May 8, 2008
Revised May 14, 2008
Arne A. Henden

The final version of the Variable Star Database (VSD) from the compDB team contained 35,816 objects. Of these, 3016 stars had NULL in their magnitude or label field, and were all indicated as variable stars. These are removed from the current update, so a total of 32,800 objects are candidates for this update.

Checking a few of these, we did find some typos and inconsistencies. The VSD has not been quality-checked, so errors are present. The first task was to double-check the work of the compDB team. For automatic photometric update, we have to know that we’re matching the proper star, so we have to be sure that all input objects are correctly identified. Matt Templeton and Michael Sallman worked on matching the object names as given by the compDB team with their appropriate astrometric catalogs. The names were things like GSC 1732-0952, so for the most part were pretty easy to match up since few source catalogs were used. We assumed a 5 arcsec matching radius to guard against bad coordinates or nearby neighboring stars – even this precise match will generate a few errors. Remember, every automatic matching scheme cannot be perfect. We added a further test that the label for the star (its magnitude without the decimal point) had to agree with the catalog magnitude within 3 magnitudes – a crude test to ensure the correct star is found.

After matching, 1810 stars did not match within the 5 arcsec error radius or had other problems of identification. We laid these aside for further review. This leaves 30,990 stars as potential candidates for this update.

We then took these 30,990 stars and matched them against a number of photometric catalogs. Each of these is described below.

1. Tycho2. We converted the matched BT/VT magnitudes into Johnson B,V using the Bessell relations.
2. 2MASS. After a query was run to find matches against the online catalog, these were merged into the final VSD to give J,H,K magnitudes for most of the stars.
3. TASS. We used the patches2 catalog available through VizieR. This gives fairly reliable V magnitudes from about V=8 to V=12, plus less reliable but still useful Cousins Ic magnitudes.
4. ASAS-V. Grzegorz Pojmanski matched the candidate list against his ASAS database and returned the matched stars. Note that ASAS has 15arcsec pixels, so there is a significant chance that a blended object is matched, or a wrong identification is made. While ASAS Vmagnitudes are generally reliable, there exists the potential for an error.
5. SRO. All 207K stars for which we have a minimum of two nights of all-sky calibration (as of May 5, 2008) were searched for matches. SRO calibrates 20x20 arcmin fields, usually from about 10th to 15th magnitudes. On occasion, we go brighter and fainter. However, the master list also includes stars that saturate on given exposures; this gives the potential of a good match, but poor photometry. Beware of bright stars in the SRO dataset.
6. NOFS. All 519K stars from USNO-Flagstaff for which we have at least two photometric nights (as of May 5,2008) were searched for matches. NOFS calibrates either 11x11 or 23x23arcmin fields, usually from 13-18mag. We can go as bright as 8th, or fainter than 22nd. Beware of bright stars in the NOFS dataset as they may be saturated.
7. BSC. The Bright Star Catalog is a compendium of 9110 stars brighter than V=6.5. It includes UBV magnitudes, plus an (R-I) index that can be either Johnson or Cousins. We just used the UBV magnitudes for this update. Note that the BSC does not include errors for the magnitudes, so all such measures are listed with an error of 9.999. The BSC supplement has not been included at this time.
8. SDSS. The SDSS catalog contains ugriz magnitudes for over 100 million faint stars. Its saturation limit is about 14th magnitude. Approximately 1000 SDSS stars were added to the compDB, primarily to fill in other filters/colors than just Johnson V. The Sloan filters were converted to Johnson/Cousins using the transformations given in Jester, et al. (2005AJ....130..873J).
9. GCPD. We did not include the General Catalog of Photometric Data (Mermillod) catalog in this release; this contains about 100K stars with Johnson/Cousins photometry. This catalog is difficult to search. We will include the results in the next release.
10. LONEOS. This photometry file from Brian Skiff was searched to find possible comparison stars. It also does not contain errors for the magnitudes.
11. WBVR. The Russian bright-star catalog was searched and matches converted to the Johnson/Cousins UBVR magnitude system.
12. DENIS. The DENIS southern catalog contains iJK magnitudes in a similar manner as 2MASS. Transformations to Johnson V will be calculated and used in the future, but not for the current update.
13. CMC14. The Carlsberg Meridian Circle catalog contains nearly 100 million stars between -30 and +50 declination. John Greaves derived a transformation to obtain Johnson V from the CMC14 r' magnitude along with the 2MASS (J-K) color index. This was used to fill in the V magnitude for a large number of fainter stars in the compDB.

All other catalogs (such as CCD charts, PEP charts, Zissell sequences, RASNZ sequences etc.) will be added on future releases. The photometric source corresponds with the item number above.

After the above matches, 20,802 stars matched by coordinate and had photometric errors that differed from their chart magnitudes by 0.2 mag or less. These are basically the ones we uploaded at the end of October, 2007. Another 6,695 had errors between 0.2 mag and 1.0 mag. About 461 stars had errors between 1.0 mag and 3.0 mag, with the remainder (428) having larger errors. Of these, the ones with errors less than 1.0 mag are probably correct; the ones with errors larger than this stand a reasonable chance of being mismatched. There are fields like Z Oph, where the 98 star on the modern charts was transposed from an 89 measure on earlier charts or fields like RZ Peg, where the 134 star is not properly identified (or has a wrong position), and matches with a fainter star, etc.

We cannot guard against all of the possible errors! If you discover an error, please use our Chart Error Tracking Tool to see if it has already been reported. If not, please report the error to us so that we will know to fix it for future releases. You must be an AAVSO member or observer and logged in to Blue&Gold to use this tool.

We have therefore included all stars with positional matches within 5 arcsec, with magnitude matches within 3 magnitudes, and with original labels that either looked like a magnitude or an index. This results in a file containing 28,277 stars with revised photometry that is uploaded to VSD. Another 109 stars have photometry that falls outside of those limits, and are likely mismatches (often close doubles). 2,713 stars have good coordinates, but do not have photometry from one of the catalogs listed above. Finally, the 1,810 stars listed at the beginning of this document that do not have good positional matches need to be checked.

The final result is that we started with 32,800 possible candidates; 28,277 stars now have at least V-band magnitudes; and 4,523 stars need to be either checked or new photometry needs to be acquired.