Custom dither patterns mitigate bad pixels and provide subpixel sampling to improve the effective spatial resolution of combined (drizzled) images for both the prime and the parallel instrument modes.
For instrument modes used as the primary in coordinated parallel observations, the APT template provides "custom" dither patterns, in addition to those made available to the prime mode for "regular" observations (i.e., observations not involving parallels).
Custom dither patterns improve the effective spatial resolution of the final combined (drizzled) images for both the prime and the parallel instruments by providing mitigation of bad pixels and flat field uncertainties, while also providing subpixel sampling.
For all coordinated prime+parallel mode combinations, the custom dither patterns involve non-zero steps along both detector axes (x and y) such that none of the dithers fall on the same detector row or column. Each dither constitutes an integer plus a fractional pixel (subpixel) step. The integer pixel component of the dither step mitigates bad pixels and flat field uncertainties, while the fractional pixel component improves PSF sampling and achievable spatial resolution in the combined image.
There are different dither patterns for different combinations of prime and parallel instruments. The sections below describe the general philosophy behind these dither patterns, followed by the specifics for the different instrument combinations.
General determination of custom dither pattern steps
Sub-pixel sampling of the custom dither patterns
The fractional pixel components of the dither patterns are chosen to sample the 2-D pixel phase in a nearly optimal way, with better subpixel sampling for patterns with more steps. Custom patterns have been designed with 2, 3, 4, and 9 steps. The desired subpixel shifts are listed in Table 1.
Table 1. Fractional pixel component of custom dither patterns.
|Number of dithers||Pixel phases (x, y)|
|2||(0.00, 0.00)||(0.50, 0.50)|
|3||(0.00, 0.00)||(0.33, 0.33)||(0.67, 0.67)|
|4||(0.00, 0.00)||(0.00, 0.50)||(0.50, 0.00)||(0.50, 0.50)|
|(0.00, 0.00)||(0.33, 0.00)||(0.67, 0.00)|
|(0.00, 0.33)||(0.33, 0.33)||(0.67, 0.33)|
|(0.00, 0.67)||(0.33, 0.67)||(0.67, 0.67)|
Determination of dither step sizes
Dither step sizes take into account two detector properties of the prime and parallel instruments:
- their relative orientation in the JWST focal plane (as measured during the 3rd cryovacuum test performed at GSFC during the summer of 2016), and
- their (mean) pixel sizes in units of "/pixel.
For the prime instrument, the custom dither patterns nominally place the target on the exact pixel phases listed in Table 1. For the parallel instrument, target placement is generally precise to within a radius of 0.05 pixels. This radius is similar to the nominal pointing uncertainty of small angle maneuvers with JWST, which is 5 mas. Exceptions to this general rule are noted in the following sections.
Custom dither patterns provide several choices for the overall size of the pattern. The largest sizes will be best for mitigating flat-fielding uncertainties, especially for extended objects. However, due to the geometric distortions of the images, the subpixel phases will vary more across the detector for the larger patterns. Also, there is less detector overlap after image combination for the larger patterns . Therefore, the choice of pattern is a compromise that will depend on the specific goals of the observations.
Figure 1 illustrates a 2-step dither pattern that works well for NIRCam imaging + MIRI imaging. The MIRI detector pixel array has an orientation on the sky that's offset by ~4.76° counter-clockwise with respect to that of the (average1) NIRCam detector. It is shown in red on top of the black NIRCam shortwave (SW) detector pixel array. The encircled black point shows a dither of size (17.50, 20.50) in NIRCam SW pixels, which yields an offset of (5.516, 5.469) in MIRI pixels. The latter is within a radius of 0.034 pixels, of the goal of (0.50, 0.50) in fractional pixels.
1 The "average" refers to the average orientation on the sky of the 10 NIRCam detectors.
Instrument-specific requirements for custom dithers
The PSF undersampling is most severe in the short wavelength (SW) channel of NIRCam: up to a factor ~3 with the F070W filter (versus up to a factor ~1.5 for the F277W filter in the long wavelength (LW) channel). As such, the custom dither patterns for all combinations involving NIRCam were designed to provide optimal benefits for the 0.0317" pixels in the SW channel of NIRCam.
For combinations in which NIRCam is the parallel instrument, the requirement that a given pixel phase is reached to within 0.05 pixels for the parallel instrument is softened to 0.11 pixel. Since the pixel size of the NIRCam SW detectors is so small, the resulting misplacement is still smaller than the nominal pointing uncertainty of small-angle maneuvers with JWST.
An additional requirement from NIRCam is that dithers cannot be placed within 2 pixels in X and Y of any other point in the same dither pattern.
Dithering for MIRI imaging requires that the distance between all steps in a dither pattern is at least 3 times the FWHM of the PSF (mainly to avoid overlap with latent images from preceding exposures). Since the size of the MIRI PSF changes significantly among the various MIRI filters, all custom dither patterns involving MIRI have been established specific to the selected MIRI filter.
For combinations in which MIRI is the parallel instrument, the requirement that a given pixel phase is reached to within 0.05 pixel for the parallel instrument is only enforced for MIRI filters with central wavelengths <10 μm (i.e., for MIRI filters F560W and F770W). This is because the MIRI PSF is well sampled at longer wavelengths.
An additional requirement for MIRI is that dithers cannot be placed within 2 pixels in X and Y of any other point in the same dither pattern.
The only specific requirement for NIRISS is that dithers cannot be placed within 2 pixels in X and Y of any other point in the same dither pattern.
For the custom dither pattern offered for the NIRSpec MOS + NIRCam imaging combination, the main restriction on dither size is for keeping the NIRSpec science targets positioned near the nominal center of the NIRSpec MSA shutters to minimize additional aperture throughput losses.
ASCII tables of custom dither patterns
Users may want to study the characteristics of the custom dither patterns for coordinated parallels in detail prior to selecting the most appropriate pattern for their needs. The patterns are available as ASCII tables at this page: Coordinated Parallel Dither Tables. All pointing offsets are relative to the reference position of the selected prime instrument's aperture ideal coordinate frame.