JWST Dithering Overview
JWST dithering is a data acquisition technique that enables data processing to remove various artifacts, fill gaps between detectors, and provide better sampling of the point spread function. Different dither strategies can be used for different purposes.
Dithering is a technique in which an observed astronomical scene's position is moved by a small amount and then re-observed, often for multiple times. When the resulting data are processed, the scene can be reconstructed without gaps or detector artifacts.
With some exceptions, dithering is recommended for all JWST observations because of how it improves data products for both the original proposers and later archival users. Data obtained from dithered observations can be processed to remove bad pixels. Depending on the number of dither steps and their sizes, dithering can also be used to fill in detector gaps, as well as sub-sample the point spread function at a given wavelength to produce improved resolution for the processed data. .
Each instrument has its own strategy and the dither parameters are, by necessity, instrument-specific. Customized dither patterns have been established for coordinated parallel observations. These dither patterns improve the data quality, simultaneously, for both the prime and the parallel instruments.
Different kinds of dithers
The terms "dither" and "dithering" are applied to a range of scene motions, with offset sizes from sub-pixels to several arcseconds.
Gap dithers to fill in detector gaps
See also: NIRCam Primary Dithers
Some instruments, like the NIRCam Imaging short channel, contain multiple detectors separated by small gaps. To fill in these regions of an astronomical scene, primary dithers are used; this results in small stripes or regions in the processed data with lower effective exposure times. However, with a sufficient number of dither steps, this non-uniformity can be minimized to produce an image with fairly uniform sensitivity. Gap dithers also move the scene sufficiently so that any detector features, such as bad pixels or modest flat-field features, can be mitigated.
Sub-pixel dithers to improve PSF sampling
See also: NIRCam Subpixel Dithers
In certain situations where pixels are not Nyquist sampled, dithering by fractions of a pixel can produce data that recovers some or all of the diffraction-limited performance. The effectiveness of this approach depends on the diffraction limit, which is a function of wavelength, and on the telescope's actual optical performance.
Roll dithers (coronagraphy)
See also: MIRI Coronograph Imaging Dithers
Roll dithers are different from normal dithers in that the scene is rotated about the target rather than offset. This technique is used in coronagraphy, where a target is placed in position behind a mask. Offsetting the target to move the scene would move the target from behind the mask, which is unacceptable to the science use case. Positioning the target behind a mask but rotating the scene provides some (but not all) of the benefit of a normal dither since it is an angular rotation rather than an offset; hence, the size of the offset in pixel space depends on the distance from the point of field rotation. In this use case, rotating the scene means that a separate guide star acquisition (and hence a separate observation) must be specified for the roll dither observation. Also, a roll slew is accounted as a regular slew even though the telescope pointing stays on the same target position. For example, a 10o roll maneuver takes the same amount of time as a 10o movement of position on the sky.
Coronagraphic sequences are groupings of observations that normally include two science target observations (including a roll dither) plus a PSF reference star observation, all done in a non-interruptible sequence (to minimize changing thermal or other conditions). Hence, roll dithers are implemented in a completely different manner from normal offset dithers. Roll dithers are not restricted to coronagraphy, but it is the primary example identified for its use.
Small Grid Dithers (Mainly but not exclusively coronagraphy)
See also: NIRCam Small Grid Dithers
Small grid dithering is a technique for making very small offsets in position (<60 mas). This technique uses the fine steering mirror on JWST to make very small, precise motions of the scene relative to a selected instrument. Its primary application is in coronagraphy and high contrast imaging. Because the JWST control system cannot place a given target at a given position with absolute accuracy, observations that require a target to be placed behind a coronagraphic mask will have some positioning error. Hence, observations of a science target and a PSF reference star cannot be aligned precisely. In order to maximize the matching of a PSF reference star observation to a given science observation, the PSF reference star can be observed at a number of small offset positions relative to the mask. Post-processing then allows the modeling and matching of the PSF reference star to the science target. Of course, the exposure time is being multiplied by the number of grid steps, so this improvement does not come for free. However, for science cases needing the highest quality PSF matching, the cost may be worth the price. Other specialized applications may also benefit from the use of this technique.
The allowable options for small grid dithers are selectable simply by choosing the appropriate dither pattern in a given template where it is defined. The details of the implementation (e.g., use of the fine steering mirror) are hidden from the user. See the individual instrument dither articles listed above for details.
Dithers and spectroscopy
Some spectroscopic modes allow dithering. In these cases, the user decides whether to step in the spatial direction, the spectral direction, or both, and by how much.
The NIRSpec Multi-Object Spectrograph, however, is a special case where integral full shutter steps are used in the offsets, and a large step is sometimes needed to obtain full spectral coverage due to the separation of the detector quadrants. In some documentation, the motion by integral shutters are referred to as nods rather than dithers. There are also some MIRI motions, such as offsets to a nearby background region, that are referred to as nods, but the intent of those motions is different from a dither. The nomenclature is unfortunately ambiguous, but can usually be understood from the context.
Interplay between dithers and mosaics
See also: JWST Mosaics
When defining mosaics, users should be aware that the footprint of a given mosaic tile will include any dither steps defined at each tile location. This may come into play when deciding on the amount of tile overlap to specify in APT, and on considerations about the uniformity of coverage desired. See the APT Mosaics article for more information.
Dithering in the ETC and APT
Each dither step produces a separate exposure as specified in the APT template. Hence, one needs to be clear about this when using the ETC to calculate the S/N for a given exposure or for the entire observation (all of the dither exposures combined). To approximate the total SNR expected, set the number of exposures in the ETC calculation to be the number of dither steps anticipated. But in transferring the ETC observation specifications into the APT template, users should make sure they are entering the values for an individual exposure. Selecting a given dither pattern in APT then automatically increases the number of exposures.
Pure parallel observations are not allowed to impact the primary observation. Therefore, they will either not be dithered or dithered with the primary observation's pattern (which would be non-optimal for the parallel observation). This is different from the situation for coordinated parallels, where dither patterns can be chosen to optimize step sizes for both instruments.
Time-series observations are normally cases where continuous time coverage with the target in a constant position on the detector is more important than improving the overall data quality. Hence, a dither selection of NONE can be used in these situations.