2.1 Dithering Strategies
What is Dithering?
A popular technique in UV, optical, and IR imaging observations involves the use of dithering, that is, spatially offsetting the telescope by shifts that are small relative to the detector size and therefore moving the target to a number of different locations on the detector.
Two of the main strategies involve,
- Offsets by an integer number of pixels to facilitate the removal of bad pixels
- Offsets by sub-integer pixels to improve spatial sampling of the point spread function (PSF)
The latter application is particularly important in the case of HST because the PSF is so small that it is undersampled by most primary science instruments.
3. A third spatial offsetting technique involves the use of large shifts, comparable to the scale of the detector, to fully map areas of the sky that are several times larger than the detector area. This is generally referred to as mosaicing, the technique used for observations like the CANDELS or PHAT fields. In this context, these refer to large mosaics created from data taken over multiple visits using different guide stars, and should not be confused with mosaic-type dither patterns. Large multi-visit mosaics involve observational considerations and methods of data analysis that are beyond the scope of this document. However, the techniques covered in this document are essential to mosaicing with HST.
Benefits of Dithering
Dithering of HST observations is hardly new; the primary data acquisition modes of the GHRS and FOS involved both sub- and multi-diode offsets to obtain well-sampled data along the spectral dimension without gaps resulting from the presence of a few dead diodes. However, dithered observations in imaging mode became routine only after the dramatic improvement in HST optics with the installation of COSTAR and WFPC2 in 1993.
Dithering often provides considerable benefits to a science program, specifically in the following ways:
- Dithering reduces the effects of pixel-to-pixel errors in the flat field or in spatially varying detector sensitivity.
- Integer shifts of a few pixels allow the removal of small-scale detector defects such as hot pixels, bad columns, charge traps, and IR blobs from the image.
- Non-integral (subpixel) dithers allow, when correctly implemented, the recovery of some information lost to undersampling by pixels that are not small in comparison to the point spread function.
The third point is of particular importance for HST imaging; almost all the imaging instruments aren't able to fully take advantage of the resolving power of HST optics. This is because instrument designers had to decide between fully sampling a small field of view, or using coarser sampling on a larger field.
Dithering was particularly important when WFPC2 and NICMOS were HST's primary imagers. The width of a WFPC2 WF pixel, about 0.1 arcsec, was already comparable in size to the optics full-width-at-half-maximum (FWHM) in the I-band and substantially larger in the blue band. Images from the NICMOS camera-3 detector were similarly undersampled over much of its spectral range. While the ACS/HRC provided an adequately sampled PSF at optical wavelengths, it came at the cost of a drastically reduced field of view–it was just 1/50th the area of ACS/WFC.
HST's most commonly used imagers, WFC3/UVIS, WFC3/IR, and ACS/WFC, have detector pixel widths comparable to the FWHM of the point spread function (PSF). In theory, a minimum of two samples per FWHM would be required for full recovery of the image resolution. While the "missing" samples could be recovered with dithering, it's not possible to completely undo the low-level blurring produced by a larger pixel. Still, dithering provides substantial improvements to the final image quality, including better removal of detector defects.
Costs and Drawbacks of Dithering
While dithering provides substantial benefits, there are a number of trade-offs that must be understood and considered when deciding whether or not to obtain dithered data. Additional details are described later in the document, but summarized below.
- Additional spacecraft overhead time is needed for small angle maneuvers between dither points. Users will have to make some judgment calls about what they can do within their allotted orbits by testing different observing scenarios using the Astronomer's Proposal Tool (APT).
- A long exposure broken into shorter exposures at each dither point will increase the amount of read noise in the final combined image.
- If the primary science goal is to measure differential changes over time, as in time-series photometry, dithering may complicate data analysis due to flat-field uncertainties.
For most HST observing programs, potential drawbacks of dithering are outweighed by the scientific benefits. There are, however, some situations where the drawbacks could outweigh the benefits, such as in programs with very limited observing time.
For questions related to how your observing program could be affected by dithering, please consult the Phase II Proposal Instructions and use the Astronomer's Proposal Tool. If you need additional assistance, please get in touch with your Contact Scientist or submit a ticket via the HST Help Desk portal.