NIRCam Short Wavelength Grism Time Series Observing Strategies
This page describes strategies for taking advantage of the improved bright source limits for the new (in Cycle 4) short wavelength (SW) grism capability in the NIRCam grism time-series observing mode, while balancing that with the higher throughput in the long wavelength channel and the increased data volume produced in this new mode (which uses all 5 detectors in module A rather than 3).
On this page
See also: NIRCam Grism Time Series, NIRCam Short Wavelength Grism Time Series, NIRCam Multistripe Subarrays for Grism Time Series Spectroscopy, NIRCam Grism Time-Series APT Template
The use of the Dispersed Hartmann Sensor (DHS) grisms in NIRCam's short wavelength channel is offered on a shared-risk basis due to limited time to obtain on-sky calibration observations prior to Cycle 4, and the use of new subarray readout operations. The capability is enabled by setting SW Channel Mode = GRISM in the NIRCam Grism Time Series template. Furthermore, additional readout patterns tailored to DHS time-series observations provide additional flexibility for balancing integration time and data volume.
Background regarding this new capability
Words in bold are GUI menus/
panels or data software packages;
bold italics are buttons in GUI
tools or package parameters.
The NIRCam grism time-series observing mode now includes the capability to collect slitless spectra in the short wavelength (SW) and long wavelength (LW) channels simultaneously, enhancing the mode to obtain time-series spectra spanning 0.7–5.0 μm, depending on the blocking filters selected in the two channels. The SW Dispersed Hartmann Sensor (DHS) grisms produce 8 spectra from the target separated on the detectors by about 125 pixels in the cross-dispersion direction. This new capability is described in detail in the article NIRCam Short Wavelength Grism Time Series.
To provide short frame times (and improved bright limits), new subarrays have been implemented that consist of multiple, spatially separated substripes (groups of rows) that exclude the ~80 rows of pixels between the DHS spectra in the SW channel. SW subarrays that sample 1, 2 or 4 of the DHS spectra per SW detector (corresponding to 2, 4 or 8 total DHS spectra, respectively) are available.
Available subarrays
The subarrays available when DHS grism spectra are being collected in the SW channel have names that indicate the number of detector rows in the subarray, the number of substripes the subarray is divided into, and the number of SW DHS spectra that are collected. In APT they are SUB41S1_2-SPECTRA, SUB82S2_4-SPECTRA, SUB164S4_8-SPECTRA, and SUB260S4_8-SPECTRA (see Table 1 below and NIRCam Short Wavelength Grism Time Series). In this article, these names will usually be shortened to SUB41S1, SUB82S2, SUB164S4 and SUB260S4, respectively, without loss of clarity. Additional detail about multistripe subarrays is provided in NIRCam Multistripe Subarrays for Grism Time Series Spectroscopy.
Challenges
The 8 DHS grism sub-apertures taken together only transmit 25% of the light collected by the JWST primary mirror, while the LW grism sees the entire collecting area, and the DHS produces multiple SW spectra while the LW grism produces a single spectrum. As a result, the count rates in the two channels are significantly imbalanced: SNR in the SW channel (even when all of the DHS spectra are co-added) is intrinsically lower than in the LW; conversely the saturation bright limit in the LW channel is significantly fainter than in the SW channel.
The first goal to designing successful observations in this new mode is to balance the SW SNR with the LW saturation limit. The second is to achieve that goal while also staying within the available downlink capacity of JWST. When the SW DHS grisms are used, data are collected on all 5 detectors in module A; when the DHS grisms aren't employed, data are collected on 3 detectors.
Multistripe subarray readout
As mentioned above and described in detail in the NIRCam Short Wavelength Grism Time Series and NIRCam Multistripe Subarrays for Grism Time Series Spectroscopy articles, in the LW channel a single substripe of pixels is read out each time one of the SW substripes is read out; if 2 or 4 substripes are defined, the LW substripe is read out 2× or 4× more frequently than any substripe in the SW channel. Figure 1 illustrates the timing of reads in the SW and LW channels. Repeat_Stripe is the parameter controlling whether the subarray is read out as spatially separate substripes (as in the SW channel) or as a single substripe to be read repeatedly (as in the LW channel), and is not user selectable.
Figure 1. Timing of reads within NIRCam multistripe subarrays in the SW and LW channels
Click on the figure for a larger view.
Schematic showing the timing of substripe resets and reads for SW and LW multistripe subarrays used in the NIRCam Grism Time Series template when SW Channel Mode = GRISM. The exposure pattern is BRIGHT1 (nFrames=1, nSkip=1), and there are 3 groups per integration. Solid blue triangles represent the frames that are saved and downlinked; blue rectangles with red borders are the skipped frames. In the short wavelength (SW) channel 4 spatially separated substripes are read out, while in the long wavelength (LW) channel a single substripe is read out 4× more often. The frame time, group time, and integration time in the two channels is identical, but the reads in the LW channel are oversampled by a factor of 4 relative to those in the SW channel because there are 4 substripes per subarray. Vertical red lines indicate the reset time for the 1st pixel in the subarrays. Sloping red dashed lines show how the start of the reset progresses through the substripes. Note that the LW substripe gets reset multiple times (once for each SW substripe). Here, neither channel experiences saturation of the signal during the integration ramp. Labels indicate the group time (tGroup) and frame time (tFrame), which are unchanged from normal subarray definitions: both are determined from the dimensions of the entire subarray (i.e., including all substripes). The stripe time, tstripe, is the time to read out a single substripe of the multistripe subarray. The integration time, tint, is also the same for multistripe and "normal" subarrays, but the figure reveals the subtlety that the time at which particular pixels in a subarray are read or reset depends on their position within the subarray (the same is true for FULL exposures). This fact may influence interpretation of the data since the SW spectrum in substripe 1 will be sampled at a slightly different time than a spectrum in a different substripe. Note that when there are 2 or more substripes, timing in the LW substripe is the same (modulo the oversampling) as in the last SW substripe.
New DHS-specific readout patterns
To mitigate both the data volume concerns (see next section) for transit observations, and provide added flexibility to avoid saturation in the LW channel, new readout patterns are available in this observing mode. Instructions for filling out the NIRCam Grism Time Series APT template using the available patterns are given below.
Table 1 provides the basic properties of these patterns, along with the time to acquire 1 group for each pattern as a function of subarray size. These patterns all have nFrames = 1, so they don't produce the extra data for Frame0 inherent in patterns with nFrames ≥ 2 (e.g., BRIGHT2, SHALLOW2). Additional detail is available in the NIRCam Short Wavelength Grism Time Series article. The "Group time" is the frame time × the sum of the saved and skipped frames for a pattern. In the final group of an integration the reset immediately follows the saved frame(s) (the final nSkip frames in the integration are themselves skipped).
Table 1. Subarrays for grism time-series observations using the DHS grisms in the SW channel with GRISMR in the LW channel
| Readout Pattern | nFrames | nSkip | SUB41S1_2-SPECTRA | SUB82S2_4-SPECTRA | SUB164S4_8-SPECTRA | SUB260S4_8-SPECTRA | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Group time [s] | Frame (Stripe) Time [s] | Group time [s] | Frame (Stripe) Time [s] | Group time [s] | Frame (Stripe) Time [s] | Group time [s] | Frame (Stripe) Time [s] | |||
| DHS3 | 1 | 2 | 0.66027 | 0.22009 (0.21484) | 1.30479 | 0.43493 (0.21484) | 2.59383 | 0.86461 (0.21484) | 4.10295 | 1.36765 (0.34060) |
| DHS4 | 1 | 3 | 0.88036 | 1.73972 | 3.45844 | 5.47060 | ||||
| DHS5 | 1 | 4 | 1.10045 | 2.17465 | 4.32305 | 6.83825 | ||||
| DHS6 | 1 | 5 | 1.32054 | 2.60958 | 5.18766 | 8.20590 | ||||
| DHS7 | 1 | 6 | 1.54063 | 2.04451 | 6.05227 | 9.57355 | ||||
Saturation limits
Taking DHS4 as the pattern with the shortest group time likely to be used for transits requiring exposures lasting 6 hours total (see "Data excess" section below), the time to save 2 groups is 5 frame times (reset at t=0, read, skip, skip, skip, read). The bright limits in Figure 5 at NIRCam Bright Source Limits are based on a RAPID integration with nGroups = 2, so if the DHS4 pattern is used instead the saturation limit would be -2.5 Log(5/2) ≈ 1 magnitude fainter in the SW channel. Similarly, in the LW channel (see Figure 4 in NIRCam Bright Source Limits), the saturation limit would be fainter by 1 magnitude for the same reason, but brighter by -2.5 Log(1.02/0.645) = -0.5 (for a total change of 0.5 mag fainter) due to the frame times for the DHS 2048 × 40 subarray vs. the 2048 × 64 subarray used in that figure.
Ultimately, saturation should be evaluated using the ETC. As explained earlier, the LW channel count rate will generally be much higher than in the SW channel. As a result, saturation typically only needs to be examined using a NIRCam LW grism time-series calculation. The ETC does, however, include a capability for checking for saturation in the brightest of the 8 DHS spectra if users wish to check that.
Data excess limitations on time-series duration
Time-series observations typically produce science data faster than any other type of observation. The volume of data that can be downlinked is finite as detailed in the article JWST Data Volume and Data Excess. APT calculates the volume of data produced for all observations and provides warnings when they stress the available downlink capacity.
Observations that produce a data excess more than 5 GB in excess of above the downlink capacity require significant effort to schedule, and proposers must justify why they cannot use a readout pattern that produces data at a lower rate (e.g., saturation of the signal at the minimum integration time for a longer pattern). Those that have a data excess of more than 15 GB generally will not be supported, and require a particularly strong science justification. Those in excess of the 30 GB data excess limit cannot be supported, and generate an error in APT.
Observers must use APT to verify the data excess for their proposed observations. For convenience, Table 2 provides a summary of the approximate rate at which excess data is produced when the exposure patterns and subarrays available for DHS time-series observations are used in short integrations (2–4 groups) appropriate for typical exoplanet host stars. It also gives the approximate duration of a time series exposure (Total Exposure Time in APT) that will exceed the 5 GB data excess threshold (which produces a warning in APT). The DHS7 exposure pattern has a negative data excess rate for all subarrays and values of nGroups (arbitrarily long exposures will not generate a data excess), and so aren't included in the table.
Table 2. Estimated time-series duration that exceeds the 5 GB data excess threshold vs. exposure pattern, nGroups and subarray
| Data Excess Rate (GB/hr) | Hours to Exceed Lower (5GB) Data Excess Threshold† | |||||||
|---|---|---|---|---|---|---|---|---|---|
| nGroups | SUB41S1 | SUB82S2 | SUB164S4 | SUB260S4 | SUB41S1 | SUB82S2 | SUB164S4 | SUB260S4 | |
| RAPID | 2 | 6.02 | 6.13 | 6.19 | 6.21 | 0.83 | 0.82 | 0.81 | 0.80 |
| RAPID | 3 | 7.16 | 7.29 | 7.35 | 7.38 | 0.70 | 0.69 | 0.68 | 0.68 |
| BRIGHT1 | 2, 3, 4 | 3.73 | 3.82 | 3.86 | 3.88 | 1.34 | 1.31 | 1.30 | 1.29 |
| DHS3 | 2 | 2.36 | 2.43 | 2.46 | 2.47 | 2.12 | 2.06 | 2.03 | 2.02 |
| DHS3 | 3 | 2.02 | 2.08 | 2.11 | 2.12 | 2.48 | 2.41 | 2.37 | 2.35 |
| DHS3 | 4 | 1.86 | 1.92 | 1.95 | 1.96 | 2.69 | 2.60 | 2.56 | 2.54 |
| DHS4 | 2 | 1.44 | 1.50 | 1.53 | 1.54 | 3.46 | 3.33 | 3.27 | 3.25 |
| DHS4 | 3 | 0.99 | 1.04 | 1.06 | 1.07 | 5.07 | 4.82 | 4.71 | 4.66 |
| DHS4 | 4 | 0.79 | 0.84 | 0.86 | 0.87 | 6.33 | 5.97 | 5.79 | 5.73 |
| DHS5 | 2 | 0.79 | 0.84 | 0.86 | 0.87 | 6.33 | 5.97 | 5.79 | 5.73 |
| DHS5 | 3 | 0.30 | 0.34 | 0.36 | 0.37 | 16.69 | 14.62 | 13.76 | 13.44 |
| DHS5 | 4 | 0.10 | 0.14 | 0.16 | 0.17 | 51.20 | 36.35 | 31.67 | 30.14 |
| DHS6 | 2 | 0.30 | 0.34 | 0.36 | 0.37 | 16.69 | 14.62 | 13.76 | 13.44 |
| DHS6 | 3 | -0.19 | -0.15 | -0.14 | -0.13 | inf | inf | inf | inf |
| DHS7 | any | < 0 | < 0 | < 0 | < 0 | inf | inf | inf | inf |
† The time to exceed the middle threshold (15 GB) is a factor of 3 longer. Observations exceeding that higher threshold are extremely rarely approved for execution, and require a very strong scientific justification.
APT 2025.3 overestimates data excess for the DHS3, DHS4, DHS5, DHS6 and DHS7 patterns by a factor of approximately (NGROUPS+1)/NGROUPS. This will be addressed in the next major release (APT 2025.5 in support of the Cycle 5 call for proposals).
Determining feasibility of an observation
To determine whether a particular time-series observation is feasible, observers must balance two things. First, the total duration of the time-series exposure(s) places a limit on the exposure pattern that can be used without exceeding the data excess limits (see Table 2). Second, the saturation limit for the target (in particular in the LW channel) places an upper limit on integration time, and therefore which exposure patterns can be used. Users will likely have to iterate to find combinations of exposure patterns and subarrays that simultaneously satisfy the data excess and saturation limits described above. Observations of particularly bright host stars with slow transits may well not be feasible using this capability.