1) Freak: Gene Kopan's "freak" program is executed on the frame extraction files for each scan (./flat/[j,h,k]SSS.fex), which produces 256^2 images (*.fex.fits) containing the source extraction count (number of frames where a source was identified) for each pixel. Calls like the following:
freak k001.fex 14.0
resulted in extraction counts for sources brighter than 14th mag.
2) Thresholding: Most pixels in the FREX images are 0 or 1 in low density scans. Therefore the images must be thresholded above at least 1. Pixels with 2 extractions (FREX=2) are also ignored. Comparison of the FREX data for different scans (for example, see Figure 1 of the initial analysis for scans S001-S005 of 970418n) show they are random events that do not persist across scans; these are point sources that happen to fall in the same array pixel for different frames [different sky positions]). Note that the vertical banding of pixels with 2-extraction is due to the fact that a point source will repeat in ~6 frames at the same X (cross-scan) position due to sub-pixel dithering. Similarly, pixels with 3 extractions (FREX=3) were not found to occur with repeatability from scan to scan, and are therefore rejected as candidates for significantly "hot" pixels.
At the level of 4 or more extractions, things start getting interesting, in that pixels which appear persistent in more than 1 scan generally have FREX>=4. This can be seen easily in Figure 1 for S001-S005 of 970418n. With these considerations, the FREX images for each scan are thresholded at FREX>=4, the remaining pixels being set to 0.
3) Stack (coaddition) of FREX images: To learn which pixels are the "hottest" (having the most frame extractions overall), and how many pixels on the array are effected, the thresholded FREX images are coadded (=stacked=summed).
4) Hot pixel lists: The coordinates and values (X,Y,Z) of all pixels with FREX>=4 are listed into tables for each scan's FREX image, as well as for the coadded FREX images at J, H, and K. This greatly reduces the volume of data and allows more flexible and efficient data visualization with 2-d plots than is possible using the direct images.
5) Visualization: Rather then simply displaying the coadded FREX images, which gives little insight into what is going on without extensive interaction ("picks") of pixel values, X-Y plots are generated to show the sum (coadded values), as well as the number of scans, in which FREX>=4 was found. Examples can be seen in the FREX sum and FREX scan counts plots in Figure 1 for dataset 970418n. The later figure shows in a concise way the persistence or intermittent nature of each hot pixel candidate. The FREX scan count plot is a concise way of visualizing the information presented as movies in Figure 1 of the initial hot pixel analysis.
6) Histograms and cumulative distribution functions (CDFs) are constructed to learn how many pixels are effected at different frame extraction (FREX) levels. Then the distribution function for each scan's FREX data is compared to the distribution for a selected reference scan to search for possible scan to scan differences. These results are presented in Table 2 of each night's report. See for example Table 2 for 970418n.
7) The key plots and statistics are presented in an HTML report generated by the software. See for example 970418n and 970423n.
Return to hot pixel analysis index.