NASA SPHEREx: NIR All-Sky Spectral Explorer

Updated 18 January 2026

NASA SPHEREx is an all-sky near-infrared spectrophotometric survey satellite that collects data across 102 spectral channels from 0.75–5.0 μm to probe cosmology, galaxy evolution, and astrochemistry.
It employs a flight-proven Ball Aerospace platform with advanced passive cooling and thermal management, ensuring precise optics stability and 0.15″ rms pointing during exposures.
The mission features automated survey planning and robust data pipelines that deliver calibrated spectral products with high voxel completeness, supporting legacy calibration and cross-mission studies.

The NASA SPHEREx (Spectro-Photometer for the History of the Universe, Epoch of Reionization, and Ices Explorer) spacecraft is an Explorer-class, all-sky near-infrared spectrophotometric survey satellite launched into a Sun-synchronous low-Earth orbit (LEO) on March 12, 2025. SPHEREx executes the first all-sky spectral survey from 0.75–5.0 μm at low spectral resolution, across 102 bands, creating an astronomical dataset to address key science goals in cosmology, galaxy evolution, and astrochemistry. SPHEREx employs a sequence-optimized, fully automated surveying approach, underpinned by advanced software and robust spacecraft engineering, to deliver uniform, calibrated, and public domain spectral data products for the global research community (Bryan et al., 28 Aug 2025, Bock et al., 4 Nov 2025, Crill et al., 2024).

1. Spacecraft and Instrument Architecture

SPHEREx utilizes a flight-proven Ball Aerospace Configurable Platform (BCP), with a launch mass of 500–600 kg and a single-panel solar array supplying 271–1000 W, depending on mission stage (Bock et al., 4 Nov 2025, Crill et al., 2024). Passive thermal management relies on three-stage, nested V-groove radiators and photon shields, stabilizing optics to below 80 K and mid-IR arrays to as low as 45 K. The attitude control system employs dual-star trackers, inertial reference units, four reaction wheels in a tetrahedral configuration, and torque rods for momentum dump, achieving pointing stability of 0.15″ rms during 116.9 s exposures (Bock et al., 4 Nov 2025).

The science payload consists of an off-axis, three-mirror anastigmat telescope (f/3) with a 20 cm effective aperture, dichroic beam splitters, and two focal plane assemblies (3×H2RG HgCdTe arrays each). Each detector (2048×2048 pixels, 18 μm per pixel) covers a 3.5°×3.5° field, and six arrays form a 3.5°×11° instantaneous field of view. Each array uses a linear-variable filter (LVF) that imparts a spatially varying passband, yielding six spectral bands subdivided into 17 independent channels per band—totaling 102 near-infrared channels sampling 0.75–5.0 μm. Optical components are coupled and thermally isolated via titanium bipods on the aluminum bus (Bock et al., 4 Nov 2025, Bryan et al., 28 Aug 2025).

Summary Table: Key Spacecraft and Payload Characteristics

Subsystem	Specification	Reference
Orbit	Sun-synchronous LEO, 700 km, 98 min period	(Bock et al., 4 Nov 2025)
Focal Planes	6×H2RG (2×3 arrays), 2048×2048, 18 μm px	(Bock et al., 4 Nov 2025)
Field of View	3.5°×11° total, 6.2″ pixel scale	(Crill et al., 2024)
Wavelength Range	0.75–5.0 μm (6 bands × 17 channels)	(Bock et al., 4 Nov 2025)
Spectral Resolution	R≈35–41 (0.75–3.8 μm), R≈110–130 (3.8–5 μm)	(Crill et al., 30 May 2025)
Cooling	Passive 3-stage radiators, 45–80 K	(Bock et al., 4 Nov 2025)
Power/Comms	271–1000 W, S/Ka-band (600 Mbps science DL)	(Bock et al., 4 Nov 2025)

2. Survey Design and Observation Sequencing

The SPHEREx survey is structured around two modes: (1) four interleaved All-Sky Surveys over 25 months, each offset by half a channel to optimize spectral sampling; and (2) continuous deep-field mapping near the ecliptic poles, covering ~100 deg² per pole per orbit (Bryan et al., 28 Aug 2025). Each orbit is divided into a sequence of exposures: 112.5–116.9 s “step-and-stare” pointings, separated by 11.8′ spatial steps for full-wavelength coverage, and large slews of up to 70° for efficient sky tiling. The deep fields accumulate 400+ exposures per channel at the field centers by orbital repetition and randomized sub-pixel dithering (Bryan et al., 28 Aug 2025, Bock et al., 4 Nov 2025).

On-orbit survey implementation is strictly constrained by Sun, Earth, Moon, and “ram” (shuttle glow suppression) angles, as well as South Atlantic Anomaly (SAA) passages and scheduled downlinks. The science team targets ≥98% voxel completeness (6.15″×6.15″ pixel × spectral channel) in the all-sky map. Simulations and initial mission operations demonstrate that >99.5% completeness is routinely achieved, and deep field sensitivity meets or exceeds extragalactic background light requirements (Bryan et al., 28 Aug 2025).

3. Survey Planning Software and Optimization Algorithms

Survey planning and execution are managed by the Survey Planning Software (SPS), a constraint-driven, Python-based ground operations system (Bryan et al., 28 Aug 2025). The SPS architecture encompasses orbit ingestion (JPL MONTE/SPICE via SpiceyPy), attitude constraint evaluations (NumPy/SciPy), target bookkeeping (Pandas, HEALPix), online optimization and scheduling, and telemetry feedback (Astropy).

Observation scheduling is formulated as a discrete, online, short-horizon optimization. At each timestep, the algorithm defines binary decision variables $x_{g,t} \in \{0,1\}$ , where each $x_{g,t}=1$ signifies the commencement of an observation group $g$ (a contiguous sequence of up to 17 spectral steps) at slot $t$ . The objective is to maximize the weighted incremental coverage:

$\max \sum_{t \in T}\sum_{g \in G} w_{g,t} x_{g,t}$

Here $w_{g,t}$ combines two priority metrics: spectral completeness gap $(17-N_{obs,g})/17$ and angular proximity to the receding zone edge $\Delta \theta_g / \theta_{ref}$ . A Figure-of-Merit (FoM) weighting parameter $F \approx 0.8$ governs the relative priority.

Scheduling constraints incorporate dynamic avoidance zones (thermal, power, stray light, ram), SAA blackout periods, mandatory downlink slots, and continuity of small-step spectral sampling. SPS operates using an online horizon (single-step greedy with revisit window $t+T_{late}$ ) to reduce the high-dimensional nonlinear program to tractable, rapid updates. Deep field/all-sky coverage balance is managed via a probabilistic selection ( $p \approx 85\%$ ) for deep fields when feasible (Bryan et al., 28 Aug 2025).

A single half-week planning cycle (∼1,000 attitude decisions) is generated in minutes on a 16-core server. Post-execution, telemetry updates are ingested to monitor coverage and adjust subsequent plans, supporting rapid replanning under environmental or operational contingencies (Bryan et al., 28 Aug 2025).

4. Data Processing, Calibration, and Release

SPHEREx data flow follows a staged pipeline:

Level 0 (on-board): Sample-up-the-ramp (SUR) fits per pixel yield photocurrent (e⁻/s) images, cosmic ray and saturation flags, and ~120 Gbit/day compressed raw data (Crill et al., 2024, Bryan et al., 28 Aug 2025).
Level 1 (ground): Data decompression, time-tagging, conversion to electrons/sec, and basic QA (~42 GB/day).
Level 2: Calibration of gain, nonlinearity, astrometry (Gaia frame), and PSF reconstruction.
Level 3: Forced-photometry on reference catalog positions (Gaia, Pan-STARRS, WISE/NEOWISE, 2MASS, DESI Legacy), extracting all-sky spectral catalogs and constructing the High-Reliability Source Catalog.
Level 4: Science-grade products—3D galaxy redshift catalogs, deep-field mosaics, ice absorption column densities.

Absolute gain calibration is performed using regression on Zodiacal light and stellar standards, achieving 3% (overall) and 2% (channel-to-channel) precision. Flat-field generation leverages both lab integrating sphere data and iterative self-calibration in deep fields. The pipeline flags ∼1.1% of pixels per exposure for cosmic rays, with a further ≲0.1% for reset transients (Bock et al., 4 Nov 2025). All data, including quick-release spectral images (published within two months of acquisition), annual reprocessed data cubes, and legacy catalogs, are distributed via the NASA/IPAC IRSA, following an immediate public release policy (Bock et al., 4 Nov 2025, Crill et al., 2024).

5. Spectral Performance and Sensitivity

SPHEREx’s sensitivity and spectral resolution are governed by its LVF design, optical throughput, and dominant background from Zodiacal light. Spectral resolving power per band is:

Band	Wavelength Range (μm)	R = λ/Δλ
1	0.75–1.12	41
2	1.10–1.65	41
3	1.63–2.44	41
4	2.40–3.85	35
5	3.81–4.43	110–112
6	4.41–5.01	128–130

The single-exposure, 5σ point-source limits vary with wavelength and sky position, but after four all-sky passes typically reach $\mathrm{AB}=19.5$ –$19.9$ mag (0.75–3.8 μm) and $\mathrm{AB}=17.8$ –$18.8$ mag (3.8–5.0 μm), with the sensitivity limited by the zodiacal background at all wavelengths (Bock et al., 4 Nov 2025, Crill et al., 30 May 2025, Crill et al., 2024). Deep fields are ∼1.5 mag deeper. The SNR for a given source is calculated as:

$\mathrm{SNR} = \frac{F_{\mathrm{src}} A_{\mathrm{eff}} t_{\mathrm{exp}}} {\sqrt{F_{\mathrm{sky}} A_{\mathrm{eff}} t_{\mathrm{exp}} + N_{\mathrm{dark}} t_{\mathrm{exp}} + \sigma_{\mathrm{read}}^2}}$

where the symbols denote source and background fluxes, telescope area and efficiency, exposure time, dark current, and read noise (Bock et al., 4 Nov 2025, Crill et al., 30 May 2025). For example, the mission achieves $\mathrm{AB}=18.4$  mag at $5\sigma$ in a single $R\sim41$ channel at $2~\mu$ m after four surveys (Crill et al., 2024).

6. Scientific Objectives and Data Legacy

SPHEREx’s design flows directly from three core astrophysical goals:

Measuring primordial non-Gaussianity via tracing 3D large-scale structure ( $\sigma(f_{\mathrm{NL}}^{\mathrm{loc}})\sim0.9$ forecasted).
Mapping the galactic and extragalactic ice inventory (e.g., H₂O, CO₂, CO) through absorption features in the mid-IR, based on all-sky line surveys.
Intensity mapping and evolution studies of galaxies, ISM, and the extragalactic background light (EBL), particularly through power-spectrum analysis in deep ecliptic fields (Doré et al., 2014, Lisse et al., 2024, Bock et al., 4 Nov 2025).

SPHEREx delivers uniform, spectral data (∼1.4×10⁹ galaxy spectra, >10⁸ high-quality photometric redshifts, >10⁸ stellar spectra, >10⁷ quasar spectra, ∼10⁴ asteroids with 0.75–5.0 μm coverage) (Crill et al., 30 May 2025). Specialized catalogs supporting planetary defense (spectral characterizations of >10⁵ small bodies) and stellar/galaxy cluster science are released after the primary mission (Lisse et al., 2024, Bock et al., 4 Nov 2025). Legacy science enabled by SPHEREx includes time-domain studies, Galactic ISM mapping, stellar-parameter improvement, and support for a wide array of missions (JWST, Euclid, LSST, Gaia), providing comprehensive photometric and spectroscopic cross-calibration (Doré et al., 2018, Doré et al., 2016).

7. In-Flight Performance and Operational Outcomes

Post-launch, SPHEREx has displayed stable optical, thermal, and pointing performance. Passive cooling brings the instrument to operating temperature without consumable cryogens, with detector stability better than 100 μK and temperature drifts <40 nK/s. In-flight calibration matches pre-launch instrument characterizations to <1% for flat-field and dark current. Jitter over exposures is 0.15″ rms, well below the requirement, and PSF quality is confirmed across the field and bands.

SPS has delivered >1.5× margin on requirement for voxel completeness in both all-sky and deep field surveys, with the first public data release (July 2025) showing <0.1% coverage deviation from simulation forecasts (Bryan et al., 28 Aug 2025). Image-pair differencing in the deep fields demonstrates photon-noise-limited performance for background mapping. Cosmic ray loss and terrestrial emission contamination are within predicted bounds (<2% for SAA/persistent loss; ~1% pixel flag rate for cosmic rays per exposure).

In all operational phases, the combination of passive-cooled optics, LVF spectroscopy, robust survey planning and constraint management, and fully open data policy, establishes SPHEREx as an archetype for scalable, high-productivity astrophysical survey missions—filling critical legacy roles for calibration and cross-mission synergy in the 2020s (Bock et al., 4 Nov 2025, Bryan et al., 28 Aug 2025, Crill et al., 2024).