Characterizing MIPS PSFs can be difficult because MIPS has five different observing modes, large arrays (distortion!), multiple mirror positions in each mode (distortion!), and large bandpasses (color terms!).
The STINYTIM software simulation of the PSFs actually matches our observed PSF quite well, as the optics are very smooth at these long wavelengths. One way in which there is a significant difference is the short-wavelength light leak at 160 microns; for more information on this, see section 7.2.6 below. Note that the STINYTIM simulation PSFs are oversampled, and the pixel sizes do not match what you actually get with default Spitzer data.
In general, if you need to do a detailed PSF subtraction, we recommend that you create one from your own data (using objects of similar color to your target) to ensure that you have PSF standards of the same color and brightness and observed with exactly the same AOR parameters. The lack of repeatability in scan mirror position from campaign to campaign makes it necessary to do the PSF observation in the same campaign as the science observation. The Archive does not maintain a library of PSFs with characteristics spanning the full range of observer needs for PSF subtraction.
The point response function (PRF) is a similar but distinct concept from that of the PSF. Where the PSF is the point spread function before sampling by the detector array, the PRF gives the relative pixel values coming out of the array. The PRF is essentially a convolution of a box the size of the image pixel with the PSF. The software package APEX uses PRFs to carry out point source extraction. More detailed documentation on APEX can be found in the Data Analysis section of the website.
There are various ways to create PRFs to carry out point source extraction with APEX. For a single frame source extraction with APEX, e.g. extraction from a MOSAIC, the PRF is slightly different from that one used in single BCDs. There are PRFs posted on the MIPS portion of the website that have been created from BCDs, from mosaics, and from STINYTIM. These PRFs are based on the SSC mosaic pixel scale: 2.45 arcsec/pix and resampled by a factor of 4. Users should keep in mind that different mosaic pixel scales and/or resampling factors produce different PRFs. The empirical PRFs were created using PRF_ESTIMATE (from the MOPEX package), which is a script that combines those stars selected by the user, from a post-BCD mosaic, into a ''mosaic PRF''. Usually these stars have a high signal-to-noise and no neighbors, i.e. they are the best and cleanest for the job.
How different is the empirical PRF compared with the one derived using STINYTIM? They are not that different, but they are not identical. The main difference is at the depth of the first Airy ring, which is deeper in the STINYTIM PRF.
Please note: There is not a ''magic PRF'' that will work for every data set.