Detector configuration

See also: Understanding Exposure Times, MIRI Cross-Mode Recommended Strategies

For time-series observations, the challenge is to achieve maximum detector stability and measurement precision over timescales ranging from several to many hours. This section provides guidance for achieving the best possible results for TSOs using MIRI observing modes. 

Exposure configuration

TSOs are best executed in a single staring exposure covering the duration of the time-variable event of interest, in order to achieve optimal detector and telescope pointing stability. 

Contrary to "regular" science observations, the total length of the exposure is not chosen based on the required signal-to-noise ratio, but rather on the duration of the event. Knowledge of the event's timing and duration is therefore essential for planning such observations (see Scheduling TSOs). In the data calibration and processing of TSOs, each integration ("ramp") is treated as an individual exposure. 

The number of groups in each integration should be calculated using the JWST Exposure Time Calculator (ETC) by optimizing the signal-to-noise ratio of each integration and/or the desired time resolution (i.e., the cadence at which the light curve is sampled). In the case of very bright targets, the user has to balance the number of groups (where, generally, more is better) against potential saturation issues. In Figure 1 we provide some guidance for these choices. In general, the minimum recommended number of groups in an integration for accurate calibration is 5; values between 2 and 5 are also permitted, though the calibration accuracy may be sub-optimal for such observations. The maximum recommended length of an integration is approximately ~1,000 s, based on our current best understanding of cosmic ray statistics and the effects of cosmic ray impacts on the MIRI detectors.

The number of integrations in the exposure should be calculated based on the required duration of the total exposure. In the case of exoplanet transits, this should include a sufficient time beyond the transit duration to reliably characterize the out-of-transit baseline. Some additional time at the start of the exposure should be included to account for detector settling (see section below). The maximum number of integrations permitted in a single exposure is 65,535.

Detector readout mode

Detector readout patterns

For the spectroscopic modes, the detector readout patterns are fixed. When used in slitless mode, the low resolution spectrometer (LRS) uses the SLITLESSPRISM subarray, which provides a higher dynamic range than is available in the FULL array configuration (see MIRI Bright Source Limits). For time-series imaging, all imaging subarrays may be selected for TSOs. 

If a MIRI TSO imaging target saturates in <5 groups, a smaller subarray should be chosen to allow for a higher number of groups for a given integration time. 

Detector settling time

We recommend that users start their exposure approximately 30 minutes prior to the time period they actually need to cover in order to allow for detector settling.

MIRI's mid-infrared detectors display a number of non-ideal behaviors such as nonlinearity, latents, reset anomalies ,and drifts (see Rieke et al. 2015, Ressler et al. 2015). There are several mechanisms behind these observed response variations, some arising in the detector substrate itself, others caused by the cryogenic readout electronics. TSOs use multi-integration exposures that may be affected by drifts, i.e., where the response measured between integrations gradually changes at the beginning of the observation, settling to a more stable response over time. While these data are still considered "science quality" (i.e., accurate ramps can be constructed), adding a settling time at the start of the observation is thought to help improve the relative photometry and enable more accurate detrending of the drift function.

To achieve the highest possible measurement precision in the time series, users should therefore add additional time in the start of their exposure to account for this initial settling. A complicating factor is that the strength of the effects and the required settling time is strongly dependent on the flux levels received in the preceding integration prior to the detector reset, so there is no general estimate for the detector settling time that will cover all cases. In addition, for TSOs where the exposure covers a wide wavelength range with a strongly varying SED (e.g., a stellar-type SED observed with the MIRI LRS), the detector settling time may even vary across the pixels in a given exposure. The value of 30 minutes is considered a conservative estimate on the settling time that should cover most observations. 

This recommendation is supported by findings during commissioning, where response drift of ~0.3% was observed during the first 20 minutes of the time series. 

Dealing with saturation

See also: MIRI Cross-Mode Recommended Strategies, JWST Time-Series Observations TSO Saturation, MIRI Bright Source Limits

The issue of saturation for MIRI and TSOs is described in more detail in the reference articles above. Observing very bright targets with MIRI can cause detector artifacts (e.g., Kendrew et al. 2015, Argyriou et al. 2023), and these issues become worse as the pixels approach saturation. In general, users should be cautious about integrating up to saturation.

When using the MIRI imager for TSO observations, it is possible to move the observations to the smaller subarrays to avoid saturation. Smaller subarrays have faster frame times, and therefore more groups can be recorded before saturation is reached. There is no perfect number of groups to completely avoid detector systematics, but getting to N_groups > ~20 has been shown to reduce detector systematics compared to having N_groups < ~10. One way to look at this is to use the largest subarray possible that allows for N_groups > 20, while keeping in mind recommendations to stay below the full well capacity of the MIRI detector. In the event that N_groups < 20 on the smallest subarray, follow the directions for bright targets below. Please note that users may have other reasons to want to use the larger subarrays or FULL array that may take precedence over the consideration of the number of groups. 

For the imager, LRS, and MRS modes, there may be situations where the benefits to the MIRI detector stability in additional groups in the integration outweighs the risks posed by saturation at the end of the ramp. For N_groups > 5, the JWST calibration pipeline does not use the final group of a ramp in the ramp fitting (due to the last frame effect), so saturation in the final group in those cases will not impact the ramp fit. The JWST pipeline is also able to flag saturated groups in an integration, and only use the non-saturated groups to construct the ramp; therefore, saturation in the final groups of an integration does not cause the data from the rest of the ramp to be "lost." 

Figure 1 shows a decision flow diagram to help the user choose the appropriate number of groups for their time-series observation in the case of very bright targets. These recommendations apply to all three MIRI TSO modes.

Dithering

See also: MIRI LRS Dithering, MIRI Imaging Dithering, MIRI MRS PSF and Dithering, Special Requirements

Dithering is not permitted for TSOs with MIRI. If the user has assigned a dither pattern to an observation when the Time Series Observation special requirement is selected in APT, an error will be returned.

Background subtraction

See also: MIRI LRS Recommended Strategies, Special Requirements

TSOs with the LRS use the SLITLESSPRISM subarray, which measures 416 × 72 pixels in size. The typical brightness of TSO targets, especially at the shortest wavelengths, creates a significant amount of scattered light (the cruciform artifact, seen in all MIRI modes). This artifact is extensive enough to fill the entire width of the subarray, creating an additive component to the background. When performing background subtraction using off-target regions of the subarray, the presence of scattered light leads to an over-subtraction of the background flux. To achieve accurate absolute flux measurements across the spectrum, a dedicated background observation is recommended. The background observation should immediately follow the science exposure, such that it does not impact the critical start time of the TSO. 

Science and background exposures should be linked as a non-interruptible sequence to ensure that the background exposure is obtained immediately following the science exposure. The background observation does not require additional timing constraints. 

Target acquisition considerations

See also: MIRI Cross-Mode Recommended Strategies, MIRI LRS Slitless Target Acquisition, MIRI MRS Target Acquisition, MIRI LRS APT Template, MIRI MRS APT Template

Because of the high stability and precision required in a time-series observation, target placement at the subpixel level is crucial. This is particularly the case when multi-epoch observations will be carried out and combined to reach the required measurement precision. Target acquisition (TA) is therefore highly recommended for such observations. 

For LRS slitless TSOs, TA is recommended. For MRS, TA is mandatory when the Time Series Observation special requirement is selected. Please refer to the articles above for detailed descriptions of the procedures and choices for TA in these modes.

Target acquisition is not currently supported for MIRI imaging TSOs. For these observations, the accuracy of target placement will be limited by the JWST pointing accuracy and stability. The smallest subarray for MIRI imaging measures 64 × 64 pixels (SUB64), or about 7" × 7". Provided that the coordinates of the target and its proper motion are well known, the telescope will be reliably able to acquire a target and place it within the MIRI imaging array, even when using the SUB64 subarray. 

When performing multi-epoch imaging TSOs without target acquisition, one may see the target placed onto a different set of pixels across the various epochs. This may introduce additional systematics when the multiple epochs are combined, as different pixels will be subject to different noise residuals from flat fielding, intra-pixel gain variations, and other detector effects. This is likely to result in somewhat degraded performance. 

Scheduling TSOs