HST changes temperature slightly as it passes in and out of the Earth's shadow. As the whole telescope shrinks and expands, the distance between its primary and secondary mirrors changes. Even though it only goes out of focus by a few microns, the ACS PSF changes significantly. Due to the different orientations of HST as it looks around the sky, the temperature (and PSF variations) change in a complex way that is essentially impossible to predict in advance. However, it can be approximately predicted using internal temperature sensors or by directly measuring the PSF pattern.
This page provides tools to (1) model the PSF at a range of focus positions, and (2) select which model is most appropriate for a given exposure.
Pre-prepared FITS images, containing a dense grid of (noise-free) model stars are available at focus positions separated by 1 micron. These were created assuming light a single wavelength of 435nm / 606nm / 775nm / 814nm (temporarily unavailable) / 850nm, with 0.03" pixels, to match the oversampled version of the ACS COSMOS images. More details are available in Rhodes, Massey et al. (2006; 2007).
The parameters of the PSF models can be easily changed, and new images can be manufactured, using TinyTim v6.3 and IDL wrappers.
Alternatively, images of real stars, taken from many COSMOS exposures at the same focus, as postage stamps, or overlaid on top of each other to create a dense starfield for each focus position, but with exactly the same image reduction parameters, solar aspect angle, etc. This assumes that the focus measurements in COSMOS are accurate (see below).
By matching a set of model PSFs to 10-12 stars in real images, it is possible to ascertain the effective focus position at the time of exposure to within 1 micron. We have mainly tested this procedure on the HST COSMOS survey. This is a uniform set of 589 ACS images taken with the F814W filter during HST cycles 12 and 13. We found the following results:
|Last updated by Richard Massey on 20th December 2019.|