Quick-Start Advice To Run Reverse Correction
STEP 1) DOWNLOAD THE SOFTWARE PROGRAM FROM HERE For full-frames: wfc3uv_ctereverse.F For subarrays: wfc3uv_ctereverse_wSUB.F Parallel processing version: wfc3uv_ctereverse_parallel.F STEP 2) COMPILE THE PROGRAM IN SOME DIRECTORY (NOT NECESSARILY THE EXECUTION DIRECTORY) For full-frames: g77 wfc3uv_ctereverse.F -o wfc3uv_ctereverse.e or For subarrays: g77 wfc3uv_ctereverse_wSUB.F -o wfc3uv_ctereverse_wSUB.e For parallel compilation: gfortran wfc3uv_ctereverse_parallel.F -o wfc3uv_ctereverse_parallel.e -fopenmp * you can also use gfortran instead of g77 * the capital F is important; it invokes the pre-compiler (which it needs) Then you can run the program. This particular routine insists that all the exposures be in the same directory. The executable can be in a different directory, but the images you're reading in and operating on *must* be in your current directory. The reason is that the routine takes the first 8+q characters as the image root, then assumes that the 3-character image type designation follows an underscore (ie, _raw, _flt, etc). If you do not have access to gfortran or g77, then let us know. It may be possible to construct a few executable versions of the routine. STEP 3) TEST THE COMPILATION AND LOOK AT OPTIONAL PARAMETERS If you run the program without any arguments, you get a list of its arguments: For full-frames: ./wfc3uv_ctereverse.e or For subarrays: ./wfc3uv_ctereverse_wSUB.e For the parallel version: ./wfc3uv_ctereverse_parallel.e The full README at the top of the program has a detailed discussion of the individual arguments. This quick-start README will only touch on the "standard" operation. STEP 4) EXAMPLE OF BASIC OPERATION The most basic way of running the routine would be to correct an "_flt" image and generate an "_flc" image. This is akin to what the ACS pipeline produces, but the UVIS routine is unable to operate on the _flt alone, since the _flt may have had some post-flash electrons removed, and these electrons are important to help the routine estimate how charge may have been redistributed during readout. ./wfc3uv_ctereverse.e ibxd90xeq_raw.fits FLC+ Running it without the FLC+ option will generate the correction, but the routine will stop after generating the "rac" file (the corrected "raw"). Without the FLC+ flag set, the program will not presume that there is an "flt" file to correct and thus will not generate an "flc" file. *** NOTES for subarray data *** a) The subarray routine does not output an "rac" file. b) The current routine requires access to pre-scan bias pixels. Subarrays that are currently supported: UVIS1-2K4-SUB UVIS1-2K2A-SUB UVIS1-2K2B-SUB UVIS1-2K2C-SUB UVIS1-2K2D-SUB UVIS2-C1K1C-SUB UVIS2-C512C-SUB c) There are some subarrays that do not have pre-scan pixels; these are not supported at this time. However, please see STAN Issue 18 for a work-around solution for CTE-correcting unsupported subarrays. d) The different timing of the readout-cadence of the various subarrays will have some impact on the CTE trailing, but at this point, the model appears to work reasonably well. STEP 5) EXAMINE THE IMAGES Blink between the "flt" and the "flc" images; if you output various intermediate images (with the "I+" option), look at what has changed at each stage. STEP 6) CONTINUE PROCESSING AS USUAL If you made the "flc" images, then just analyze them or astro-drizzle them as you would "flt" files. ADDITIONAL INFORMATION The forward model is also available here: wfc3uv_cteforward.F A brief document describing the correction is available here: wfc3cte_alpha.docx The document describes the data used to constrain the model and shows what the model can and can't do. In particular the model has to be conservative to avoid amplifying readnoise. As such, it under-corrects faint sources on low backgrounds. At the same time---and somewhat counter-intuitively---the underlying model over-predicts losses of faint sources on low backgrounds, so users of the forward model should beware of this. We think we understand the cause of this and improvements are in the works; but the fix will require a tweaking of the algorithm itself, and not just parameter adjustment. REGISTRATION AND FEEDBACK Since this is an alpha version, we reserve the right to make significant changes. Some improvements are planned, but many will likely come as a result of user input regarding tweaks needed to make it run on different platforms. As such, we will be updating the code and providing notes below on changes/improvements. If you would like to be notified when updates are made, send an e-mail to email@example.com with "subscribe update_wfc3uv_cte" in the body.
|05/20/2013||1.0||Added capability for correcting subarrays (wfc3uv_ctereverse_wSUB)|