NIRCam Grism Time Series

JWST NIRCam's grism time-series observing mode performs rapid spectroscopic (R ~ 1,600 at 4µm) monitoring of bright, time-variable sources at 2.4–5.0 µm.

On this page

See also: NIRCam Grism Time-Series APT Template

The NIRCam grism time-series observing mode is designed to monitor bright, isolated, time-varying sources at 2.4–5.0 µm with spectroscopic resolving power R ~ 1,600 at 4µm. It is one of 2 modes available for NIRCam time-series observations (TSO), the other being time-series imaging. Both of these modes allow very long exposures (>50 hours with some interruptions and limitations) and disallow dithering and mosaics. (A separate non-TSO grism mode is available for wide field grism observations.)

In grism time-series mode, the module A "grism R" is used to disperse the target's spectrum along (parallel to) detector rows. The grism is used in conjunction with one of 4 wide filters in the long wavelength channel (2.4–5.0 µm): F277W, F322W2, F356W, and F444W.

Simultaneous short wavelength imaging is obtained with a weak lens to defocus the image. This choice offers saturation limits similar to the long wavelength grism for a given integration time (which must be identical for both wavelengths).

Integration times may be shortened for rapid cadence monitoring by using detector subarrays and/or multiple detector outputs. The shortest integration times enable observations without saturation of some stars visible to the unaided eye (~5th magnitude).

Figure 1. Sample NIRCam Grism R + F444W data

Sample NIRCam Grism R + F444W data, obtained during Integrated Science Instrument Module (ISIM) testing at Goddard. The bright test source included a CO2 absorption feature at ~4.25 µm. Wavelength increases left to right.


Table 1. Available combinations of filters, weak lenses, and grism

Short wavelength channel (0.6–2.3 µm)
Long wavelength channel (2.4–5.0 µm)

Weak lens offering +8 or +4 waves defocus
combined with a filter among these allowed combinations:

  • WLP8 + F070W
  • WLP8 + F140M
  • WLP8 + F182M
  • WLP8 + F187N
  • WLP8 + F210M
  • WLP8 + F212N
  • WLP4+F212N2

Grism R plus a wide filter:

  • F277W
  • F322W2
  • F356W
  • F444W

The WLP4 weak lens is joined with a F212N2 narrowband filter with a 2.3% bandpass, to form a single optical element. (Note that this F212N2 filter is wider than the 1% bandpass F212N filter in the filter wheel that can be paired with WLP8.)

Figure 2. Throughput response for NIRCam grism and long wavelength filters (1st order)

First order throughputs of the module A and B grism and the 4 wide filters (module A is shown, module B is similar) for grism time series, including all JWST and NIRCam optics and detector quantum efficiencies. The grism throughput must be multiplied with that of a selected filter (not shown). The module B grisms are AR coated on only one side and therefore have throughputs ∼33% lower than the module A grisms.
Figure 3. Throughput response for NIRCam grism and long wavelength filters (2nd order)

Second order throughputs of the module A and B grism and the 4 wide filters (module A is shown, module B is similar) for grism time series, including all JWST and NIRCam optics and detector quantum efficiencies. The grism throughput must be multiplied with that of a selected filter (not shown).
Grism throughput tables for both modules and for both 1st and 2nd orders can be downloaded: here. Note that these are grism-only throughputs. Filter-only throughputs, can be dowloaded here. To obtain the full system throughputs, the grism-only throughputs need to be interpolated at the same wavelengths as the filter-only throughputs and multiplied by the filter-only throughputs. 

Subarrays and readout times

Bright science targets require short integration times to avoid saturation. To shorten detector readout times, a subset of the detector rows may be read out (and the rest discarded). These subarrays will contain the entire spectrum for isolated compact science targets. Module A available subarrays for this mode contain 64, 128, or 256 pixel rows. Each of the subarrays include all 2,048 pixel columns (2,040 sensitive to light + 8 reference). The full 2048 × 2048 pixel array may be read out instead if desired.

The detector may be read out either through 4 outputs simultaneously (to speed readout and minimize saturation) or a single output (to reduce the data volume rate). This is the only NIRCam observing mode to offer this option; other modes always use 4 outputs to read the full detector (and a single output for subarrays). When 4 outputs are chosen, the detector output is split into 4 columns, or "stripes," each 512 pixels wide that are read simultaneously.

Table 2.  Detector read out times for available subarrays and numbers of outputs

Readout time (s)
1 output
Readout time (s)
4 outputs


Each spectrum is dispersed by 1 nm/pix. The undeviated wavelength is 3.95 µm.

As shown in Figure 4, for wavelengths below 4 µm (F277W, F322W2, F356W), sources are positioned near the right of the detector (x = 1581), and the spectra disperse to the left (towards shorter wavelengths). Above 4 µm (F444W), a different reference position is used (x = 952) because the source disperses to the right (towards longer wavelengths).

Vertically, the sources are positioned at y = 35 on the long wavelength detectors when using the 64, 128, and 256 subarrays, and at y = 280 when using the full array. At short wavelengths, the defocused images land in the vertical center of the corresponding subarrays, which traverse two short wave detectors (with a 4″–5″ gap).

Figure 4. Grism dispersions within a grism detector subarray

Footprint of the spectra in grism time-series mode. Red boxes indicate the 256 pixels subarray for this mode (to scale). The footprints for the different filters are spaced vertically for clarity (the actual subarray on the detector is the same for the 4 filters). The reference position is marked with an X symbol and a vertical dotted line. The reference position corresponds to the location of the undeflected wavelength of 3.95 μm, or alternatively would correspond to the direct image position if the grism was not in the beam. Filters F277W, F322W2 and F356W share the same reference position, while the F444W reference position is shifted to smaller x values in the science reference frame. The thick portion of the footprint corresponds to wavelengths for which the total throughput is larger than 50% of the maximum throughput for that setting. The thinner part of the footprint corresponds to wavelengths for which the total throughput is larger than 1% of the maximum throughput for that setting.

This figure refers to pre-mission expectations and will be updated when further analysis of calibration data is completed.


The reddest part of the F444W footprint (above 5.03 μm, where the throughput drops below 20% of the maximum) extends beyond the detector. The loss of data in this low-throughput region is balanced by 2 factors. First, the high throughput portion of the spectrum has more overlap with the corresponding portions for the other 3 filters. This allows a better characterization of systematics at the subpixel level in the region in common. Second, this choice of the reference position would allow a larger fraction of the short wavelength DHS (Dispersed Hartmann Sensor) spectra to land on the SW detector [see Schlawin et al. 2017]. SW DHS spectroscopy in parallel with LW grism time series is not enabled for cycle 1 but may be a future upscope of NIRCam capabilities. Maintaining the same reference positions for the LW grism time series with and without parallel DHS spectroscopy is advantageous as it would provide better overall LW F444W grism time-series characterization.

Saturation limits

See also: NIRCam Bright Source Limits

Based on preliminary estimates, A-type main sequence stars as bright as K ~ 4.5 (Vega mag) may be observed with the NIRCam grism without saturating the module A detectors at any wavelength when using the smallest subarray (2048 × 64 pixels), stripe mode (4 outputs), and a short integration time of 0.68 s (2 reads). Still brighter stars may be observed at the longest wavelengths, as shown in Figure 5.

Figure 5. Grism saturation limits in the 2048 × 64 pixel subarray
Approximate grism saturation limits in K-band Vega magnitudes for 3 stellar types (A0V, G2V, M2V) in the module A 2048 × 64 pixel subarray with stripe mode readout (4 outputs), assuming a detector reset and 2 reads (0.68 s integration). Results are from the Exposure Time Calculator (ETC) v2.0. Please use the ETC to obtain saturation limits for your proposed observations.
For larger subarrays and/or a single detector output, the minimum integration times increase, and a K ~ 4.6 Vega mag star would saturate the detector. Table 3 shows approximate saturation limits for the various subarrays, again assuming 2 detector reads between resets (RAPID readout pattern). These limits are given for 2.7 µm, the wavelength most prone to saturation. F277W and F322W2 observations will experience such saturation. Longer wavelength observations may observe somewhat brighter stars without saturating. Please consult the Exposure Time Calculator (ETC).

Table 3. Subarray saturation limits


Approx. saturation limit
(K Vega mag)
1 output

Approx. saturation limit
(K Vega mag)
4 outputs

Tilt events

The NIRCam time-series imaging and grism modes are the only modes currently supported that allow a simultaneous high-cadence wavefront sensing along with the time series. An analysis of commissioning observations, covering a transit of HAT-P-14b (PID 1442) have revealed a jump in the target flux around the mid-transit in both the SW (weak lens) and LW (grism, band-integrated) light curves (Figure 6 and 7). The jumps in flux correspond to changes in the mirror surface, known as “tilt events”, which have been hypothesized to be structural microdynamics in the telescope that may occur when stresses in the backplane structure behind the mirrors are suddenly relaxed. As these stresses are released, the frequency of tilt events is expected to decrease in time. 

The short wavelength Weak Lens +8 PSF is part of the wavefront sensing tools aboard JWST and is thus highly sensitive to changes in the mirror surface. Identification of tilt events can be done by choosing an arbitrary reference image from the SW time series of rate images and dividing the rate from all other integrations by it. The tilt events show up as clear wave patterns in these differential images due to changes in the mirror surface (Figure 6). For the HAT-P-14b data, the mirror changes appear on a timescale faster than the 27 second cadence of integrations. Photometry analysis on the pairwise-subtracted group images shows that each of the two tilt events happen on a timescale as fast or faster than the 1.351 second time between BRIGHT2 detector groups, which contain 2 frames each. Tilt events can also be sensed from the LW grism time series by measuring the FWHM as a function of time (Figure 7). Observers making use of the NIRCam time-series and grism time-series modes are strongly encouraged to examine the data products described above for their observations and routinely identify tilt events, if any.

Figure 6. Time-series imaging and grism transit light curves with tilt events
The light curves exhibited jumps most noticeably in the long wavelength spectroscopy, shown here as a broadband time series. One way to detect these jumps is by measuring the FWHM of the long wavelength spectroscopy.

Figure 7. Tilt events in the SW time series imaging
Weak Lens ratio images before and after two tilt events identified in the commissioning transit observation of HAT-P-14b.

Detector settling time

Transit observations of exoplanet HAT-P-14b, obtained during commissioning (PID 1442) suggest that the settling time for the short wavelength channel is in the range from 5 to 15 minutes and even shorter for the long wavelength channel. Observers are advised to continue assume the more conservative pre-flight settling time of 30 minutes in the preparation of their transit observations until a further notice.

SW and LW timing synchronization

The timing of the short and long wavelength channels can be offset from each other from zero seconds to the frame time, which can be smaller than the group time.


Beatty et al. in prep (reference will be updated when ready)

Greene, T. et al. 2017, JATIS, 035001

λ = 2.4 to 5 μm spectroscopy with the James Webb Space Telescope Near-Infrared Camera

Rigby, J. et al 2022, eprint arXiv:2207.05632 
Characterization of JWST science performance from commissioning

Schlawin et al. 2017, PASP, 129, 971
Two NIRCam Channels are Better than One: How JWST Can Do More Science with NIRCam’s Short-wavelength Dispersed Hartmann Sensor

Schlawin et al. 2022 (PASP in rev., reference will be updated when ready)

Notable updates
    Updated Figures 2 & 3 with the in-flight throughputs

    ETC v2.0 updated grism saturation limits Figure 5 
    Added subsections "Tilt events," "Detector settling time," and "SW and LW timing synchronization"
    Updated Table 3 using ETC v2.0

    Updated grism saturation limits (ETC v1.5); Table 3 corrected

    Corrected frame times for Grism time series Noutputs = 1

    Edited information to match recent paper (Greene et al. 2017)
Originally published