The calstis pipeline includes a step to automatically reject cosmic rays (CRs) in CCD exposures taken with the parameter CR-SPLIT >= 2. Many users have noted that this step doesn't do a particularly good job; the most obvious failing is an overly aggressive rejection that ends up deleting real data. After extensive testing to optimize the CR rejection parameters (see the ISR 98-22, Cosmic Ray Rejection in STIS CCD Images) a new parameter table for CRREJTAB has been installed that implements the recommendations from Table 1 of ISR 98-22. One can use "getref" to obtain this new table from the archive, or one can edit the old table using ISR 98-22 as a guide for your particular circumstance. Tests with the new table show a marked improvement, and a much lower false rejection rate while still eliminating most cosmic rays in an image. (The algorithmic changes to calstis2 suggested in ISR 98-22 have not yet been implemented.)

We note that the intent of the CR rejection step is not to produce high-quality science in all cases, which is not possible given the wide variety of data and modes possible with the STIS CCD, but rather to give a first cut that is at least aesthetically pleasing. If the needs of your science match the simple algorithm used, then this step can also be scientifically useful as is. One should, however, examine the pipeline product critically to determine if CRs are being rejected in a way that is compatible with your scientific needs. Simple, qualitative tests that one should do are to "blink" the CRSPLIT data frames and the data quality extensions in the "_flt" file (windowed to z1=8000, z2=8200, to show the CRs) to see which data points are being rejected as CRs, and whether this positively or negatively impacts the science you wish to do. Customized CR rejection can be done independently of the pipeline using the task "ocrreject" in the stis package, or with many other data editing tasks available in IRAF and STSDAS.