How Dark the Sky: The JWST Backgrounds

For near-infrared wavelengths (0.6–5 μみゅーm), the astrophysical stray-light requirement was set to be less than the natural background from zodiacal dust. Stray-light modeling, mostly using reverse ray-tracing, with some limited use of forward ray-tracing, was performed by Ball Aerospace and was cross-checked using a different modeling code at NASA's Goddard Space Flight Center. These stray-light models incorporated detailed mechanical models of the structures and material properties of each surface (Lightsey et al. 2014; Lightsey 2016).

To define the stray-light requirement, a benchmark position on the sky was determined that would be a stressing case for stray light; it has a fairly high (−73°) ecliptic latitude, a zodiacal background about 1.2 times higher than the celestial minimum, and (the stressing part) positions the center of the Galaxy toward the front and above the observatory. ¹³ The celestial coordinates of this 1.2 minimum zodi pointing are given in Table 2. The stray-light requirements were set for this 1.2 minimum zodi field, as listed in Table 1, to be close to but lower than the zodiacal dust contribution at this reference pointing. As discussed below in more detail, the three main contributors to near-infrared stray light are (a) light scattered from particulates on the primary or secondary mirrors, (b) light scattering off the secondary mirror support struts, and (c) background from the so-called truant path, defined below.

Particulate scatter, especially off the primary and secondary mirrors, represents a significant fraction of the predicted stray-light background in the near-infrared (see the top of Figure 3). An extensive program of contamination control was instituted to maintain highly clean conditions throughout integration and test, including at the launch site (see Wooldridge et al. 2014a, 2014b). A cover for the entrance to the AOS was only removed when necessary for testing purposes, to keep the tertiary and fine-steering mirrors as well as the science instruments clean. The primary and secondary mirrors were visually inspected for particulates and spot cleaned as needed. This work continued throughout telescope and observatory integration and testing as well as at the launch site. Since in situ contamination measurements were not possible on the mirrors, surrogate contamination wafers were used to monitor the cleanliness of the environment throughout the telescope development. These wafers were placed as close to the flight telescope hardware as possible and used conservative assumptions when applying the measured contamination to the mirrors. For example, wafers were facing up, whereas the telescope optics were often perpendicular to the ground or facing toward the ground to mitigate the contamination accumulation. The stray-light models were used to derive a particulate percent area coverage (PAC) requirement of 1.5% for the primary mirror and 0.5% for the secondary mirror. Contamination monitoring and assumptions for redistribution during launch predicted an end-of-life primary mirror PAC of 0.754% and secondary mirror PAC of 0.149%—both levels are significantly below the requirements.

Zoom In Zoom Out Reset image size

Molecular contaminants were estimated at 32 Å for the primary mirror and 38 Å for the secondary mirror at launch via ellipsometry on optical witness samples that traveled in close proximity to the flight hardware. Analytical estimates of launch exposure and deposition during cooldown of the observatory predicted totals of 45 Å and 59 Å, respectively, at end-of-life for the sorts of molecules (e.g., organics) commonly referred to as nonvolatile residue (far below the 350 Å requirement). No spectral features have been detected in flight that would indicate unexpected deposition. Similarly, the carefully controlled cooldown of the observatory resulted in predictions of water-ice deposition below the 42 Å requirement on that component for each of these optics. There is no clear detection of spectral features from water ice on the optics in flight, although the breadth of the expected water-ice features and the uncertainties in the intrinsic throughput of the instruments preclude setting precise quantitative limits on the achieved deposition on individual optics. Upper limits on the total equivalent width of the (undetected) strong water-ice absorption feature in the 3.1 μみゅーm region indicate compliance with the total allocation for water-ice deposition on the summed elements of the optical train.

The coverings on the secondary mirror support struts were designed to provide a high degree of thermal stability for the secondary mirror positioned by the long struts. The cover design incorporated a strip of black Kapton on the upper side of each strut, with the rest of the strut profile covered by vapor deposited aluminum (VDA), including particularly the view to the warm sunshield. While meeting the thermal stability requirements, celestial sources can specularly reflect or scatter light from the undersides of the struts onto the primary mirror, which constitutes one of the larger sources of near-infrared background (see Figure 3).

An internal telescope pupil stop was located near the fine-steering mirror. This stop was oversized compared to the image of the primary mirror at the stop to allow the full primary mirror area transmission, and to provide a well-defined pupil for wave-front sensing. However, this left a gap between the stop and the primary mirror pupil image. Light from celestial sources behind the primary mirror that passes near the primary mirror periphery could reflect off the SM and pass through the gap at the stop, creating a stray-light source known as the truant path. A supported, black Kapton frill, extending outward around the periphery of the primary mirror, was installed to substantially block this truant-path light. Approximately 4 m² of frill area was projected to the internal pupil stop, as compared to 25 m² for the primary mirror. Particulates on the frill will scatter in the same manner as those on the primary mirror, and frill cleanliness is therefore also important. Two notches in the frill near the upper wing hinges were retained to ensure noninterference during wing deployments. Two other small gaps between the projected frill and the stop remain. These allowed for some margin for sunshield deployment, between the projected sunshield shadow line and the edge of the frill itself. The notches and gaps can be seen as truant-path sources at the bottom of Figure 3. Analysis showed that the residual truant-path contribution, while still one of the larger near-infrared background contributors, would be acceptable (see the bottom of Figure 3). Good alignment between the primary mirror image (and frill) at the fine-steering mirror mask also had to be achieved to provide the maximum suppression of the truant path.

For the mid-infrared, thermal emission from observatory surfaces was expected to establish the background level beyond about 15 μみゅーm; shortward of this, the thermal self-emission was expected to be lower than the zodiacal background. Observatory contributions to mid-infrared requirements of 3.9 and 200 MJy sr⁻¹ were set at 10 and 20 μみゅーm, compared to the expected in-field zodi contributions of 15 and 31 MJy sr⁻¹; see Table 1. To predict the mid-infrared background levels, reverse ray-tracing was performed, with structural temperatures provided by thermal models and emissivity and scatter characteristics from material properties.

By design, the large JWST sunshield (see the companion PASP paper by Menzel et al., submitted) keeps the primary and secondary mirrors in the shadow of the sunshield, which allowed them to cool to 35–55 K, establishing the background level for the longest operable wavelength of 28.5 μみゅーm. As expected, of the 18 primary mirror segments, the 4 segments closest to the sunshield have the highest temperatures and therefore contribute most to this long-wavelength background.

The sunshield itself, and particularly the top layer, L5 (the coldest), establishes the background level for shorter (∼15 μみゅーm) mid-infrared wavelengths. Modeling of the deployment established alignment tolerances for the relative positioning of the five layers with respect to each other. This was important to ensure that the sunshield background contribution was only from L5, the coolest layer, and not via scattering from an exposed edge of any warmer layers. In addition, each layer should see only the adjacent layer in order to maintain the appropriate temperature distribution throughout the sunshield assembly; for example, L5 should have a view only to L4, but not to the much warmer L3–L1 layers. A third consideration was maintaining the separation of the sunshield layers to permit sufficient escape of heat to meet the thermal requirements. Design and analysis of the propagation of heat and sunlight along the spreader bars, which cause and maintain the separation of the sunshield layers, ensured that they and their capping epaulets contributed only a low level of stray light.

Thermal emission from the warm region of the deployable tower assembly (DTA) and the core region of the sunshield near the DTA sets the background near 10 μみゅーm. The view from these areas to the secondary mirror is blocked by the bib, a lightweight, black Kapton assembly that is attached to the lower part of the primary mirror backplane and was deployed along with the DTA. The size and final design shape of the bib was a compromise to ensure proper deployment, but still provide a sufficiently small view to the core region to meet the mid-infrared stray-light requirements. Since modeling showed that at equilibrium, the upper portion of this bib would run near 60 K, a smaller, deployable flap was added to block direct visibility to this warm portion of the bib.

The ISIM Electronics Compartment (IEC) was placed within the cold region of the observatory to house critical electronics units operating at 273 K to 313 K. The science instruments had to be isolated from the IEC's thermal emission.

Measuring the background levels is one of the simplest measurements that can be made with JWST. In our experimentation, a simple median of imaging data agreed so closely with backgrounds estimated by first removing sources through sigma clipping that we chose to use the simple median as the calculation method.

All we can measure for any pointing is the summed spectrum of all the components listed in Section 2. While what we wish to know are the levels of specific components—the amount of astrophysical stray light, observatory stray light, and thermal self-emission—these photons add to the backgrounds just like the more pedestrian in-field zodiacal emission and ISM light. Therefore, the commissioning observations that were designed to characterize the stray light (PID 01448; PI Smith) observed the total background spectrum for multiple pointings, some that should be dominated by the zodiacal emission, some that should be dominated by Galactic ISM, as well as some relatively deep fields, and the 1.2 minimum zodi benchmark field. The idea was to first characterize that we were correctly measuring the (well-known) zodiacal and ISM emission, and then examine the backgrounds in the deep fields where the stray-light contribution should be a significant fraction of the total backgrounds.

In designing these commissioning activities, we were quite aware that while the telescope is susceptible to astrophysical stray light from many directions (see Figure 3), all that can be measured is the summed stray light from all those directions. The telescope's susceptibility to astrophysical stray light is directional, while the measured backgrounds are integrated over all these directions. We can therefore not hope to empirically create a new stray-light susceptibility map on orbit. Instead, for commissioning, our two goals were to test how well the predicted stray-light model was working, and determine a factor by which to scale the susceptibility map, to better match the on-orbit data, in time to support the Cycle 2 Call for Proposals.

Similarly, while in the models observatory stray light comes from scattering from many different parts of the observatory, all that we can measure is the total mid-infrared spectrum and its time variability. This is why we modeled the observatory stray light as we did, dividing the predicted backgrounds into 21 major components of the observatory—so that we could scale these component spectra up or down to better fit the observed total mid-infrared spectrum. Our goals for commissioning were to determine whether the mid-infrared backgrounds were consistent with prelaunch expectations, identify any exceedances, and determine the on-orbit thermal spectrum in time to support the Cycle 2 Call for Proposals.

In the longer term JWST science data can be harvested to monitor the levels of astrophysical stray light and observatory stray light. This monitoring might hypothetically uncover certain pointings where the astrophysical stray light is anomalously high, which might reveal any large deviations from the susceptibility model. Monitoring of observatory stray light might hypothetically reveal any degradation of the sunshield or baffling in the core region.

The stray-light commissioning program was designed as eight fields with a range of expected background levels across the sky, observed with shallow integrations of parallel NIRCam and MIRI observations for all fields. Additionally, five of these fields were observed with NIRISS, and three fields were observed with NIRSPEC. Table 2 lists the stray-light commissioning fields and the dates of observation. Observations were taken in MIRI/NIRCAM parallel mode, using three-point MIRI dithers, obtaining eight NIRCam bands (F070W, F115W, F150W, F277W, F356W, F444W, and F480W) with 3 × 150 s exposures in SHALLOW 4 mode, and also obtaining four MIRI bands (F560W, F770W, F1000W, and F1280W) with 3 × 139 s exposures in FASTR1 mode. Parallel mode resulted in a small offset between the MIRI and NIRCAM fields. As the fields were selected for background measurements (rather than specific features of an astronomical field), these offsets did not affect the data analysis or results. The NIRISS observations were single undithered 330 s exposures in NISRAPID readout mode for each of the F115W and F200W filters.

To supplement these dedicated stray-light observations, we added archival data covering three deep fields. We choose three data sets that should have low backgrounds, which were observed early in Cycle 1, and that have several-hour integrations in several NIRCam and NIRISS broadband filters, and where the data are already public. These deep fields are listed in Table 3.

The "MIRI Sky Background and Variation with Telescope Pointing" commissioning program measured the background seen by the MIRI imager through filters spanning the 5–28 μみゅーm wavelength range at the nominal hot and cold observatory pointing attitudes listed in Table 4. These observations were conducted as part of the commissioning test of thermal stability, in which the observatory was allowed to thermally equilibrate at a relatively hot attitude for 4 d, and then slewed to a cold attitude and again allowed to equilibrate for 7 d (see Section 5.4.4 of McElwain et al., submitted.) Data from a blank region of the sky in both attitudes were obtained using the F770W, F1280W, F1500W, F1800W, F2100W, and F2550W MIRI filters, with exposures of about 850 s in FASTR1 and a three-point dither using the cycling pattern. A 2 × 2 mosaic with 50% overlap in the F1800W filter was obtained to evaluate the presence of spatial structure in the thermal emission at longer wavelengths, and to correlate it with the measurements in all other filters. Table 4 gives pointing information and the dates of the observations.

To examine the variability of the long-wavelength background at 25.5 μみゅーm, we collect background measurements from ten commissioning, calibration, and Cycle 1 programs, as listed in Table 5.

We measured the JWST backgrounds from commissioning and early Cycle 1 data as follows.

For both NIRCam and NIRISS imaging data, we downloaded Level 1b data (uncalibrated files, labeled "UNCAL") from the JWST archive, and reduced the data using the first stage of the JWST pipeline with stock parameter files to create Level 2a (count-rate files, labeled "RATE") files in units of digital numbers per second (DN/s). The JWST calibration pipeline version was 1.7.2. The CRDS context using the PUB server was jwst_0967.pmap. ²⁰

To convert from measured count-rates (DN/s) into the flux-calibrated units of MJy/sr (a flux density surface brightness), we followed one path for NIRCam and a different path for NIRISS. The reason for this was that the JWST initial flux calibration of NIRCam was unsettled for much of this analysis; it converged in 2022 October. For NIRISS imaging data, we used stage 2 of the JWST pipeline (same pipeline version and CRDS server as above) with the stock parameter files to generate Level 2b files (calibrated files, labeled "CAL") in units of MJy/sr. We performed all photometry of NIRISS images on these flux-calibrated data products. We verified with the NIRISS instrument development team that this approach uses the latest available flux calibration. For NIRCam imaging data, we performed all photometry on the count-rate images (Level 2a), and then converted the measured count-rates into DN/s to surface brightness in MJy/sr by using the PHOTMJSR flux calibration conversion factors that were delivered to CRDS on 2022 October 3, ²¹ some of which are described in Boyer et al. (2022). To emphasize, we did not use the PHOTMJSR values from the headers because they had not yet been updated at the time this analysis was performed.

Similarly, for MIRI imaging, the Level 1b data, uncalibrated files, in units of DN were processed to Level 2a files (rate files in DN/s) and finally to 2b (CAL) flux-calibrated data, in units of MJy/sr, using the JWST calibration pipeline version, and the 1.6.2 CRDS context using the PUB server was jwst_0969.pmap. ²²

For imaging data, for each exposure and each detector, we took the median value of all illuminated pixels as the background. We then took the median value of these measurements over the multiple exposures per detector, with the standard deviation as the plotted error bar. In the case of NIRCam short wavelength, which has eight detectors, we label the median for each detector (labeled as module A or B, and numbers 1–4), and also plot the median over all detectors (plotted as purple stars.)

The JWST flux calibration at the time of this analysis was based on preliminary in-flight calibration data, not ground-test data. The NIRCam calibration used in this work was based on only two stars (Boyer et al. 2022). The MIRI calibration used in this work was based on 1–2 stars depending on the filter. This is compared to the JWST Cycle 1 calibration plan, which calls for at least nine stars (generally three from each of these categories: A dwarfs, hot stars, and solar analogs) to be observed by the end of cycle 1 (Gordon et al. 2022). For the calibration context used in this work, the flux calibration uncertainties are thought to be at the level of ∼5% for MIRI (K. Gordon 2023, private communication) and no more than 5% for NIRCam (M. Boyer 2023, private communication).

We predict the JWST background levels using the JWST_backgrounds tool; the inputs are the celestial coordinates and date of observation. For cases where observations were spread over multiple days, we plot the predicted backgrounds for each day the target was observed in the figures.

Five of the commissioning stray-light pointings were predicted to have low background levels (below 0.3 MJy sr⁻¹ at 1–5 μみゅーm). All five of these pointings tell a consistent story: at almost all wavelengths from 1–4.5 μみゅーm, the observed backgrounds are lower than expected. Further, the overlapping photometry from NIRCam and NIRISS at λらむだ = 1.1 and 2.0 μみゅーm agrees fairly well; this is encouraging because their flux calibrations were determined largely independently.

There is no evidence that the predicted in-field backgrounds (yellow curves in Figure 4) are incorrect, except at the shortest wavelengths (λらむだ < 1 μみゅーm). Indeed, Figure 4 shows that the NIRCam photometry closely matches the expected shape and normalization of the zodiacal emission at 3–5 μみゅーm, including the valley between the emission from reflected sunlight and the emission from sunlight that was thermally reprocessed by dust.

The simplest explanation for why the 1–3 μみゅーm points in Figure 9 lie below the prediction, is that there is less astrophysical stray light than expected. Scaling down the predicted stray-light component by 20% provides a better fit to the commissioning stray-light data. We will explore this scaling factor in more detail in the next subsections.

The shallow observations of the eight commissioning stray-light fields also show that the total background scales as expected with ecliptic latitude and Galactic latitude.

At the shortest wavelengths (λらむだ < 1.2 μみゅーm), there appears to be an issue with the background model: the shape of the zodiacal spectrum may be incorrect. This is most obvious for Abell 2744 (Figure 6 panel (A)), where the observed data are below the predictions from zodiacal emission (yellow curve); it also is obvious in the shortest wavelength data points of several of the commissioning stray-light fields (Figure 9). The model at these wavelengths was extrapolated from COBE/DIRBE data at 1.2 μみゅーm; it appears there may be an issue with the scattering function that was assumed in that extrapolation. We plan to investigate this issue and address it in the background model to support future proposal calls.

A second issue affecting short wavelengths (λらむだ ≤ 1.5 μみゅーm) is that the observed backgrounds in the Galactic bulge field are much lower than the predictions. We suspect that this arises because JWST resolves much of the faint background into individual stars.

Since the stray-light requirements were designed around a particularly stressing pointing, the 1.2 minimum zodi benchmark, we devote specific attention to the results in that field. Figure 5 compares the predicted amount of stray light for the 1.2 minimum zodi benchmark field (blue line) to the maximum amount of stray light that would have met requirements (green points at 2.0 and 3.0 μみゅーm). Subtracting the expected zodiacal and Galactic contributions, we infer for this field at 2.0 μみゅーm a stray-light level of 0.06 MJy sr⁻¹ from NIRCam and 0.035 MJy sr⁻¹ from NIRISS. Scaling the model-predicted value by 80% yields 0.048 MJy sr⁻¹, which splits the difference between the measured values from NIRCam and NIRISS. Similarly, at 3 μみゅーm, we infer a stray-light level of 0.03 from NIRCam compared to 0.051 MJy sr⁻¹ by scaling the stray-light prediction by 80%. These values are listed in Table 1. All these values are lower than the requirements.

Zoom In Zoom Out Reset image size

The 1.2 minimum zodi field was observed on day-of-year 123, a day when the backgrounds were predicted to be relatively low. According to the background model, had the field been observed on the day with the highest predicted background, the stray-light levels would have been 36% higher at 2 μみゅーm. This is still lower than stray-light requirements, which are 48% higher at 2 μみゅーm than what was predicted for the day of observation. Thus, given the observed measured amount of stray light on the day the field was observed, the stray-light model predicts that there is no observable day when the stray light would exceed the requirements in the benchmark field.

The stray-light commissioning observations were intended to test whether the stray light behaved as expected with ecliptic latitude and Galactic latitude, and whether the stray-light levels met requirements. As the previous section makes clear, the answers are Yes in both cases. Given this experimental design, these shallow (3 × 150 s) NIRCam observations are not ideal to precisely measure the amount of stray light because they suffer from significant 1/f noise. Therefore, we now examine the near-infrared backgrounds in three deep pointings from early in Cycle 1, with integration times of hours rather than minutes.

Figure 6 shows the measured backgrounds in the three deep fields, sorted from highest to lowest background level. These observations were not designed to measure backgrounds, but they certainly serve the purpose. The results are consistent with what was seen in the commissioning fields, with greater precision due to the long integrations: the backgrounds observed in almost all wavelengths are lower than predicted; NIRISS and NIRCam photometry agree extremely well for wavelengths of overlap; and at 4.1 and 4.4 μみゅーm where the zodiacal emission should be much brighter than any astrophysical stray-light contribution, the photometry is indeed quite close to (but generally lower than) the total predictions, which is exactly what one would expect if the photometry is accurate and the stray-light levels are lower than expected.

Zoom In Zoom Out Reset image size

Indeed, the North Ecliptic Pole pointing was chosen because it is one of the darkest parts of the infrared sky (Windhorst et al. 2022). As such, panel (C) of Figure 6 indicates the blackest the sky can appear to JWST.

While two of the three deep fields are centered on lensing clusters, with intracluster light that could potentially bias the backgrounds high, the similar backgrounds measured from the two NIRCam modules (one centered on the cluster in each case) indicates that the intracluster light does not significantly bias the measurement of the median background.

We scale the stray-light component of the backgrounds for each field to better match the photometry. The rough scaling factors are 0.4 for Abell 2744, 0.8 for SMACS0723, and 0.9 for NEP-TDF. These scalings are consistent with the 0.8 scaling factor found for the commissioning fields. The result is clear: over the full range 0.7 ≤ λらむだ ≤ 4.6 μみゅーm, astrophysical stray light has been successfully controlled such that it is less important than the unavoidable in-field astrophysical backgrounds. In other words, as it was designed to be, JWST is limited by the zodiacal light, rather than by stray light, for λらむだ < 5 μみゅーm.

We now consider JWST backgrounds for λらむだ > 5 μみゅーm. Figure 7, panel (A), shows the results of the thermal stability test. The photometry is almost identical for the hot and cold pointings.

Zoom In Zoom Out Reset image size

The backgrounds of long-wavelength MIRI images show low-frequency spatial structure at the 2% level that is independent of the pointing direction. In the future, this structure may be removed by subtracting a noiseless model of the structured background. Depending on the science goals, including the target's spatial extent, Cycle 1 observers should consider choosing a dither pattern or obtaining off-target background images at the same exposure time to correct the science images by subtraction. The total observing time will increase depending on the strategy employed. This spatial structure is a complexity of the backgrounds realized on-orbit; prelaunch analysis assumed the stray-light illumination would be uniform.

In panels (B) and (C) of Figure 7, we subtract the expected astrophysical components of the background from the measured photometry to estimate the residual thermal component. Overall, the observed thermal background spectrum closely matches the prelaunch predictions (blue and red curves in panel (C) of Figure 7), which is testament to the thermal engineering of JWST, in particular, the integrated thermal and optical modeling of a complex system.

There may be some excess emission at 6–15 μみゅーm compared to prelaunch predictions (see panel (B) of Figure 7). The thermal model is made of 16 nonzero subcomponents (colored thin curves) that we derived from the reverse ray-traced models. The observational constraints available are not nearly sufficient to rigorously fit the photometry with this model. Instead, we adjust the prelaunch end-of-life model in an ad hoc manner by adjusting the temperatures and effective areas of three of these components (the sunshield, the primary mirror, and the DTA tower) to better fit the observed residual backgrounds. This ad hoc revised thermal model is plotted as the black curve in panel (C) of Figure 7. This revised thermal curve currently is the best characterization of the JWST thermal spectrum for observation planning and scheduling.

JWST carried a level 1 requirement to provide the thermal environment needed such that the imaging science instruments would be limited by zodiacal light background over the wavelength range 1.7–10 μみゅーm. The green curve in panel (C) of Figure 7 shows that this is indeed the case for λらむだ ≤ 12.5 μみゅーm.

JWST carried level 2 requirements, listed in Table 1, for the mid-infrared thermal emission at 10 μみゅーm and 20 μみゅーm. Given current constraints on the thermal spectrum, we estimate the 10 μみゅーm thermal background as in the range 1–6 MJy sr⁻¹. With the current amount of precision available from subtraction of the astrophysical backgrounds, and the current uncertainty in the MIRI imaging flux calibrations, it is not yet clear whether the requirement is met at 10 μみゅーm; a larger number of archival data sets from Cycle 1 will need to be analyzed. At 20 μみゅーm, the requirement is clearly met at present, with an estimated equivalent monochromatic flux density of 155 ± 15 MJy sr⁻¹.

In the case of background-limited observations, integration time will scale with total background level. Panel (D) of Figure 7 shows the expected change in background levels and therefore exposure times due to predicted degradation of the sunshield over time (red curve), as well as the difference in exposure times when using the thermal spectrum that was adjusted to fit the commissioning data (black curve), rather than the thermal spectrum that was assumed prelaunch. This plot shows that for these two beginning-of-life and end-of-life models, predicted degradation of the sunshield over the mission lifetime should increase exposures times by up to 25%. This scaling would place the 20 μみゅーm measured background just below the requirement value at the end of life.

We also infer that, considering backgrounds alone, the exposure times for background-limited observations at λらむだ ∼ 10 μみゅーm may need to be 30% longer than prelaunch expectations, due to an apparent excess of short-wavelength emission. However, the better-than-expected overall throughput of MIRI (see Wright et al. 2023) may largely compensate for this effect.

These results are based on the latest calibrations and reference files available at the time of submitting this paper. We expect the calibration files to evolve and converge as Cycle 1 progresses, and therefore, small variations from the reported mid-infrared background measurements are expected.

The amount of solar energy input to the hot side of the sunshield must vary as the cosine of the pitch angle; it is not obvious how this variation in input energy translates into the temperatures of the cold side of the sunshield and the primary mirrors, both of which should dominate the mid-infrared backgrounds.

Figure 8 shows that the measured background variation in the F2550W filter is about 7%. We searched for but did not find a correlation between this background and the telescope orientation relative to the Sun. Thus, we conclude that the mid-infrared backgrounds are quite stable with time; they still vary enough that the variation must be accounted for in the data reduction. The MIRI team will trend the long-wavelength backgrounds throughout the life of the mission.

Zoom In Zoom Out Reset image size