In order to acquire data suitable for super resolution post-processing, we take data in a manner that provides a high level of spatial sampling of the Airy pattern and redundant imaging at all three wavelengths. Simulations (Bippert-Plymate, Rieke, & Paul 1992; Astronomical Data Analysis Software and Systems I, A.S.P. Conf. Series, Vol. 25, D. Worrall, C. Biemesderfer and J. Barnes, eds., p. 205) show that such oversampling enhances the ability of computer processing to improve the angular resolution that is achieved in final data products. The standard 24 µm photometry observation described above is suitable for super resolution because of the ~ pixels and highly redundant images. Such is not the case at 70 and 160 µm; the nominal 70 µm pixel size is too large, and the 160 µm observation provides only 2 images of a source, which is inadequate for good image reconstruction. Here we describe the procedures used for obtaining good super resolution observations in these two bands. Table 3.4 provides a summary of this mode in all bands.
Figure 3.13: Source positions on the 70 µm array during the super resolution AOT for a compact source. These frames cover the source region; every other frame is of a sky position roughly 4´ away, and are obtained using scan mirror deflections. The pixel scale here is 5´´, so the circles representing the FWHM size of the PSF are correspondingly larger than for figures illustrating observations using the nominal (9.98´´) pixel scale at 70 µm.
At 70 µm, the default (coarse) pixel scale under-samples the Airy pattern. However, a separate ''narrow field'' optical train provides imaging at ˝ the coarse pixel scale, and is specifically designed for studies requiring high angular resolution. Presumably the source of interest is at least potentially extended, so separate sky images are obtained by moving the source on and off the array with the scan mirror. As a result of the on-orbit realities, we made changes to our pre-launch plans for this mode; it resembles the pre-launch AOT, but shifted to the less noisy side (side A) and with the amplitudes of the offsets correspondingly scaled down in the cross-scan direction.
To begin, the scan mirror is offset to one of the ends of motion permitted for the high-resolution optical train. Defining the center of the array as [0, 0], the telescope is pointed to place the source on pixel [x,y] = [-6, +2.5] and an exposure obtained. The scan mirror is moved to the other extreme of the permitted range and a sky exposure made. Next, the scan mirror is moved to pixel [-6, +0.5] and an exposure obtained, followed by a sky exposure, followed by one at [-6, -1], followed by sky, by one at [-6, 2.5], and by a sky. The telescope is then repointed in the cross-scan direction, and the sequence repeated with on-source exposures at positions [-7.5, +2.5]; [-7.5, +0.5]; [-7.5, -1]; and [-7.5, -2.5]. The actual source positions may differ somewhat from those given above and in Figure 3.13, which illustrates this mode of obtaining data; Figure 3.14 shows a visualization of the data frames relative to a hypothetical source.
This sequence gives two independent images sampled on half-pixel centers. The 1.5 pixel spacing has been adopted to provide some resilience to any dead pixels, which should then not remove two adjacent pieces of data. Additional data to increase the net integration are obtained from a slightly different starting position to dilute further the effects of any dead pixels on the final image. The observer only needed to select the number of basic observation cycles required. The procedure for obtaining the data was automatically optimized to achieve the best efficiency and data quality.
Figure 3.14: Source detections on the array during the 70 µm compact source super resolution photometry AOT shown in Figure 3.13. Time runs from left to right, top to bottom. The observations start with an image in the sky position, then a stimulator flash, then the first source image. Four on source - off source pairs of images are taken, then a final stimulator flash. The spacecraft is then offset, there is a stimulator flash, four more on - off image pairs are taken, and the cycle is completed with a final stimulator flash. Repeat cycles of the AOT omit frames 1 and 10, but still provide 8 source - sky image pairs per cycle.