SPHEREX · ALL-SKY INFRARED SPECTRAL SURVEY

The sky is breathing.

SPHEREx maps the whole sky in 102 colors of infrared light, every six months. We made it a time machine anyone can use — and measured what comets exhale.

Open the sky Meet the comets

Scroll to the colors

3I/ATLAS · real SPHEREx frames centred on the comet, common stretch, no continuum subtracted SPHEREx D5 · 4.14 µm · 2025-08-01 · obs 2025W31_2A_0219_1 SPHEREx: NASA/JPL-Caltech/IRSA (DR1 + QR3, L2 l2b-v27) · Ephemeris: NASA/JPL Horizons (SPHEREx observer)

visit 1 / 12

102 colors of infrared light

Every SPHEREx exposure sees the sky through a strip of filters: 0.75 to 4.99 µm, split into channels. Taller bars resolve finer color.

H₂OCH₄CO₂ ν₃CO (1–0)

Hover a bar to read its channel.

Bar height: resolving power R. Dashed bars: frames the DR1 all-sky cube leaves blank (dichroic).

§1

Pluto was found by blinking two plates.

In 1930, Clyde Tombaugh flipped between two photographic plates of the same patch of sky, taken days apart. Everything that stayed put was a star; the dot that jumped was Pluto.

SPHEREx returns to every part of the sky every six months, so the same trick works across the whole infrared sky.

Barnard's Star moves faster across our sky than any other star. Blink the first and last visits: the crosses are where Gaia predicts it on each date.

Barnard's Star · SPHEREx D4 · 24 frames from 7 visits · cropped around the star, pixels not resampled

A SPHEREx D4 · 2.71 µm · 2025-08-28 · obs 2025W35_1A_0658_4 Crosses: Gaia DR3 4472832130942575872 propagated to each date (dim = A, bright = B). Arrow: A → B. SPHEREx: NASA/JPL-Caltech/IRSA (DR1 + QR3, L2 l2b-v27) · Gaia DR3: ESA/Gaia/DPAC

Gaia-predicted shift A → B: 1.7 px = 10.5″

§2

A rainbow on a chip.

SPHEREx has no filter wheel. A linear variable filter sits on each detector, so every row of pixels sees a slightly different wavelength. A star is measured in many colors as the telescope sweeps past it.

The catch: on two visits the same star lands on different rows, so it is seen in a different color (3.74 µm versus 2.46 µm here). Subtract the two frames and almost every star leaves a residue.

Diff λ compares each frame with the static sky at the same wavelength, built from other visits. What survives is what really changed — here, the star that moved.

Diff λ reference: static sky at the frame's wavelength from 5 other frames · both panels share one ±5σ color scale

Raw subtraction of two SPHEREx frames at different wavelengths: residues at almost every star — **Raw B − A** D4 2.46 µm (2026-09-02) − 3.74 µm (2026-03-11) **Diff λ** D4 2.46 µm · 2026-09-02 · obs 2026W36_1A_0510_1 SPHEREx: NASA/JPL-Caltech/IRSA (DR1 + QR3, L2 l2b-v27)

Diff λ of the same frame: only the moving Barnard's Star leaves a positive–negative pair — **Raw B − A** D4 2.46 µm (2026-09-02) − 3.74 µm (2026-03-11) **Diff λ** D4 2.46 µm · 2026-09-02 · obs 2026W36_1A_0510_1 SPHEREx: NASA/JPL-Caltech/IRSA (DR1 + QR3, L2 l2b-v27)

Drag to compare raw subtraction (left) with Diff λ (right)

−5σ 0 +5σ saturated σ = 0.020 MJy/sr

§3

What the sky is made of.

Each SPHEREx color is a slice of a spectrum. Molecules absorb or glow at their own wavelengths: water, methane, carbon dioxide and carbon monoxide all leave fingerprints between 1 and 5 µm, and hydrogen marks itself in sharp lines.

This is a real SPHEREx spectrum of a pair of brown dwarfs, separated from their neighbors with Gaia. Pick a band to see where it shows up.

Band definitions: §12.3 of the build guide · shaded widths are the regions annotated, not fits

Loading… SPHEREx: NASA/JPL-Caltech/IRSA

Select a band to learn which object in this set shows it.

§4

Comets breathe.

Near the Sun, a comet's ices turn to gas. In SPHEREx frames the dust glows as a smooth continuum; subtract it and the gas remains, brightest at the carbon dioxide band.

From that glow we estimate how many molecules leave the nucleus each second, always with an uncertainty range and a stated caveat.

Interstellar comet 3I/ATLAS is our validation case: its numbers are compared with published measurements in §5.

SPHEREx plate of 3I/ATLAS — 3I/ATLAS · SPHEREx · 4.14 µm · MJD 60888.67 · obs 2025W31_2A_0219_1 SPHEREx: NASA/JPL-Caltech/IRSA (DR1 + QR3, L2 l2b-v27) · Ephemeris: NASA/JPL Horizons (SPHEREx observer)

Q(CO₂), August 2025

1.17 × 10²⁷ s⁻¹

range 8.60 × 10²⁶ – 1.38 × 10²⁷

Statistical error ±3.0e+25; χ²ᵣ 27.5. The 16–84% range comes from the research notes and Q scales with the assumed outflow speed (about −40% for 0.50 instead of 0.85 r_h^−0.5 km/s).

Research fit of the CO₂ ν₃ band (research/11_band_physical.py), reproduced 2026-11-14

Q(CO₂), December 2025: 3.42 × 10²⁷ s⁻¹ (range 2.78 × 10²⁷ – 4.18 × 10²⁷); Band present (S/N 41.1) but the model fits poorly (χ²ᵣ 185): Q is not trusted at its quoted precision.; Comet census (co2_census_physical.csv)

§5

Measured, then checked.

A number is only useful if you know how far to trust it. Every SPHEREx measurement here is compared with an independent reference before it is shown.

Comet gas production
Species	SPHEREx (this work)	Published range	Check
3I/ATLAS · CO₂ · 2025-08	1.17 × 10²⁷ s⁻¹	9.40 × 10²⁶–1.60 × 10²⁷	within range
χ²ᵣ 27.5. Consistent with the published range; not an independent confirmation. Lisse et al. 2025, RNAAS, arXiv:2508.15469 (0.94e27, preliminary); Lisse et al. 2026, ApJL 1000, L52, arXiv:2512.07318 (1.6e27 +/- 0.16e27)
29P/Schwassmann-Wachmann · CO · 2025-12	4.50 × 10²⁸ s⁻¹	2.90 × 10²⁸–5.60 × 10²⁸	within range
χ²ᵣ 127.3. Consistent with the published range; not an independent confirmation. Bockelee-Morvan et al. 2022 (IRAM, 2.9-5.6e28); quiescent 1.8e28 (Festou et al. 2001)

Photometry against 2MASS and WISE · 41 stars · fields CDFS, COSMOS, NEP
Band	Catalogue	Stars	SPHEREx / catalogue
J	2MASS	38	0.982 ± 0.020
H	2MASS	39	1.008 ± 0.038
Ks	2MASS	29	1.011 ± 0.033
W1	WISE	41	0.972 ± 0.014
W2	WISE	41	0.996 ± 0.014

SPHEREx flux divided by catalogue flux, per band, ± σ. Mean of the bands 0.994; band ratios span 0.972–1.011. Method: flux = sampled-ePSF forced fit (R=4 px, free background), NO aperture correction; err^2 = sigma^2 + (0.03 F)^2; official photometry mask

§6

Five places to start.

The five objects of the challenge, each with a real SPHEREx plate and field notes. Every number sits next to its reference and its caveat.

D4 · 2.71 µm · 2025-08-28 obs 2025W35_1A_0658_4

01

Barnard's Star

The fastest star across our sky.

Proper motion (SPHEREx)

10.6 ± 0.4″/yr

Gaia 10.39″/yr
Parallax

not measurable (saturated core)

547 mas

Caveat:formal error ±0.10″/yr; ±0.4 reflects frame systematics (saturated core, 3 useful visits)
Explore Barnard's Star
D6 · 4.85 µm · 2025-10-06 obs 2025W41_1A_0052_1

02

Epsilon Indi B (Ba+Bb)

A brown dwarf pair, with methane in its spectrum.

Proper motion (SPHEREx)

4.75 ± 0.04″/yr

Gaia DR3 4.68″/yr
Parallax (differential vs 4 Gaia field stars)

271 ± 20 mas

Gaia DR3 270.7 mas (ε Ind A: 274.8 mas)
Methane at 3.3 µm

5.4× fainter than the 4.0 µm peak

T1+T6 dwarfs

Caveat:error from per-visit bootstrap (9 visits); dropping one control star gives 277–294 mas
Explore Epsilon Indi B (Ba+Bb)
D6 · 4.82 µm · 2026-04-26 obs 2026W17_2A_0591_1

03

WISE 0855−0714

The coldest known object outside the Solar System.

Below 3.8 µm

≈ 0 (deblended from Gaia star at 6.8″)

J ≈ 25
Synthetic W2 (real W2 response curve)

0.45 ± 0.02 mJy

published W2 ≈ 13.9 mag (0.48 mJy); CatWISE2020 13.82 (0.51 mJy)

Caveat:14% of the W2 band lies beyond SPHEREx's 5.0 µm limit; the AllWISE catalogue value (13.70 mag, 0.57 mJy) is blended at the 2010 position
Explore WISE 0855−0714
D6 · 4.80 µm · 2026-02-21 obs 2026W08_2A_0361_1

04

RW Aurigae

A young star dimmed and brightened by its disk.

2025-09 → 2026-09

optical ×1.8 · 4–5 µm ×0.5 (anti-correlated)

inner-disk hot dust
SPLICES

ice target (Class II YSO)

IRSA

Caveat:three epochs only
Explore RW Aurigae
D6 · 4.68 µm · 2026-04-11 obs 2026W15_2A_0483_3

05

R Draconis

A pulsating red giant.

Variability timescale (3.8–5.0 µm, wing photometry)

≈ 229 d (225–241)

catalogued period 245.6 d
Peak-to-peak range (5–95%)

2.0 mag (0.9 µm) → 0.7 mag (4.4 µm)

Mira

Caveat:only 1.9 cycles observed: consistent with the catalogued period, not an independent measurement (1.1–2.4 µm gives 236 d, 230–519)
Explore R Draconis