No valid WCS values were available in the header, so I manually entered center coordinates of M5 cluster and pixel sizes of 7.7 arcseconds, calculated from assumed focal length of 400 mm and physical pixel size of 15 micrometers. To detect stars and make catalog matching, a program called fitsblink was used. Only pixels which are both inside the radius of 3 pixels from the brightest pixel of a star and above 3 sigma of the local background were taken into account for the calculation of positions and magnitudes. In the catalog matching, maximum residual of 5 arcseconds was allowed. After making the match with the GSC-ACT catalog, the following information was revealed:
| Description | m5v.fts | m5i.fts | unit |
|---|---|---|---|
| Center right ascension | 15:18:23.50 | 15:18:24.48 | hour angle |
| Center declination | +02:09:28.6 | +02:09:23.7 | degree |
| pixel size | 7.74 | 7.66 | arcsecond |
| Rotation | 89.73 | 89.68 | decimal degree |
| No. of stars matched with the catalog | 146 | 198 | |
| Average residual | 1.66 | 1.37 | arcsecond |
| Deviation from average residual | 1.05 | 0.94 | arcsecond |
| Smallest residual | 0.05 | 0.02 | arcsecond |
| Largest residual | 4.82 | 4.25 | arcsecond |
| Saturated objects | 0 | 3 |
No positional dependence of residuals could be observed, so no attempt to use higher order plate constants was made.
Size distribution of residuals is given in the figure below:
Next, for identified stars the dependence of residuals on magnitudes is given in the next figure. Some explanation may be necessary here: points correspond to individual measurements. For each magnitude bin (10-11, 11-12, etc) average magnitude and average residual were calculated. These pairs of values are used as coordinates of points which are connected by lines. For the curious, here is an AWK program which calculates these values from the fitsblink output files.
awk 'NR>1&&NF==10{i=int($10);n[i]++;r[i]+=$9;s[i]+=$9*$9;m[i]+=$10}\
END{for (a in n){rs=r[a]/n[a];print m[a]/n[a], rs, sqrt(s[a]/n[a]-rs*rs)}}' \
m5i.dat | sort -n
Fitsblink is not particulary ambitious about magnitude measurements. Nevertheless, the relation between the GSC magnitudes and instrumental magnitudes extracted from the two images is given in the next image. Note that the filters do not match the one used in GSC so it is difficult to get some idea about the photometric accuracy. Also, multiple measurements of the same stars would give a better feeling about photometry. Instrumental magnitude was defined for this purpose as
where I is the measured intensity (AD count).
Finally, let us inspect positional dependence of magnitude differences (GSC - instrumental) which would not be surprising due to simple flatfielding used and due to the presence of the unresolved background of stars in the cluster. The positional variation of background for the image m5v.fts is seen in the following figure. For each star the median value of pixels lying between the circles with the center at the star's brightest pixel and the radii of 3 and 10 pixel was determined. This value is plotted as function of x coordinate. As one might guess, we get something similar for the y coordinate, so this is not plotted. There is a strong increase of background around the cluster position.
This variation is not so prominent when we plot the difference between the instrumental and GSC magnitude. This difference may be obscured due to other errors. I did not try to find out what is the coalition of the detached (from the main line) stars on the right hand of next image. Maybe they really are just of different color. Also, GSC magnitudes are not exactly known to be gloriously accurate, so it is difficult to say without further investigation.
