# Destello against published values

Destello finds the same source in every SPHEREx epoch, centroids it and fits a linear motion. We compare the result with values from the literature and from Gaia. Every row below is read from the latest validation run, nothing is typed by hand.

The text below is the validation record kept in the repository (README.md). The page itself reads the latest run from the server.

`data/validation.json` (generated 2026-11-15 20:04 local time, 2026-11-15T23:04:13Z, by `scripts/run_validation.py`, pipeline 0.3.0) holds 30 rows for V-01 to V-15: 26 pass, 1 fail and 3 are not measurable. Changes made on 2026-11-15: V-14 for the August 2025 campaign of 3I/ATLAS passes with a method change made after its fail was known (the centroid now fits the Gaia DR3 stars inside the fit box together with the comet; the rule was written before it was run, the criterion is unchanged, the December campaign is the control, and both results are kept). V-01 is now reported for every preset (it was computed for RW Aurigae and R Draconis but not written; RW Aurigae passes, R Draconis has no usable reference star and is not measurable). V-15 for Barnard's Star passes with its WISE companion sequence (10372.7 mas/yr against 10398.6 from Gaia DR3, 1.4 sigma per axis; the fit does not recover the parallax). V-13 for the August 2025 campaign of 3I/ATLAS passes after an estimator correction: the code required each continuum window at 3 sigma, which the registered difference test does not ask for. The CO2 band is 88.4 +/- 13.5 mJy against a continuum of 1.0 +/- 0.4 mJy that is not detected, 6.5 sigma; no band to continuum ratio is reported. The correction was written after the window values were known, and both results are kept. The data are the five presets and the two 3I/ATLAS campaigns rebuilt on SPHEREx Data Release 1 (DR1, DOI 10.26131/IRSA678), with Quick Release 3 for the newest epochs. No tolerance was changed. V-16 (NEOSSat) was not attempted. The criteria and tolerances of each row are in `scripts/run_validation.py` and are shown on the validation page of the site. History: the build of 2026-11-12 on Quick Release 2 and 3 passed 22 of 22 checks.

Failures:

- V-10 RW Aurigae: one unexplained dipole (lobes +8.0 and -10.5 sigma) on the wing of the G 11.88 star Gaia DR3 156429997483589120, 9.6 arcsec away, whose core overflows in a few exposures; rebuilding the first epoch from the Quick Release 2 cutouts gives no dipole, so the DR1 pixel values changed.
- V-14 comet August 2025, earlier method: 43 of 63 measured exposures within the bound (median offset 1.75 arcsec, maximum 20.3 arcsec), a fail. In 19 of the 20 failing exposures a Gaia star sat inside the fit box. With the stars fitted jointly (method version 3) 78 of 79 measured exposures are within the bound (median 1.43 arcsec, maximum 6.09 arcsec). December 2025 passes with both methods (103 of 103).

Not measurable:

- V-12 Epsilon Indi B: DR1 defines a new HALO flag bit, set on 1 to 4 core pixels at the target and part of the photometry mask, so the first epoch keeps 1 of its 3 points in the band window (2 are needed). The earlier 0.424 is the Quick Release 2 value.
- V-13 3I/ATLAS CO2 band, December 2025: 1 point in the band window (2 are needed), because the exposures of that pass step 0.036 um. No detection is claimed for December.

| Check | Target | Destello | Reference | Result |
|---|---|---|---|---|
| V-02 proper motion | Barnard's Star | 10351.9 +/- 598.4 mas/yr | 10393.3 mas/yr (Gaia DR3) | pass, 0.4 sigma |
| V-03 proper motion | WISE 0855-0714 (2 epochs) | 8680.8 +/- 1338.8 mas/yr | 8087.8 mas/yr (CatWISE2020) | pass, 1.1 sigma |
| V-04 proper motion | Epsilon Indi B (unresolved pair) | 4575.5 +/- 211.0 mas/yr | 4684.2 mas/yr (Gaia DR3) | pass, 0.5 sigma |
| V-11 blind motion search | Epsilon Indi B | 4643.9 +/- 148.1 mas/yr | 4684.2 mas/yr (Gaia DR3) | pass, 0.3 sigma |
| V-15 long baseline (WISE/NEOWISE and SPHEREx) | WISE 0855-0714 | 8157.8 +/- 15.7 mas/yr | 8087.8 mas/yr (CatWISE2020, not independent) | pass, 2.5 sigma (worst axis) |
| V-15 long baseline (WISE/NEOWISE and SPHEREx) | Barnard's Star | 10372.7 +/- 17.5 mas/yr | 10398.6 mas/yr (Gaia DR3 at the fit epoch) | pass, 1.4 sigma (worst axis); reduced chi-square 6.2 and the parallax is not recovered |
| V-06 variability | R Draconis | 0.302 +/- 0.020 mag between epochs | change expected (Mira variable) | pass, 15.4 sigma |
| V-07 variability | RW Aurigae | 0.378 +/- 0.011 mag between epochs 1 and 2 (largest pair) | change expected | pass by the stated-result rule: the detectors disagree in sign (D1 to D3 brighter, D5 and D6 fainter), so no consistent change is claimed |
| V-08 flux scale | Field stars against 2MASS | -7.6 +/- 2.8 percent | within 10 percent per band | pass |
| V-09 null test | Stationary field star near Barnard's Star | 123.4 +/- 147.8 mas/yr | 4.8 mas/yr (Gaia DR3) | pass, 0.8 sigma |
| V-01, V-05 astrometry | All presets (V-01 not measurable for R Draconis: no usable reference star) | recalibration residuals 0.14 to 0.28 arcsec; position offsets 0.67 to 4.06 arcsec | below 1 arcsec; within 6.2 arcsec | pass |
| V-10 difference images | Barnard's Star, Epsilon Indi B, R Draconis, WISE 0855 | 0 unexplained dipoles | 0 | pass |
| V-10 difference images | RW Aurigae | 1 unexplained dipole | 0 | fail |
| V-14 coma centroid | 3I/ATLAS, August 2025 | 78 of 79 within the bound (43 of 63 with the earlier method, without the neighbour stars) | at least 90 percent | pass |
| V-14 coma centroid | 3I/ATLAS, December 2025 | 103 of 103 within the bound | at least 90 percent | pass |
| V-12 methane index | Epsilon Indi B | not measurable | below the field stars | not measurable |
| V-13 CO2 band | 3I/ATLAS, August 2025 | band 88.4 +/- 13.5 mJy, continuum not detected, 6.5 sigma | band above the continuum by more than 3 sigma | pass |
| V-13 CO2 band | 3I/ATLAS, December 2025 | not measurable | band above the continuum | not measurable |

For V-15 on WISE 0855, the proper motion in declination agrees with the reference only because the CatWISE2020 error is large, and CatWISE2020 is built from WISE data, so the comparison is a consistency check. The WISE epochs of 2010 to 2011 are missing because IRSA TAP did not answer.

Data note (2026-11-14): IRSA replaced Quick Release 2 with DR1 on 2026-11-12 and removed `spherex_qr2` from its TAP service. The backend queries `spherex_dr1`, `spherex_qr3` and their deep variants.
