3.2	Pipeline Overview

3.2.1 calacs: Image Calibration

3.2.2 multidrizzle and pydrizzle: Distortion Correction and Dither Combination

3.2.3 When is OTFR not Appropriate?

Pipeline processing is carried out by two separate image processing packages: calacs corrects for instrumental effects to produce calibrated products. multidrizzle corrects for geometric distortion, performs cosmic ray rejection on combined images, and for dithered images, also removes hot pixels.

3.2.1

calacs: Image Calibration

calacs controls the image calibration steps based on the type of images and/or associations:

•	for CCD images, initial basic reductions such as overscan and bias subtractions are performed on each image using the task acsccd.

•	If the association has CCD images created from “CR-SPLIT” observations, or from repeated non-dithered exposures (several sub-exposures per observation), the task acsrej is used to combine the images and reject cosmic rays.

•	The task acs2d continues with routine image reductions; MAMA images are flat-fielded. CCD images—single images and images combined with acsrej—are flat-fielded and, in most cases, dark current is subtracted.

•	SBC MAMA images in the association, created from repeated non-dithered exposures (several sub-exposures per observation), are summed using the task acssum.

Fully calibrated data products from calacs (with suffixes flt.fits, crj.fits, sfl.fits) are in units of electrons.

While intermediate steps in calacs make use of sky subtraction values to perform certain steps, such as in identifying cosmic rays, all data products created by the pipeline will not be sky subtracted.

Calibrated products from the pipeline may still contain some artifacts such as hot pixels, cosmic rays, amplifier crosstalk, and in the case of post-SM4 WFC images, bias shift and bias striping.

Hot pixels and cosmic rays can be removed from dithered single images (with suffix flt.fits after calibration by calacs) using multidrizzle to process associations created from “POS TARG” or dither “PATTERN” special requirements in Phase II proposals.

As of spring 2011, calacs software does not correct for amplifier crosstalk, bias shift, or bias striping. To correct these artifacts, it is currently necessary to use stand-alone STSDAS routines outside calacs. These are described in Section 4.2.1.

calacs and Single Exposures

Each single-exposure raw image in an association undergoes standard detector calibrations in calacs, such as bias subtraction, dark subtraction, and flat-fielding (see Section 3.3), to create an flt.fits image. This is done regardless of whether those single images will be combined in later calacs steps. Data in the “SCI” (science image) and “ERR” (error image) extensions of a calibrated flt.fits image are in units of electrons, whereas the raw ACS images are in units of counts.

calacs and Combining of Sub-exposures

Depending on how multiple sub-exposures were executed, calacs has two different ways to combine them.

1.	If CCD images are flagged in an association table as belonging to a “CR-SPLIT” or repeated observations set,1 the following steps are performed by calacs:

-	bias subtraction, and dark subtraction for most CCD images, are performed on each raw image,

-	images are combined with cosmic ray rejection, and

-	the combined image is flat-fielded to create a calibrated image file with suffix crj.fits.

2.	If SBC MAMA images are flagged in an association table as belonging to a set of repeated sub-exposures, calacs takes the following actions:

-	each image is fully calibrated and flat-fielded to produce flt.fits files.

-	The flt.fits images are summed to create an image with the suffix sfl.fits.

Note that each single exposure image from a “CR-SPLIT” or repeated sub-exposures set will also be calibrated individually to produce a flt.fits image for later use in multidrizzle if the header value EXPSCORR=“PERFORM” which is currently the default.)

calacs and Dithered Exposures

calacs produces a calibrated flt.fits file for each single-exposure image in an association, including those created from using dither “PATTERN” and “POS TARG” special requirements in the Phase II proposal.

If more than one exposure was taken at a pointing using “CR-SPLIT” (as is sometimes the case for CCD observations), calacs produces a cosmic ray-rejected combined image, crj.fits, for that pointing.

If there were two or more repeated sub-exposures at a pointing, calacs produces a cosmic ray-rejected combined image, crj.fits, for CCD data. For SBC MAMA data, a summed image is created with the suffix sfl.fits.

However, calacs will not combine images from multiple positions. Later in the pipeline, after calacs processing is completed, flt.fits images will be corrected for geometric distortion and combined, with cosmic ray and hot pixel removal, by multidrizzle (crj.fits and sfl.fits files are not used in multidrizzle)

Table 3.1: Input and Output Image Suffixes from calacs and multidrizzle for Various Observing Modes

Image Type		Image Suffixes (suffix.fits)
		calacs Input	calacs Output	calacs Cosmic Ray Rejected?	MultiDrizzle Input	MultiDrizzle Output	Multidrizzle Cosmic Ray Rejected?
Single		raw	flt	No	flt	drz	No
Repeated observations	CCD	asn & raw	flt & crj	Yes	asn & flt or flt	drz	Yes
	MAMA	asn & raw	flt & sfl	n/a1	asn & flt or flt	drz	n/a1
CR-SPLIT		asn & raw	flt & crj	Yes	asn & flt or flt	drz	Yes
dither PATTERN or POS TARG		asn & raw	flt, flt & crj, or flt & sfl	Maybe2	asn & flt or flt	drz	Yes

---

1

SBC MAMA detectors are not sensitive to cosmic rays.

2

Depends on the image type. For “CR-SPLIT” exposures, calacs creates crj.fits combined images. For repeated MAMA exposures, calacs creates a summed sfl.fits file. However, combined images are not used as input to multidrizzle. Only flt.fits files are the primary input to multidrizzle. They can also be represented by an association table, if one is available.

3.2.2

multidrizzle and pydrizzle: Distortion Correction and Dither Combination

All ACS data are automatically corrected for distortion during pipeline processing using the multidrizzle task which automatically performs image registration, cosmic ray rejection and final Drizzle image combination. This task relies on the IDCTAB reference table for a description of the distortion model.

Observations that use dither “PATTERN” and “POS TARG” offsets in a single visit are described in the pipeline-created association table. If the association table is present in the working directory and provided as input to multidrizzle, the task will process the flt.fits files specified in it. multidrizzle also accepts, as input, wildcard-specified filenames, a comma-separated list of images, an ascii file containing a list of images, or just one image name.

multidrizzle performs geometric distortion corrections on all flt.fits input images. Using WCS information in the image headers, the images are registered and combined with cosmic ray removal. For dithered images, hot pixels are also removed. The resulting combined image, in units² of electrons/second, is written to a file with the suffix drz.fits. (For WFC, data from the two chips are mosaicked together as one image.)

It is important to recognize that multidrizzle is built around the pydrizzle task, which was capable of aligning the images and correcting for the geometric distortion but did not remove cosmic rays. Instead, pydrizzle relied on crj.fits products as input to create cosmic ray-free images.

The use of multidrizzle supersedes pydrizzle, and uses the original flt.fits files directly as input to combine images with cosmic ray rejection, and for dithered images, also performs hot pixel rejection. This has significant advantages for observations with one exposure, or very few sub-exposures at each pointing, taken at few dither positions. In these cases, multidrizzle will use the information from overlapping fields in all flt.fits images to produce the best possible cosmic ray and hot pixel rejection for the output combined image. The resulting drizzled images should generally be useful for science as-is, although subsequent reprocessing off-line with multidrizzle may be desirable to optimize the data for some specific scientific applications. For more information, please see Section 3.5.1.

3.2.3

When is OTFR not Appropriate?

The goal of the ACS pipeline is to provide data calibrated to a level suitable for initial evaluation and analysis for all users. Observers require a detailed understanding of the calibrations applied to their data and the ability to repeat, often with improved products, the calibration process at their home institution. There are several occasions when data processed via OTFR from the Archive is not ideal, requiring off-line interactive processing:

•	running calacs with different reference files than those specified in the image header.

•	Running calacs with non-default calibration switch values.

•	When images must be cleaned of artifacts such as new hot pixels, cosmic rays, crosstalk, bias shift, or bias striping.

Images combined with MultiDrizzle in the pipeline were produced using parameters that are suitable for the widest range of scientific applications. Some datasets, however, would benefit significantly from further processing off-line. For example, a different pixel scale or orientation may be desired, or cosmic ray rejection parameters might need to be slightly modified. The same muultidrizzle task used in the pipeline is also available to users in STSDAS for off-line processing of flt.fits images retrieved from the Archive. multidrizzle provides a single-step interface to a complex suite of tasks in the STSDAS dither package. It is built around the pydrizzle task and needs to be executed within the PyRAF environment. For more information, please refer to the Multidrizzle Handbook.

---

1

The Phase II proposal’s exposure logsheet line parameter "Number_of_Interations" has an integer value greater than 1.

2

The final output unit type is set in the multidrizzle task parameter final_units. The choices are “cps” (the default value) for counts per second or “counts.” For ACS, the unit of counts is electrons, as specified in the image group header keyword BUNIT.