Comparison of astrometry made by different catalogs in 3 astrometric calibration regions

It is of interest for astrometric observer to know the accuracy of reference catalogs. Although TYCHO and its derivative, ACT, are known of their high accuracy, they can not be used in relatively narrow fields because of insufficient number of stars. Observer still depends on less accurate, but denser catalogs. To get a glimpse of systematic catalog errors, I present a comparison of astrometry results for the GSC 1.1 and the ACT catalogs in three astrometric calibration regions prepared by Stone et al. from USNO. Astrometry of stars in these regions is assumed to be "exact" in a sense that it is much more (factor of 10) accurate than typical routine astrometry with modest equipment and currently available catalogs. In addition to the above catalogs, I examined a derivative of GSC, prepared by Bill Gray from Project Pluto. This catalog, GSC_ACT, is a recalibrated GSC 1.1 by using the ACT positions as reference.

Images were taken by Herman Mikuz at the Crni Vrh Observatory in the night from May 9 to 10 May 1999 with the ACIT robotic telescope under the same conditions as used for the asteroid search.

Optics19-cm, f/4 Schmidt-Cassegrain, flat field
CCDHISIS-44 1536 * 1024 pixels
Temperature-30o C
Pixel size2.365" * 2.365"
Field of view1 degree * 0.7 degrees
Exposure time60 s

Astrometry was made automatically with fitsblink, using the above mentioned catalogs. After the star lists were stored a match was found with the stars from astrometric reference regions. This match is based on absolute coordinates. Only stars within 3 arc seconds from the catalog stars were considered the same.

A note: If you wish to download image files or star lists, click on links with the right mouse button and choose Save Link As. Image files are compressed with gzip and are approx. 1.4 MB long. For the structure of star lists please refer to the fitsblink manual.

Field name a1238 (star list) a1329 (star list) a1524 (star list)
Astrometric region I J K
Targeted coordinates 12:38:00 00:00:00 13:29:00 00:00:00 15:24:00 00:00:00
Actual coordinates 12:38:02.49 +00:00:14.1 13:29:01.61 -00:00:17.015:23:57.11 -00:00:20.9
No. of detected objects 547 496 795
ACT average residual for ref. stars 0.16 " 0.12 " 0.16 "
No. of ACT ref. stars 9 9 6
GSC average residual for ref. stars 0.34 " 0.45 " 0.38 "
No. of GSC ref. stars 91 92 90
GSC_ACT average residual for ref. stars 0.34 " 0.46 " 0.38 "
No. of GSC_ACT ref. stars 91 94 90

It is interesting to plot residuals of star astrometry for all fields, separately for each catalog. It is very clear that, while the scatter remains roughly the same, there is a large systematic shift for the GSC catalog, which depends on the field taken. The extent of this shift is somewhat surprising for unsuspecting astrometrist. It is clear from these results, that ACT catalog is a preferred solution to GSC 1.1 even for fields with a rather small number of reference stars. However, results from the GSC_ACT seem to be very promising and it could probably effectively replace GSC 1.1. The systematic shift is almost completely removed. Analysis of three images only is not enough to claim improvement for the complete catalog but this partial result at least shows a significant improvement in comparison to the "raw" GSC 1.1. There are other questions, which could be answered by an analysis of deeper and more narrow images without any ACT stars. Above all, it is a question what was the random error of GSC stars fainter than the limiting magnitude of ACT. Unlike the systematic error, this error was not removed in the process of recalibration. Also, some magnitude-dependent systematic error may exist in the GSC catalog that would also be unaffected by the recalibration process for stars fainter than the ACT limiting magnitude.







One would naturally expect that the astrometry is not equally good for bright and faint stars. This is indeed a case, as it is evident from the next image. Only results from the image a1238 is presented here, other two give essentially the same results.

Results for all images and catalogues used are in the following table, both for all stars (limiting magnitude 17.5) and for relatively bright stars (< 15 mag). It is apparent that the scatter of star positions is not dependent on the catalog used, so this is clearly the result of the astrometric process.
ImageCatalogLim. mag.No. of starsAvg. RA res.Std. dev.Avg. DEC res.Std. dev.
a1238 ACT 17.5 404 0.044 0.72-0.1390.79
a1238 GSC 17.5 403 -0.329 0.700.4970.75
a1238 GSC_ACT 17.5 405-0.093 0.71-0.1690.77
a1238 ACT 15 124 -0.0090.310.013 0.26
a1238 GSC 15 124 -0.3950.310.676 0.26
a1238 GSC_ACT 15 124 -0.1420.32-0.018 0.27
a1329 ACT 17.5 416 0.055 0.73-0.2370.64
a1329 GSC 17.5 416 0.289 0.730.7280.63
a1329 GSC_ACT 17.5 416-0.0320.72-0.2410.65
a1329 ACT 15 142 -0.089 0.28-0.0720.22
a1329 GSC 15 142 0.150 0.280.9210.22
a1329 GSC_ACT 15 142-0.1830.27-0.0790.23
a1524 ACT 17.5 719 0.033 0.77-0.3150.68
a1524 GSC 17.5 717 0.282 0.750.5430.67
a1524 GSC_ACT 17.5 719-0.0050.75-0.2410.68
a1524 ACT 15 199 -0.050 0.34-0.1040.22
a1524 GSC 15 199 0.199 0.300.7870.23
a1524 GSC_ACT 15 199-0.0950.310.0280.22

Relation between the measured intensities and the catalog magnitudes is given in the next figure, again for the image a1238. Note that the intensity scale is logarithmic.

intmag.gif To software index