The MultiDrizzle Handbook


6.1 Limitations of Pipeline Processing

The goal of instrument pipelines are to provide data calibrated to a level suitable for initial evaluation and analysis for all users. Improvements to the calibration methods are immediately applied to the data when they are retrieved through the HST archives On-The-Fly reprocessing (OTFR) system. These same calibration script updates are propagated in releases of the IRAF/PyRAF STSDAS software package. If it has been a long time since your data was retrieved from the archive, it is advisable that you request them again to ensure that they contain the most up-to-date header information and calibrations.

There are presently two fundamental separate steps in the OTF process. The first is the calibration of individual data sets, and the second is combination of dithered datasets into a single output image using MultiDrizzle.

The second step cannot succeed without good results in the first. In some cases pipeline calibration of individual images will be insufficient. There are several occasions when On-The-Fly reprocessing is not ideal and when off-line interactive processing by the user is required. For instance the user may with to

NICMOS data in particular may require special attention by the user. Datasets often contain additional signal in the sky, persistence or pedestal (differing bias levels between quadrants in the chip) that require processing to be removed after the data have been calibrated through CALNICA. For more detailed information on recognizing and removing these effects in your NICMOS data see the data analysis chapter of the NICMOS Data Handbook which can be downloaded from:

Even when their individual datasets have been well calibrated, many users may wish to rerun the standard MultiDrizzle performed by the pipeline. The pipeline uses coarse values for both the output pixel size (scale) and drizzling kernel (pixfrac). This speeds up processing of the pipeline, and is sufficient to give the user a very good quick view of the field. However, when the user has well dithered data, he or she may wish to rerun MultiDrizzle at his or her home institution using parameters better suited for optimal cosmic ray removal and good final image resolution. In this chapter the user can find numerous examples showing how she or he may benefit from running MultiDrizzle on data from the archive.

The following sections illustrate specific examples of using the drizzle software with each of the instruments that it currently supports: WFPC2, ACS, WFC3, STIS/CCD, and NICMOS.

The intermediate and final drizzled images can take up a lot of space, and running the tasks in the dither package can be quite CPU-intensive, so care should be taken to make sure enough disk space is available before retrieving the images and running through the examples. All of the initial input images for each of these examples can be retrieved from the Hubble archive at

Space Telescope Science Institute
Voice: (410) 338-1082