JWST Time-Series Observations TSO Saturation
The optimal number of groups and integrations for time-series observations are defined by instrument-specific strategies & signal limits — the so-called saturation limits.
When preparing time-series observations (TSOs), the concept of saturation is one of the central ones when trying to optimize the number of groups and integrations. This, in turn, helps define not only which targets are actually feasible scientific cases with the instruments onboard JWST, but also strategies to optimize the efficiency of the observations (i.e., the photon collecting time).
This article provides an overview of the concept of saturation in the context of JWST for TSOs: what it means, how important it is for the different instruments and strategies to cope with its impact for scientific observations.
What is saturation?
Saturation in the JWST context
The strict physical definition of saturation for a pixel in a detector is related to its full well capacity: the signal level at which any additional photon reaching a given pixel can no longer be counted.
However, in the JWST context, there are 2 additional definitions of saturation that are of relevance both for data analysis and proposal preparation: the one used by the JWST Science Calibration Pipeline, and that used by the JWST Exposure Time Calculator (ETC) uses.
Although the full well capacity of the pixels has been directly measured for all the detectors onboard JWST, the closest a pixel is to this limit, the more challenging it is to calibrate various detector-level effects. Based on this, different instrument teams have defined different signal limits set to ensure that data can be adequately calibrated, which is what defines the concept of saturation both in the JWST Science Calibration Pipeline and the JWST ETC. While this signal limit is defined for every pixel on every JWST detector in the JWST Science Calibration Pipeline (and can be accessed through the saturation reference files in the JWST Calibration Reference Data System—CRDS), the JWST ETC considers a single representative limit for all the pixels. The most relevant for proposal preparation is the JWST ETC signal limit which, depending on the instrument, is around ~70% to ~80% of the detectors' full well capacity for TSOs.
For the Cycle 3 Call for Proposals, it is expected that users use this signal limit as a guide and follow the instrument-specific guidelines to interpret those limits.
Comparison to other saturation definitions
Definitions of saturation or signal limits in other space-based instruments like, e.g., the Hubble Space Telescope's (HST) Wide Field Camera 3 (WFC3) or Spitzer's Infrared Array Camera (IRAC), have defined observational strategies for TSOs in the past. The detectors used by WFC3 and IRAC are similar to the ones used by JWST's near-infrared instruments and MIRI, respectively; but the detectors on board JWST have benefited from more recent technology development. Users should therefore be cautious when translating their HST or Spitzer experience to JWST. Here we compare how the concept of saturation differs between these generations of instruments.
For HST/WFC3, a commonly adopted definition of saturation involved measuring each pixel's deviation from linear behavior: when the non-linearity correction for a pixel is above 5%, the pixel is considered "saturated" (Hilbert 2008; signals above this limit are simply not corrected for non-linearities in HST/WFC3 data per contractual specifications). Initial considerations for planning observations with HST/WFC3 for TSOs (typically for transit spectroscopy) suggested some observations could benefit from targeting counts much lower than this saturation level to avoid the non-linear regimes of the detector (McCullough & MacKenty, 2012). However, recent methods for dealing with lightcurve systematics (which come mainly from persistence/charge trapping; see e.g., Stevenson & Eck 2019 and references therein) have mostly lifted this constraint for HST/WFC3. Observations targeting around ~70% of the "saturation" level were still able to deliver (differential) high-precision spectrophotometry during TSOs at the highest fluences (Stevenson & Fowler, 2019). This suggests that near photon-limited precision is achievable even in possible non-linear regimes of the detector for that instrument configuration. Spitzer/IRAC uses the "traditional" definition of saturation (i.e., the signal level at which full well is achieved). Recommended signal limits are also based on the level of non-linearity of the detector: correction at the 1% level is only possible up to ~90% of full well, after which the signals cannot be linearized.
Translating the above HST and Spitzer experience to JWST is, however, not straightforward. If the same saturation definition for HST/WFC3 was applied to the analogous JWST near-infrared detectors (i.e., the ones used by the NIRCam, NIRISS and NIRSpec instruments), they would cut the dynamical range of the detectors by about 75%, as the JWST detectors present higher non-linear responses. Similarly, the same statement applies to the MIRI detectors when compared to the Spitzer/IRAC ones. Ground-testing and new calibration methods, however, should allow for non-linearity corrections which are more precise than the ones performed for Spitzer/IRAC and HST/WFC3 (see, e.g., the case for NIRCam; Canipe et al. 2017). This suggests that the limitation for most TSOs — at least for differential measurements — will not be the instrument's non-linearity behavior, but most likely various other sources of systematic noise. The contributions of various systematics for TSOs are also different between JWST and its predecessors, as they depend on numerous design and performance parameters, such as pixel scale and sampling, pointing stability, and flat field stability.
Testing these complex and often correlated issues in a physically representative way is exceptionally challenging on the ground due to the required measurement precision. Full on-orbit characterization of JWST detectors is currently ongoing.
Target signal limits for TSOs
Several tests and analyses on the on-orbit performance of JWST detectors with fluence levels close to or at the saturation level are underway. Early results on those analyses suggest that for proposals that can afford it, targeting from 50% saturation up to the saturation limit in the JWST ETC should be on the safe side, with the former being the safest choice for Cycle 3. Some guidance is provided below based on data and analyses performed on publicly available TSO exposures obtained both during commissioning and Cycle 1.
For the JWST near-infrared detectors (NIRCam, NIRISS, and NIRSpec), fluences close to or at the saturation level of the instruments have been reached by at least 3 TSO programs with publicly available data: PID 1541 (Observation 1; PI: Espinoza with NIRISS/SOSS), PID 1366 (Observations 4 and 21; PI: Batalha with NIRspec/BOTS and NIRISS/SOSS, respectively), and PID 2589 (Observation 6; PI: Lim with NIRSpec/BOTS). Analyses performed on those programs suggest that degradation of the signal-to-noise ratio starts to be seen when counts reach roughly the saturation limits defined by the JWST ETC. One important caveat with those analyses is that the group number at which saturation occurs is very relevant. Integrations with a relatively large number of groups (> 5–10 groups) with pixels saturating in the last few groups are not as impactful on the final precision of TSOs as integrations with a small number of groups (< 5 groups) with pixels saturating in the last few groups.
When defining a number of groups per integration considering these signal limits, it is important to be aware as well that, due to the non-destructive reading of the JWST detectors, the JWST Pipeline can flag and disregard saturated groups, allowing users to use the remaining groups in the ramp to construct the final ramp. This adds an additional level of freedom in that if a number of groups saturate during the ramp, this does not imply the whole integration is lost. This is useful, e.g., for MIRI observations, where saturating in the last group can actually be helpful.
Users are encouraged to check the instrument-specific strategies that have been written by the instrument teams in the links provided below.
TSOs of very bright targets
If the observations being planned with JWST involve a bright object which is at the limit of what is observable with the observatory according to the signal limits defined by the Exposure Time Calculator, it might be wise to first revise the observational strategy being proposed and be absolutely sure that there is no other possible mode to observe the target in order to lower the maximum fluence attained by the observations. For example:
- If performing spectrophotometry: is it possible to use an instrument mode with a higher resolving power? This might help to spread the light of the object in more pixels.
- Is it possible to use smaller subarrays? These can help decrease the read times.
- Is it possible to use a faster readout mode for the detector?
- Is saturation happening in the spectral region of interest? If not, mild saturation might be an option; note, however, charge diffusion can occur for neighboring pixels (see, e.g., Plazas et al. 2018).
In general, if after revising all these points the target still saturates (i.e., the object achieves saturation in less than 3 groups), then observations could possibly enter a regime of risk in which the quality of the to-be-acquired data is uncertain, and recommendations to follow are strongly instrument-specific. Guidelines for these cases are provided in the instrument-specific pages, which are linked below so users can explore options in this direction.
Listed below are articles on recommended strategies relevant to TSOs of each of the instruments onboard JWST. These contain important information that must be reviewed by proposers when deciding to use a given instrument during proposal planning and submission.
- MIRI TSO Recommended Strategies
- NIRISS SOSS Recommended Strategies
- NIRCam TSO Recommended Strategies
- NIRSpec Detector Recommended Strategies
Canipe A., Robberto M., Hilbert B. 2017, JWST-STScI-005167, SM-12
A New Non-Linearity Correction Method for NIRCam
Hilbert B. 2008, Instrument Science Report WFC3 2008-39
WFC3 TV3 Testing: IR Channel Nonlinearity Correction
McCullough P. and MacKenty J., 2012, Instrument Science Report WFC3 2012-08
Considerations for Using Spatial Scans with WFC3
Plazas A. A., Shapiro C., Smith R., et al. 2018, PASP, 130, 065004
Laboratory Measurements of the Brighter-fatter Effect in an H2RG Infrared Detector
Stevenson K. B. and Eck W. 2019, Instrument Science Report WFC3 2019-13
Pre-Flashing WFC3/IR Time Series, Spatial Scan Observations
Stevenson K. B. and Fowler J. 2019, Instrument Science Report WFC3 2019-12
Analyzing Eight Years of Transiting Exoplanet Observations Using WFC3's Spatial Scan Monitor