Assessment of image quality

In order to measure the stability of the image scale and PSF across the field and make a rudimentary assessment of the performance of Lucky Exposures images in astrometric measurements, the observations used for Figure 5.16a were re-analysed.

The $60$ best exposures from the same dataset on $M13$ used in Chapter 5.5.3 were separated into two groups of $30$. The exposures in these two groups were observed during two separate time windows, so the atmospheric effects should be uncorrelated for the two datasets. The exposures in each group were shifted and added together to give two independent images of the field in M13. I used the same approach to accurately measure the positions of the stars in these two images as was used for measuring the location of the brightest speckles in individual short exposures - the two images were filtered using the modulation transfer function of a diffraction-limited telescope (shown in Figure 5.6) and resampled, and the peak pixel in the resulting stellar images was taken as the position of each star. The relative star positions calculated for the two independent datasets were compared and found to agree within $6$ $mas$ for eight of the brightest stars, without accounting for changes in plate scale or orientation. The stellar magnitudes agreed within $0.02$. Clearly for astronomically useful measurements the optical distortions in the instrument would need to be accurately determined, and any shift in the stellar positions due to limited charge transfer efficiency would also have to be characterised. Given good instrument characterisation, the accuracy of the astrometry and photometry would improve substantially with increased observing time, potentially allowing accurate measurements of globular cluster velocity dispersions and photometric variability studies.

Some of the Lucky Exposures images generated from data taken using L3Vision CCDs were found to show evidence for smearing at low signal levels in the direction of CCD serial transfers. The dependency of the smearing effect on both the position in the image and on the signal level gave weight to the hypothesis that problems with charge transfer efficiency might be to blame. In order to investigate the smearing effect in more detail I chose an image of M15 which was quite badly affected, taken from our observing run in July 2002 through the $810$ $nm$ HiRac I filter at the NOT. Figure 5.22 shows a region around the core of M15. The cluster centre is marked by a green cross toward the right-hand side (as determined by Guhathakurta et al. (1996)). The image has been contrast stretched to highlight some of the fainter stars. Most of the stars show some evidence of horizontal smearing, particular the fainter stars toward the right-hand edge. Cross-sections through two stars toward the lower right-hand corner of the image are shown in Figure 5.23. The horizontal cross-section along line C shows a long tail to the right of the stellar centroid, reminiscent of the charge distributions measured in the laboratory under conditions of poor transfer efficiency (one of these distributions was shown in Figure 4.13).

Figure 5.22: Lucky Exposures image from the core of M15. The region in the red box is enlarged in Figure 5.23a.

Figure 5.23: Selected exposures from the core of M15. Cross sections through the image are shown in panels b) and c) along the dotted lines in a).