The appearance of the HST aberrated PSF is, by now, so familiar to everyone that I see no point in depressing people with more pictures of it. Many of the problems created by the current PSF have already been addressed. This section is simply a very quick summary and emphasis of these effects, after which we will take a more forward-looking approach. A more detailed discussion of these topics is available in Baxter et al. (1993).
The FOC PSF produced by the aberrated HST comprises a very sharp and narrow
core (FWHM 4 pixels (F/96)), containing approximately 15-20%of the
total
light. This is surrounded by large, highly structured halo, extending out to
radii in excess of 120 pixels (although relatively faint at this radius).
(Note: It is estimated that 2-5%(depending on wavelength) of the total light
is scattered very thinly to radii greater than this, but we shall ignore
this.) With this in mind, the first point to emphasize is that any science
data (i.e., the science target itself) which lies within this distance of the
edge of the format will have had some of its light scattered out of the format,
from where it cannot be recovered. This is what we refer to as ``edge effect''
and obviously, the closer the target is to the edge, the larger the effect.
The second effect related to edge proximity, and already mentioned, is geometric
correction. The halos of PSFs close to the frame edges may retain residual
distortions, tending to put halo structures out of alignment with equivalent
structures in the restoring PSF. Finally, nonlinearity and saturation may
significantly effect your science data and/or the restoring PSF. All three of
these effects, if present (and they almost certainly are, to some degree), will
lead to residuals being left in your restored images.
The second major topic which users should be aware of is that of OTA ``breathing'' which manifests itself as a pseudo-sinusoidal variation of the OTA focus. A plot of secondary mirror despace (de-focus) against time shows that the effect has a period equal to the HST orbital period, and an amplitude which varies between about 2-6 microns (see Fig. 4). Investigation of this phenomena has indicated that the cause of breathing is primarily thermal (Bely 1993), resulting from the presence of a temperature gradient between the interiors and surfaces of certain HST structural members. As well as the very clear orbital dependency, it appears that there is a significant dependence on attitude of the spacecraft with respect both the Sun and the bright Earth. The effect of this variation in the secondary mirror despace on the HST point spread function is that it produces a radial movement in the structures in the halo. Also, diffraction rings close to the core of the PSF tend to increase and decrease in radius.
From the point of view of image restoration, this could represent a serious constraint since it has been shown (Baxter et al. 1993) that restoration of perfect data (noise-free, undistorted, linear, and high S/N) with a PSF differing by only 4 microns in focus (otherwise identical) leaves significant residuals in the restored image, and that these are only enhanced and amplified by continued iteration. OTA breathing will continue to be a problem, even after the HST repair mission.
It would be expected that the successful deployment of COSTAR should remove the
need for extensive applications of image restoration techniques to FOC data,
however there will, no doubt, still be occasions when they will be used. COSTAR
will greatly alter the appearance of the PSF, most notably in that
the halo should contract in both power and radius. Some 70-80%of the total
power of the PSF should reside within a radius of 10 pixels with the peak
count increasing by a factor of 4-5 relative to the pre-COSTAR PSF (see Fig.
5). The halo of the post-COSTAR PSF should, for all practical purposes,
disappear at a radius of about 30 pixels (0 4 for F/151), which will
also
define the new perimeter width for edge effects. In fact, the 400-500%increase in the point source peak count rates, combined with the non-linearity
constraints outlined in § 6, should act to suppress the PSF halo even
further, making it significantly visible only when relatively long exposure
times are employed.
Unfortunately, these improvements in the PSF do not come without cost. In
particular, COSTAR introduces a position dependence into the FOC point spread
function which we have not had to deal with previously (see Fig. 6). The FOC
was designed to image the HST focal plane at a distance of 6 56 from
the optical axis (V1). At this distance the focal plane is tilted at an angle
of 10
with respect to V1, and it is this plane which the FOC images.
Unfortunately the focal surface produced by COSTAR is itself tilted with respect
to this plane and introduces field-dependent focus variation and
astigmatism (Jedrzejewski et al. 1993). Both of these effects
increase linearly with distance from the fully corrected field point. In Fig.
6 we show a simulation which compares the post-COSTAR PSF as it should appear at
the center of the photocathode (the fully corrected field point), and at the
bottom right corner of the 512
1024 (zoomed) format (the worst
case). It can be seen that the effect of this field dependence is small, but
is still likely to create problems for stellar photometry and image restoration.
