MIRI Parallel Observations

JWST's Mid-Infrared Instrument (MIRI) can be used in parallel with the Near-Infrared Camera (NIRCam) during cycle 1. 

On this page

MIRI coordinated parallel observations are planned as part of a primary program (as opposed to pure parallel observations that are part of distinct, separate programs).  For cycle 1, Mid-Infrared Instrument (MIRI) coordinated parallel imaging observations are possible only with the Near-Infrared Camera (NIRCam). These coordinated observations are useful for sensitive images that require near-infrared through mid-infrared coverage. The design of coordinated observations has to consider the instruments characteristics, type of science, and operational aspects and limitations. 

Figure 1. Comparison of the MIRI and NIRCam FOVs

The MIRI imager is depicted in red, including the coronagraphic mask. The blue lines show the FOV covered by the NIRCam short wavelength A and B modules. The characteristic NIRCam 5” inter-SCA gaps are depicted as well.
Figure 2. Possible pointing for proposed deep observations

In this configuration, NIRCam is prime in the HUDF and MIRI is located in the GOODS South. The instrument overlays have been generated with the JWST Astronomers Proposal Tool. The NIRCam and MIRI overlays are highlighted in red and blue, respectively. Green crosses mark viable guide stars. Background image credit: STScI.

Operational aspects

The operational aspects to be considered are both observatory and instrument driven. From the observatory perspective, there are 2 basic rules: (i) no mechanism movement is allowed, and (ii) no calibration lamp can be switched on while another instrument is acquiring an exposure. These rules prevent both spurious vibrations and thermal instability, and require the 2 instruments in use to coordinate their mechanism changes (e.g., filter wheels) in between exposures. Calibration activities such as darks can be simultaneously carried out in the other instruments.  

From the instruments perspective, it is important to adjust the dwell time per dither position such that it minimizes the detector "dead time" per instrument (i.e., the time that one instrument is waiting for the other instrument to complete its exposure). The length of time that an instrument should acquire data (single or multiple integrations) in a single dither position is dictated by the detector performance. The initial plan for NIRCam deep field programs is to integrate for few thousand seconds. As for MIRI, recent analysis of data from the cryo-vacuum campaigns carried out at NASA Goddard Space Flight Center have shown that these exposure lengths fall into the regime in which the MIRI detectors' performance is optimal at short wavelengths. This guarantees that both instruments can be operated with detector dead time driven to a minimum.

Observing strategies that are designed to minimize the presence of persistence are crucial for deep fields programs. Although there are usually no bright sources in the field of interest, signal can build up making the identification of faint sources challenging. Observations that use a combination of short-to-medium length ramps with multiple integrations, and background matching and self-calibration techniques have been identified as useful strategies to both prevent and mitigate persistence. 

Dither pattern definitions

See also: MIRI Dithering

The JWST ground system will offer specific sets of dither patterns optimized for each instrument/mode. Similarly, new dither patters that effectively work for both instruments taking into account their individual detector characteristics, PSFs, FOV, and relative orientations need to be defined. In all cases, the prime instrument will drive the number of instrument configurations and length of the exposure times.

This section discusses preliminary dither patterns that have been defined using the NIRCam short wavelength channel and the MIRI imager, a configuration expected to be used in deep fields. Specific aspects and assumptions that have been considered are: 

  • The MIRI PSF is Nyquist-sampled at wavelengths longer than about 6.3 μm. Subpixel sampling may be needed at shorter wavelengths to properly recover the PSF. At longer wavelengths individual dither pointings should ideally be separated by at least 4× FWHM. This will improve both the PSF and the image cosmetics. The final optimal separation between dither positions is still not fully defined; testing with different distances using simulated data in all filters is underway. Aspects like detector effects (cross-like structure on the detector substrate) or the importance of diffraction spikes (also from sources outside the MIRI FOV) that may be relevant for faint sources are also being taken under consideration. 

  • NIRCam needs subpixel sampling for almost all wavelengths bluer than 2 μm in the short wavelength detector and 4.4 μm in the long wavelength detector. 

  • The dither pattern design of choice must offer a compromise between spatial coverage and depth, and that should be driven by the particular science objective. In this case, one basic need for MIRI is to have a reasonably large area covered, without having a negative impact on the depth of the NIRCam data.

  • We have used 4.772° offset between the MIRI and the JWST V2 axis (as measured in the cryo-vacuum campaigns), whereas NIRCam is assumed to be aligned with the JWST V2 and V3 axis.

The following dither examples target the MIRI filters that are under-sampled or close to Nyquist-sampled (5.6 μm, 7.7 μm, and 10.0 μm) and the NIRCam short wavelength channel. So far. no distortion has been considered, and there is no optimization for the NIRCam long wavelength channel. 

Figure 3. Four-point dither pattern example suggested for all MIRI filters

Left: Each position is at least 4× FWHM from any other position, to improve PSF and cosmetics. Each symbol/color represents one MIRI filter. The black circle marks the initial position for all filters. Dither points are distributed such that the peak of the PSF does not fall in the same row/column for any of the points. The central wavelength of the filter increases radially. These simulations serve to define the optimal separation.
Center: Simulated image of an extended source observed with MIRI in the 5.6 μm filter.
Right: Image recovered using the 4-point dither pattern.
Figure 4 shows an example combining a 4-point dither pattern that enables subpixel sampling with a 16-point pattern. This covers the 5” NIRCam inter-SCA gaps, but the large NIRCam intra-module gap is still present. Although a small area is mapped in both instruments (close in size to each instrument footprint), this pattern shows uniform coverage leading to a consistent depth in the observations: about 7 hours in the most exposed areas. Because of its size, this is not the best option for a deep field aimed at covering a large FOV with NIRCam and MIRI, but it is an excellent pattern of choice for science cases that do not need a wide coverage.
Figure 4. Example of a deep exposure with NIRCam primary and MIRI in parallel

The moves combine a 4-point pattern to enable subpixel sampling both in NIRCam and MIRI with a 16-point pattern that covers the NIRCam inter-SCA gaps. The scale on the bottom indicates the total integration time in seconds, the field is centered at (RA,DEC) = (00:00:00, 00:00:00).

Figure 5 shows a dithering example where larger spatial coverage can be achieved by adding 2 large dithers on each axis to cover the NIRCam inter-module gap. This survey design combines optimal subpixel sampling and a larger spatial coverage, about twice as much as in the previous one. However, this dither concept is not favorable for a MIRI deep field because of the varying coverage in the footprint, ranging from about 7.5 to 15 hours of total exposure time.

Figure 5. Survey design combining the optimal subpixel sampling for both NIRCam and MIRI

This example also includes the 16-point dithers to cover the inter-SCA gap and 2 large dithers along each axis to cover in the intra-module gap. The combination of the latter two cause MIRI to have quite a non-uniform footprint. The scale on the bottom indicates the total integration time in seconds, the field is centered at (RA,DEC) = (00:00:00, 00:00:00).
Figure 6 shows a dithering example that appears to be a good choice for deep observations. It not only allows for a more uniform coverage in MIRI but also surveys a large area in a 3 × 3 mosaic-like pattern. This pattern also provides a large spatial coverage for NIRCam and reasonably deep exposures in about 70% of the observed area.
Figure 6. Example of a 3×3 pattern design where NIRCam moves are planned to allow more uniform coverage of MIRI

The scale on the bottom indicates the total integration time in seconds, the field is centered at (RA,DEC) = (00:00:00, 00:00:00).

Choice of a MIRI filter

See also: MIRI Filters and Dispersers

The filters selected to carry out the observations impact the dither pattern definition. In the case of MIRI, the filter determines the minimum separation between different positions and the need of subpixel sampling.

The example dither patterns defined above favor the NIRCam short wavelength channel and the MIRI short wavelength filters (the most sensitive ones). However, there are other considerations driven by the science case and scientific aim of the proposal. 

For example, using more than 5,000 simulated galaxies, Bisigello et al. (2016) studied how the NIRCam broadband and the MIRI 5.6 μm and 7.7 μm filter combinations recover photometric redshifts for galaxies in the z = 0–10 range. They concluded that this combination of NIRCam broadband and MIRI 5.6 μm and 7.7 μm photometry not only lessens the impact of not having ancillary observations at λ < 0.6 μm but also refines the photometric redshift estimation. However, the time availability and desired depth of the exposures may constrain projects to use only a single MIRI filter. In that case, the filter selection needs careful consideration.

The filter of choice for MIRI is still a complex open question, and to make a decision, further data analysis, simulations and on-orbit performance should be considered. 


Bisigello, L., et al. 2016, ApJS, 227, 19 (arXiv:1605.06334)
The impact of JWST broadband filter choice on photometric redshift estimation

Latest updates
Originally published