C.1 Why Mosaicking and Dithering are Needed

The sizes of telescope pointing offsets between successive exposures can be very different, depending on whether the purpose is “mosaicking” or “dithering.” Mosaicking is done with the aim of increasing the area of sky covered by a particular set of exposures, usually with the goal of providing a seamless joining of contiguous frames. The angular offsets used when mosaicking are generally large, up to the size of the field of view. Only programs observing targets larger than the field of view of the detector need to use mosaicked exposures.

Dithering generally involves much smaller telescope offsets, often on the order of a few pixels in size. Most imaging programs are advised to use dithering for several reasons, including:

• removal of hot pixels and other detector blemishes (Section 6.10.2)
  
• improving sampling of the PSF (Sections 6.11.1 and 7.10.1)
  
• improving photometric accuracy by averaging over flat-fielding errors (Sections 5.4.3, 5.7.4, and 6.11.1)
  
• bridging over the gap between the chips in the UVIS channel (Section 5.2.2).
  

Dithered and mosaicked exposures can be combined using software included in DrizzlePac. Several documents provide examples of how to use this software, including the ReadTheDocs software documentation and the DrizzlePac Handbook.

WFC3 ISR 2015-04 describes a methodology for optimizing the parameter pixfrac and shows the results of tests conducted for the Frontier Fields program. WFC3 ISR 2015-09 shows how mosaic alignment can be achieved in a single step in DrizzlePac 2.0 by building up an expanded reference catalog, and uses the WFC3 observations of the Eagle Nebula (M16) as an example.

Instead of using DrizzlePac, WFC3 ISR 2014-23 describes the procedure by which the individual images in the Frontier Fields program were aligned using galaxies and provides the FORTRAN source code hst2galign that accomplishes the alignment. Further uses of hst2galign involving PSF-fitting and faint source location and photometry are discussed in WFC3 ISR 2014-24.

Note that it is sometimes necessary to use software like that in DrizzlePac to combine even CR-SPLIT or repeat exposures, when pointing drift causes slight misalignment of exposures and differences in how PSFs are pixelated, or when gradual changes in focus over the course of an orbit produce changes in the observed PSF.

In some programs, especially those observing time-variable phenomena, combining dithered exposures to correct for cosmic rays and transient bad pixels may be scientifically infeasible. In such cases, single-image based methods must be used. These methods use statistical properties of cosmic-ray brightness or sharpness to identify and interpolate across cosmic rays. Single-image cosmic ray rejection schemes are not available through the standard WFC3 calibration pipeline.