Part of the NIRCam GTO program includes deep field imaging of the GOODS-S field (Figure 1). The goal is to study galaxy evolution from the first steps (z > 10) through the end of the dark ages (7 < z < 9) and through the epoch of galaxy assembly (2 < z < 6). Specific objectives include:
This science use case describes what data is required to achieve the program's objectives, including discussion of the required wavelengths, spatial resolution, sensitivity, and spatial coverage. The Exposure Time Calculator (ETC) section describes how to estimate the necessary imaging depth, and the Astronomer's Proposal Tool (APT) section describes how to design the observation parameters. This program was presented as part of the JWST Proposal Planning Workshop at STScI in May 2017. Presentations can be found online. Note that the program presented here is slightly different from the one presented at the workshop.
NIRCam Imaging (0.6–5.0 μm) is the primary observing mode for this program, with MIRI imaging (5.6–25.5 μm) in parallel. The spatial area required is determined by the expected density of galaxies at a specified redshift. For a sample of 10 galaxies at z = 12, simulations from Crowley et al. (2017) show that 167 square arcmin and a depth of 100 ksec is required. A sample of 100 galaxies at z = 7, requires 33 square arcmin at 10 ksec depth. Here, the goal is to cover approximately 25 sq. arcmin at 30–50 ksec depth.
Dithers are highly recommended for NIRCam imaging to protect against cosmic rays, improve the angular resolution, cover detector gaps, compensate for detector artifacts, and improve image quality. NIRCam is Nyquist sampled at 2 and 4 μm, so subpixel dithers are required for good spatial sampling for most of the wavelength coverage. Larger NIRCam dithers are also required to cover the detector gaps (see below). MIRI is oversampled over most of its wavelength range and Nyquist sampled at F770W, so MIRI subpixel sampling is not required.
The NIRCam observations will be deeper than the Hubble observations and have superior spatial resolution at λ > 1 μm. MIRI limits are 10x those achieved with Spitzer and will have much better spatial resolution.
Exposure Time Calculator
With ETC, we will compute the signal-to-noise ratio (S/N) achieved for a z = 8 galaxy and a z = 1.5 galaxy in all of the desired filters. For detailed help with using the ETC, see the online help pages.
Setting up a scene
In the "Scene and Sources" tab, you can edit the sources within your scene. In the "Scene Editor" box, there are tabs for setting the source continuum, shape, and flux renormalization. For the continuum, we'll use the galaxy spectra from Brown et al. (2014).
Once the sources are created, add them both to the Scene and give them offsets so they do not overlap. There is also an option to set the rotation.
With only two sources, ETC will use those sources to define the edges of the frame and will therefore give warnings that the aperture is partially outside of the field of view. One way to avoid this warning is to add more sources to the field with larger offsets so that the galaxies are well inside the field of view. Figure 3 shows an example.
Setting up a calculation
The signal-to-noise estimates are carried out under the "Calculations" tab. This program requires NIRCam Short-Wavelength (SW) imaging, NIRCam Long-Wavelength (LW) Imaging, and MIRI Imaging calculations. For each calculation, we must specify the Scene (which we defined above), the Background, the Instrument Setup, the Detector Setup, and the Strategy. The aim will be to detect both galaxies with exposure times from about 30 ks to 50 ks.
In the "Backgrounds" tab, there is the option to choose a pre-calculated background level (low, medium, or high), or to choose the background for the specific sky coordinates and date for the observation. Here, we use the "medium" background configuration for all calculations (50th percentile).
In this tab, specify the filter used in the calculation. This program uses 10 filters: 4 NIRCam SW, 5 NIRCam LW, and 1 MIRI filter, and one calculation is needed for each. These can be set up separately or users can set up one and then choose to "Expand Filters" from the "Expand" menu at the top of the ETC window. This option copies the selected calculation once for every available filter. The filters considered here are: F090W, F115W, F150W, F200W, F277W, F356W, F444W, F770W, F335M, and F410M (see above).
A galaxy at redshift 8 has a sharp drop off in flux at λ < 1 μm (see Figure 2). At λ > 4 μm, the S/N is limited by a combination of the lower filter throughput and the higher background. The exposure times in the short- and long-wavelength filters should therefore be longer than those at 1 μm < λ < 4 μm. Here, we will test the expected S/N for the galaxies in our scene using about 30 ks in F200W, F277W, F335M, and F356W and about 50 ks in F090W, F115W, F150W, F410M, and F444W. The F770W MIRI filter will be observed in parallel with all observations for a total of about 200 ks (see APT section), so we compute the S/N for that exposure time. To set up these exposure times in ETC, we need to choose the readout pattern, including the number of groups, integrations, and exposures.
NIRCam continuously and non-destructively samples the detectors every 10.7s (one "Frame") when using the full array. To limit the total data volume, readout patterns average multiple frames (a "Group") and skips others. The length of an "Integration" is determined by the number of groups specified. At the end of an integration, the detector resets. Multiple integrations can be combined into a single "Exposure". These terms are illustrated on the Readout Pattern documentation page. When choosing a readout pattern, this program must balance a few competing factors:
The best choice for this program is the DEEP8 readout pattern, which is designed for the deepest imaging. With 7-9 GROUPS/INT (depending on the filter) and 2 INTEGRATIONS/EXP, this program can achieve the required depth while mitigating effects from cosmic rays. Dithers also increase the depth; these can be mimicked in the ETC by increasing the number of Exposures. This program will use the 9 dithers total (see APT section), which are designed to cover the SW detector gaps and improve the sampling of the PSF. To achieve 30 ks, we set Groups = 9, Integrations = 2, and Exposures = 9. To achieve 50 ks, we increase the number of integrations to 3. The detector setups are summarized in Table 1.
MIRI also samples its detector non-destructively and continuously, this time every 2.8 s. This program uses the FAST readout pattern, which is recommended for background-limited observations. The number of Groups, Integrations, and Exposures are set to match the NIRCam exposure times because the MIRI observations will be done in parallel. NIRCam observes a SW and a LW filter simultaneously, so the total amount of exposure time available for MIRI is about half the total NIRCam exposure time, or a bit more or less depending on how the filters are paired. Here, we show the S/N for a total exposure time around 200 ks.
Table 1. Summary of ETC detector setup and resulting S/N
The "Strategy" tab specifies the aperture information used to extract the source flux and compute the S/N. Choose the option to center the aperture on the source, and select the source of interest from the pull-down menu (either the z = 8 or z = 1.5 galaxy). The Aperture radius should be large enough to encompass the source, and no smaller than ≈1.5–2x the FWHM of the PSF. Too-small apertures result in overly optimistic S/N estimates. The NIRCam PSF FWHM increases with wavelength. For the F770W MIRI filter, the PSF FWHM is ≈0.25″. Table 1 summarizes the selected aperture sizes. For the background, there is the option to set a background sky annulus or to use a noiseless sky background. To avoid flux from other sources in this scene, we use the noiseless sky background option.
Setting up the proposal in APT
Instructions on how to start a proposal in APT can be found in the online help pages. This program uses the NIRCam Imaging APT Template. All information must be filled out before submitting the proposal, but is not necessary to fill out all "Proposal Information" while setting up the detector specifications. Note that red x's will appear in the "Tree Editor" on the left side of the GUI until all required information is filled out.
Specify a target
This program is targeting the GOODS-S field with a 2x2 NIRCam mosaic. There are a few options for entering a target in APT; we will specify an individual Fixed Target. To start, create a new Fixed Target at RA = 03h 32m 42.7s, Dec = -27d 47m 59.7s. By setting this specific position (instead of the center of the GOODS-S field), the NIRCam and MIRI observations will fall completely within the CANDLES coverage (Figure 1), given a specific position angle (see Special Requirements).
Defining NIRCam imaging observations in APT
To get started, create a new observation inside an Observation Folder. From there, select NIRCam as the Prime Instrument and 'NIRCam Imaging' as the Template. Check the Coordinated Parallel box and select "NIRCam-MIRI Imaging". Finally, choose the target defined above from the pull-down menu. When editing the "Observation", the tabs on the lower half of the GUI screen contain the detector specifications.
To set up the NIRCam imaging, proceed to the "NIRCam Imaging" tab (leftmost tab on the lower half of the APT screen). Since this program will be imaging a large area, choose ALL from the "Module" menu and the FULL from the "Subarray" menu, which together encompass all pixels in all 10 NIRCam detectors, providing the largest possible spatial coverage. See the NIRCam field of view page for details. For smaller areas, users can select individual modules and/or subarrays.
Some considerations when designing this APT program
For deep imaging that uses all 10 NIRCam detectors plus parallels, data volume can add up quickly. For this particular program, the data volume is high enough that the observations must be split into an individual observation for each filter instead of including all filters within a single observation in the "Filters" box. This increases the overheads slightly, but makes up for some of the extra overhead by allowing for more flexibility in mosaicking, as described in the next paragraph.
The filter wheels on JWST have limited lifetimes and users must take care to minimize the number of filter wheel moves required by their programs. Normally, this means ensuring that dithers are used in place of mosaics. This is because dither patterns are executed in their entirety before changing to the next filter, while mosaics cycle through all filters at a single position, potentially requiring a large number of filter wheel moves (see the NIRCam Dithers and Mosaics Overview Page). Since this program has divided its observations up by filter, this is not a concern: each observation will include all dithers and mosaics for a single filter, requiring no filter wheel moves within the observation. Altogether, the program requires only up to 7 filter wheel moves, or 5 moves if the observations are executed in sequence.
The primary dither pattern employed here is the INTRAMODULE pattern, which is designed to cover detector gaps in the SW channel. Three INTRAMODULE dithers is the minimum required to cover all SW gaps. This program also uses a 3-point subpixel dither pattern to improve the spatial resolution of the images. Since this program employs coordinated parallels with MIRI, it uses a coordinated parallel subpixel dither pattern, which is designed to provide good sampling for both instruments when used in parallel. Here, we use the 3-POINT-WITH-MIRI-F770W option since all MIRI observations here will be with F770W. Note the coordinate parallel subpixel dither options are not available unless the coordinated parallel box is checked in the Observation.
In the "Filters" box, specify the required filters, readout patterns, and exposures. NIRCam has a short-wavelength and a long-wavelength channel that produce simultaneous imaging (via a dichroic) over the same field-of-view. Programs can therefore select one filter for each channel for each exposure sequence in the Filters box, and both will be observed with the identical readout patterns and total exposure time. Filters used in this program are summarized in Table 2.
NIRCam readout patterns
As described in the ETC section above, this program uses the DEEP8 readout pattern, which is designed for the deepest imaging. To achieve the required depth, this program uses 7-9 Groups/Int (depending on the filter) and 2 Integrations/Exp. The Total Exposure Time displayed at the end of the row in the Filters box includes all integrations and dithers, and is set up to achieve about 30 ks in some filters and 50 ks in others. See the ETC section above for an explanation about the selected exposure times. Since some observations had to be split to allow for greater data volume, the combination of groups and integrations is slightly different than what was used in the ETC section, but the resulting total exposure times are similar. Readout patterns are summarized in Table 2.
The parallel MIRI observations are specified in the "MIRI Imaging" tab. The program uses only the F770W filter and the FAST readout pattern. For this program, FAST is preferred to the SLOW pattern because these observations are background limited. The number of groups and integrations are set to match the NIRCam exposure times (Table 2).
Table 2. Summary of detector parameters in APT
To cover the necessary spatial area, this program uses a mosaic with 2 rows and 2 columns. The "Row Overlap" is set to 20% and the "Column Overlap" is set to 78%. The column overlap ensures 2 things: (1) that the large gap between the NIRCam modules is covered, and (2) that the MIRI coverage is continuous. The resulting NIRCam mosaic has 2 wide stripes of increased (2x) depth where the modules overlap. Figure 6 shows the Aladin visualization. The NIRCam mosaic covers approximately 25 square arcmin, and the MIRI mosaic covers approximately 7 square arcmin.
To ensure that the NIRCam and MIRI maps fall within the CANDLES region and includes the ACS Ultra Deep Field, we must restrict the position angle of the observations. In the "Special Requirements" tab, add a PA Range of 280°–300° for each of the observations. It is important to set the PA range for each observation instead of using the Same PA Link option – because the visits are so long (see below), it is impossible for them all to be observed at the exact same position angle.
Final data volume and schedulability
The data volume per visit is restricted to 58 GB maximum, which is the limit of the solid state recorder onboard the telescope (not including some space reserved for telemetry data). JWST downlinks in two 4-hour windows each day. Each window can transfer 28 GB of data. Individual visits can exceed 28 GB, but these may be difficult to schedule since they require 2 downlink windows to transfer all of the visit data to the ground. This program has 4 visits per observation (28 visits total for 7 observations), and each visit requires 36–46 GB of data and 8–12 hours of observing time. These large data requirements, long visits, and restrictive position angles result in some difficulty in scheduling. The position angle requirement restricts the observations to a 32 day window in September/October 2019.
Final program length
The total program length is 274.34 hours, 218.2 hours of which is NIRCam exposure time. The efficiency is therefore 79.5%. The final APT file can be downloaded at: Final_DeepField.aptx.