 cross linkage, plotting of specific variability classes listed in the GCVS overtop the full dataset will readily reveal any abberant classifications for many variability types (although care has to be taken for borderline cases, but whilst many types of star can undergo interstellar extinction at all passbands, making them appear redder than true, there is no such mechanism that makes red stars look blue in the optical and/or near infrared.  As already noted, BY Dra stars classified as red dwarfs should not be cropping up in the blue part of the plot).  People interested in revising such data can thus generate a short list of candidate erroneous objects.

The whole is, of course, somewhat reliant upon good quality magnitudes, and also upon reliable cross matching between optical and near infrared photometric datasets.  All of it can be done online.  The UKDISS server at ROE and the SDSS casjobs engine both carry "neighbour" batch cross matching options for those not able to cross match locally.



John Greaves

More information about the vsnet-alert mailing list