MIRI Generic Recommended Strategies
A general set of guidelines can be applied to all MIRI observing modes.
Certain aspects of the proposal planning process are common to all MIRI observing modes: every program has to consider (i.e., how to choose an ideal combination of groups and integrations per exposure), that will help to mitigate the spurious sky + thermal telescope emission, and .
The following set of questions are meant to guide users navigating this page.
Specific best practices for the main MIRI observing modes and MIRI Time Series Observations can be found in the following JDocs pages: MIRI Imaging Recommended Strategies, MIRI MRS Recommended Strategies, MIRI LRS Recommended Strategies, MIRI Coronagraphy Best Practises and MIRI-Specific Time-Series Observations.
Detector readout best practices
Like other instruments on JWST, MIRI detectors use MULTIACCUM readouts. Once the final group in an integration is read, the detector circuit is immediately reset and, if defined, a new integration starts. The on-sky time of each individual exposure (i.e. the time spent in a single dither position) is defined by the number of groups and integrations. Note that time series observations (TSO) are a specific science case where very long non-dithered exposures are obtained, so some traditional rules may not apply.
How many groups and integrations should I use?
For MIRI, the optimal combination of groups and integrations depends on the science case, usually influenced by target brightness.
What is the recommended ideal, minimum, and maximum length of an integration?
The MIRI Si:As detectors exhibit a set of non-ideal, but known, detector effects (see e.g. Ressler et al 2015). The first few groups in a integration are most strongly affected. An increasing number of groups per integration can mitigate these effects.
- Ideal number of groups: Depends on the source brightness. In FAST mode, 100 groups (about 280 seconds in FULL array, shorter when using subarray mode) is suggested as a starting point. Increase or decrease the integration to achieve your desired signal-to-noise and/or avoid saturation. In SLOW mode 25 groups (about 600 seconds) is a good starting point.
- Minimum number of groups: Five groups is typically the minimum number required to obtain a reasonable calibration. For very bright sources, observers are permitted to use 2 to 4 groups, but there is no clear guideline for the noise and photometric accuracy we can expect in these very short integrations. When using fewer than 5 groups, you should plan to observe a calibration star with exactly the same exposure parameters if photometric accuracy is important to your program. If, instead of accuracy, repeatable precision is your primary need (e.g., transits), then this step may not be necessary.
- Maximum recommended integration length: 360 groups in FAST/FULL mode, 42 in SLOW mode (approximately 1000 seconds). This is not a hard limit, but it is based on three independent studies that consider cosmic ray rates from the Spitzer data. These studies have shown that during a 1000 s integration approximately 60% of the detector pixels are expected to be affected by cosmic rays (i.e. they will be usable but their noise properties will likely be changed). This not only includes the pixels that are directly impacted by a cosmic ray, but also the four adjacent pixels. There may be, however, cases where having integrations longer than 1000 seconds may be beneficial to achieve the science goals. For instance, deep cosmological imaging observations that run in parallel with other instruments may be better coordinated with longer ramps.
In many cases, the maximum integration length will be set by saturation of bright sources within the field (or emission lines within a spectrum). Because MIRI samples many points "up the ramp", saturating mid-way the integration does not mean that data is lost; the pipeline will fit only the initial unsaturated data points and return a correct flux value. This can be an advantage since you can integrate longer to gain better signal-to-noise on fainter areas of the field or spectrum. However, saturation, as well as bright sources, will leave latent images in subsequent exposures to deal with, although they do not appear to be a strong function of the integrated signal level.
Should I use multiple integrations?
The first integration within an exposure differs from the second and subsequent ones. This is because there are multiple detector resets between exposures (the system goes into a continuous reset while the telescope is dithering, filter wheels are moving, etc.), allowing some of the above-mentioned transient features to be cleared out of the first integration of the exposure. Between integrations within a single exposure, however, there is only a single reset; this leads to a flux-dependent decay effect for the first few frames of each successive integration. This effect is greatly reduced by the data processing pipeline, but the longer the integration, the more accurate the result will be (i.e., the errors in the correction become less significant with more samples). It is thus best to select integration lengths which are as long as your observations allow; (i.e., a few long integrations are better than many short ones).
Choosing the readout mode: FAST vs. SLOW
See also: MIRI Detector Readout Overview
- FAST mode reads out the detector every 2.775 seconds in full array mode, vs. 23.889 seconds in SLOW mode. FAST mode offers finer time sampling that allows better characterization of detector effects and more samples for cosmic ray correction.
- SLOW mode provides about 9 times less data volume.
- FAST mode is better suited for most imaging observations, it is thus the default (not mandatory) mode offered by APT.
- SLOW mode may deliver slightly lower noise and fewer subtle spatial artifacts in MRS observations. For this reason, it is offered as the default (not mandatory) readout mode by APT.
- SLOW mode integrations should not be shorter than 10 groups (~ 240 s); use FAST mode if shorter integration times are needed.
Users are encouraged to consider these points when choosing the readout mode, and to follow the general guidelines of integration length optimization.
Background subtraction best practices
Does my program need background observations?
Observers should carefully consider the JWST background's impact this extra emission (modeled by the Exposure Time Calculator Old) will have on their data. The recommended strategy to account for and remove the background depends on the source structure:
- For point source studies, the background will be included in the science data, and thus the observer does not need to include extra background observations to account for it.
- For extended sources, when the emission of the science target covers the entire FOV, it is recommended to define a background target and acquire data in every spectral configuration used for the science data. Specific considerations for each MIRI observing mode are given in this MRS page and the imager and LRS Best Practices pages.
To assess whether the background limited special requirement in APT should be used please read this Background-Limited JWST Observations JDocs page.
How often do I need to get a background?
The JWST mission defines visits as individual schedulable units, used to build up the observations timeline. Visits that are not linked by special requirements can be planned at different times of the observing cycle and thus background variations are expected. Observers should plan on taking background data for every observing period (i.e., non-linked visit) in their programs. The Zodiacal background will vary with a timescale of weeks. Pre-launch, it is difficult to predict the degree to which the telescope self-emission will vary temporally. Observing programs that need low backgrounds can request when the background is predicted to be relatively low. Observing programs that require high accuracy relative calibrations in the results (better than 1%) and use extended targets may consider taking background data before and after the science exposures.
Target acquisition best practices
See also: MIRI Target Acquisitions
An overview of the MIRI Target Acquisition process is given elsewhere in the documentation and discussed in the mode-dedicated Best Practices pages (Imaging, MRS, LRS, and TSO). The aim of this section is to give details on choosing the TA source and its environment in a way that ensures the centroid algorithm is carried out on the source requested by the user. The on-board centroid algorithm for MIRI works as follows:
- The raw image is pre-treated (background-subtracted and flat-fielded).
- It then finds the brightest 3 × 3 pixel region of the detector region of interest (ROI, see Table 1). The checkbox size (3 × 3) has been defined to encompass the imager PSF.
- After that, it performs a fine location by calculating the centroid in the previously located brightest area.
The success of this procedure depends on several aspects:
- The TA target should be a point source.
- The integrated signal-to-noise ratio should be ~30.
- The TA source should be the brightest source in the ROI. If there is a source brighter (by any factor) than the one selected by the observer, TA will be carried out on that. If there is a source with the same brightness the algorithm will perform TA in the first one encountered following the direction in which the detector is read out.
- Users should be aware that regions with bright diffuse emission may also result in false identification of the TA target. To reduce the possibility of this happening, the checkbox area is a very small one.
Table 1: Sizes of detector regions of interest used to perform target acquisition.
|TA Region of Interest||Size|
|MRS (TA performed in imager detector)||64 × 64 pixel2|
|LRS (slit and slitless)||64 × 64 pixel2|
|Coronagraphs||64 × 64 pixel2 (16 × 16 pixel2 secondary TA)|
Note that the MIRI detector plate scale is 0.11 arcsec/pix.
Target Acquisition can be performed using three different readout modes: FAST, SLOW (planned for upcoming APT/ETC version) and FASTGRPAVG. The selection of readout mode should be driven by the volume of data that can be handled on-orbit to perform the centroid algorithm, and the SNR achieved. As data volume can become a concern when TA requires a large number of groups, FASTGRPAVG is included to avoid on-board memory problems. In this mode, each group consists of 4 co-added FAST mode frames (see details in JWST ETC MIRI Target Acquisition). The ETC will also give warnings when the SNR achieved is not sufficient to successfully finish the TA procedure.
Ressler et al. 2015, PASP, 127, 675R
The Mid-Infrared Instrument for the James Webb Space Telescope, VIII: The MIRI Focal Plane System
Rieke et al. 2015, PASP, 127, 665R
The Mid-Infrared Instrument for the James Webb Space Telescope, VII: The MIRI Detectors
Lightsey, P.A., 2016, SPIE, 9904, 99040A
Stray light field dependence for the James Webb Space Telescope
Krick J. E. et al., 2012, ApJ, 754, 53
A Spitzer/IRAC Measure of the Zodiacal Light