In order fully to exploit the imaging performance of UKIRT (or any other modern high-performance telescope) it is necessary to bear in mind a few principles which affect imaging performance, espaecially insofar as it is affected by the performance of the tip-tilt image stabilisation system.
Guiding on a star which is a little way away from your target means that the fast guider is not looking through exactly the same column of atmosphere as the instrument looking at your target. This means that the image motion caused by the atmosphere will be different for guide star and target: the further away the guide star the more different the measured and desired atmopheric corrections. Eventually, when the guide star is seen through an atmospheric column which is completely uncorrelated with that through which the target is viewed, the correction is just additional noise and will actually degrade the image.
The degree to which one ought to worry about this is not yet clear, and certainly varies from site to site. So far we have failed to detect any anisoplanatic effects at UKIRT, but have not had a chance to look for them in good seeing (< 0.”6 FWHM). There is other evidence that tip-tilt seeing effects are small on Mauna Kea.
- We therefore suspect that anisoplanatism is probably not very important except in the very best conditions (<0.”3, say); and in any case
- it affects only the upper-atmosphere seeing contribution, which as a rule does not seem to have a lot of tip-tilt power.
However the fast guider also corrects for:
- Telescope vibrations (e.g. windshake) up to a few Hz, which are the same for any source
- The tip-tilt component of facility and boundary-layer seeing effects, which are near the telescope, i.e. will therefore be fully corrected almost irrespective of the distance of the guide star from the target.
Telescope vibrations are much more serious than the tip-tilt component of upper-atmophere seeing, so that fast guiding always improves the image quality, irrespective of anisoplanatism or even photon-starvation effects (q.v., below), as long as the latter allow the guider to stay locked onto the guide star.
The precision with which the fast guider can determine the centroid of the guide star image is a function of the S/N of the detected signal, which is determined by the number of detected photons in each readout of the guider CCD. In dark or “grey” conditions the sky background is small and the dominant noise sources are the read noise of the CCD (~3.1 e– per read) and the photon noise of the signal from the star.
The adverse effect of photon starvation is more readily apparent than that of anisoplanatism, and appears to be detectable (in the best seeing) at guide star magnitudes V > 14, at which level it is customary amongst UKIRT TSSs to begin to reduce the sample rate of the fast guider from the conventional 100Hz used for brighter guide sources (corresponding to 6.5 ms exposure per sample), which provides a closed-loop bandwidth of ~10 Hz.
The fast guider system has been extensively optimised and since late 1997 it has utilised a sophisticated Kalman filter algorithm to minimise the adverse effects of photon starvation. This, plus the remarkably sensitive low-noise guider CCD, permit guiding at 50 Hz on stars down to V ~ 18.6 in good conditions. (This sample frequency offers almost a threefold increase in counts-per-read but only a two-fold reduction in the closed-loop bandpass.)
Obtaining the maximum spatial resolution usually involves some form of PSF correction (often subtraction, sometimes deconvolution). If this is planned, the selection of PFS template stars should be done so as to produce a PSF as similar as possible to that affecting the target.
This implies that the PSF star should be observed by guiding in a manner similar to the target, i.e. not on the star itself, unless the target was also thus observed.
In general the guide star for the PSF template should be selected to have (in decreasing order of importance):
- The same apparent V or R magnitude as the target’s guide star unless both are brighter than V ~ 15 or so; the fainter the GS the more a good magnitude match is desirable, probably to <0.5 mag at the faint end.
- The same radial angular distance from the target (so that any anisoplanatism affects both equally).
- The same position angle (because of, e.g., wind, anisoplanatic degradation may not be circularly symmetric.)
The most important of these is the match of guide star brightness. As indicated above, image degradation due to anisoplanatism (off-axis guiding) (let alone angular anisoplanatism = guiding at a different position angle) has not yet been detected at UKIRT.