betaFeedback

WorksPhysics and Astronomy

Comprehensive Visible and Near-infrared Keck and Gemini Monitoring and Characterization of Interstellar Comet 3I/ATLAS during its Encounter with the Solar System

Abstract

We present 14 optical imaging and eight optical spectroscopic epochs of interstellar comet 3I/ATLAS from Keck/LRIS and Gemini/GMOS, two near-infrared (NIR) imaging epochs from Keck/MOSFIRE, and four NIR spectroscopic epochs from Gemini/Flamingos-2 and GNIRS, obtained between 2025 July and 2026 May. Mean GMOS colors of g−r=0.732±0.029g-r=0.732\pm0.029, r−i=0.237±0.025r-i=0.237\pm0.025, and i−z=0.109±0.022i-z=0.109\pm0.022 mag show no significant change across perihelion. Phase-corrected red-band activity follows Afρ∝rh−2.10±0.14Afρ\propto r_h^{-2.10\pm0.14} inbound and rh−3.34±0.43r_h^{-3.34\pm0.43} outbound. Transverse-width and sunward-plume speeds imply dust losses of 2.7×1092.7\times10^9 and 1.0×10101.0\times10^{10} kg over 2025 July 3–2026 May 15 for 1-μμm–1-mm grains, equivalent to global layers of 0.26 and 0.99 m under the adopted grain and nucleus assumptions. The concentrated coma, compact sunward feature, and antisolar tail seen on 2025 August 27 broadened by 2025 November 26; extended dust persisted through 2026 February 17. C2_2/CN nearly doubled from 0.257±0.0140.257\pm0.014 on 2025 September 14–15 to 0.481±0.0800.481\pm0.080 on 2025 November 26, while C3_3/CN fell from 0.327±0.0240.327\pm0.024 on 2025 August 28 to 0.214±0.0090.214\pm0.009 on 2025 September 14–15. MOSFIRE NIR imaging on 2025 July 11 gives Y−J=0.466±0.143Y-J=0.466\pm0.143, J−H=0.239±0.052J-H=0.239\pm0.052, H−Ks=0.454±0.150H-K_s=0.454\pm0.150, and J−Ks=0.693±0.202J-K_s=0.693\pm0.202 mag. Spectra give J−H=0.24J-H=0.24 mag on 2025 September 5–6 and J−H=0.27J-H=0.27–0.28 and J−Ks=0.21J-K_s=0.21–0.22 mag on 2025 December 1–2 over the retained passband intervals. Model-calibrated imaging on 2026 February 6 gives near-solar J−Ks=0.303±0.051J-K_s=0.303\pm0.051 mag. No water-ice absorption is robust to the tested continuum and calibration choices in the NIR spectra. These observations provide a lasting record of the comet's evolution through the Solar System and a benchmark for studying and monitoring future interstellar objects.

Typeset using LATEX default style in AASTeX7.0.2

Introduction

Interstellar objects (ISOs) carry material formed in other planetary systems into the Solar System ([40]. Their ejection links this population to planetary scattering and the solid inventory of their parent systems ([99, 86, 101, 73]. Radial drift, migrating ice lines, and volatile redistribution affect the material available for ejection ([124]); white-dwarf pollution provides independent evidence for volatile-rich extrasolar planetesimals ([102]. Irradiation modifies ISO surfaces, while solar heating releases gas and dust that can obscure the nucleus ([41, 47]. Observations across an encounter help separate this evolution from compositional differences, although size, activity, and survey sensitivity bias the small detected sample.

1I/‘Oumuamua showed a red continuum and a large-amplitude, irregular lightcurve consistent with a strongly nonspherical body in non-principal-axis rotation ([80, 20, 38]. Its featureless optical/NIR spectrum resembled primitive Solar System surfaces ([41], and its nongravitational acceleration occurred without a resolved dust coma ([81]. 2I/Borisov had a coma and tail with comet-like optical colors ([65, 47, 19]. Although the coma complicated nucleus and rotation measurements ([18], HSTHST profiles constrained its radius to <0.5<0.5 km for geometric albedo 0.04 ([62]. Ultraviolet and millimeter spectra revealed high CO/H2_2O abundance ([10, 34]. Optical–millimeter observations favored compact grains exceeding approximately 1 mm and found evolution in CO/H2_2O across perihelion ([121]. Ni I was also detected in its cold coma ((P. Guzik & M. Drahus 2021). Comparing the surface-dominated spectrum of 1I with 2I’s coma requires accounting for grain size, phase angle, and activity.

L. Denneau et al. [37] announced the ATLAS discovery of 3I/ATLAS, followed by early characterization by B. T. Bolin et al. [12] and D. Z. Seligman et al. [110]. The comet showed a red continuum and resolved coma, with weak or undetected canonical gas bands near discovery ([96, 4] (C. Optiom et al. 2025; R. de la Fuente Marcos et al. 2025). Rubin observations cover 2025 June 21–July 20 ([32]); ATLAS resolves changes in the inbound growth law ([115], and TESS, HATPI, and smaller telescopes extend the inbound baseline ([76, 53, 117]. Optical monitoring tracks the changing continuum, morphology, and gas emission ([5, 97, 8, 106, 3], including the CN and Ni coma observed with UH88/SNIFS and Keck/KCWI ((W. B. Hooegendam et al. 2026a, 2025, 2026b). JCMT constrains parent molecules ([54], while infrared observations probe ice, volatiles, and dust mineralogy ([120, 6 b]. The red visible continuum flattens in the NIR ([79, 120], becoming neutral or slightly blue in IRTF spectra ([70]. Polarimetry reveals an unusually deep, narrow negative branch ([44].

We use Keck and Gemini imaging and spectroscopy before and after perihelion to measure 3I/ATLAS’s morphology, colors, dust activity, and gas production and constrain water ice. Comparisons use a common physical aperture, phase correction, and orbital-plane geometry, following the 2I/Borisov and P/2019 LD2_2 studies ([19, 14]. Prediscovery observations of 2I demonstrate the value of extending the baseline beyond efficient water @references 天天爱彩票 sublimation (Q. Ye et al. 2020). We compare our measurements with preliminary reports ([22, 25 c,a]); the December 22 GMOS spectra also appear in Appendix A of [7].

Observations

The campaign comprises 14 optical imaging epochs (13 GMOS, one LRIS), eight optical spectroscopic epochs (seven GMOS, one LRIS), two MOSFIRE imaging epochs, and four NIR spectroscopic epochs (two Flamingos-2, two GNIRS; Table 1). These counts include excluded observations and treat each session, including the August 27–28 imaging and other sequences crossing UTC midnight, as one epoch. Alternative extractions, orders, and combinations share exposures.

Optical observations

GMOS-S observations were obtained under GS-2025B-DD-103 (PI: Elena Sabbi) on 2025 August 27–28 and GS-2025B-DD-102 (PI: Bryce T. Bolin) on September 5, 6, 8, and 14–15. GMOS-N observations used GN-2025B-DD-102 (PI: Bryce T. Bolin) on 2025 November 26, December 19 and 22, and 2026 January 20–21 and February 5 and 7; February 17 and March 27 used GN-2025B-FT-113 (PI: Bryce T. Bolin).1 LRIS spectroscopy used Keck program 2025B W353 (PI: Jeffrey Cooke); imaging on 2026 May 15 UTC used the target-of-opportunity program 2026A N041 (PI: Ian Wong). Liverpool IO:O data ([45]) were obtained under PATAG AZ22A01 (PI: F. Romanov).

We obtained uu, gg, rr, ii, ZZ images with GMOS-S, gg, rr, ii, ZZ with GMOS-N, and GG, RR, II with LRIS. The GMOS setup resembles those used for 2024 YR4 and 2024 PT5 ([17 b]). GMOS long-slit spectra used 540- and 550-nm central settings. We evaluate the 2025 September 1 LRIS spectroscopic geometry at 00:00 UTC because reduced headers lack exposure timestamps; LRIS imaging uses mean exposure midpoints.

The electronic GMOS log contains timestamps and airmasses for 148 exposures contributing to 49 date/filter stacks. We use exposure-weighted midpoints, assigning parent-stack midpoints where subset membership is unknown. The 14 spectral settings retain midpoints, integration times, and slit position angles. GMOS-S used B480+ G5327 and GMOS-N B480+ G5309, both in first order with open filters and a 1-arcsec slit. Their spatial scales are 0.1600 and 0.1614 arcsec pixel−1^{-1}, respectively. The 34 spectral exposures total 10,180 s.

The 2025 August 27 Gemini South run included the public Shadow the Scientists (StS) session “A Close Encounter with an Interstellar Visitor from Gemini South.”2 The 2025 November 26 GMOS-N observations accompanied “Exploring Comet 3I/ATLAS: A Rare Encounter with an Interstellar Visitor,” led by B.T.B., Laura-May Abron, and Matthew Belyakov.3

*Imaging reduction and calibration—*The GMOS count-rate stacks are comet-aligned on a north-up, east-left grid with 0.16-arcsec pixels (Figure 1 and 2). Science, variance, quality, exposure-count, and exposure-map extensions are retained. Each calibrated derivative uses its own exposure time. Diagonal variances omit interpolation, background, and calibration covariance. Field-source minor-axis FWHM spans 0.63–1.68 arcsec for GMOS-S and 0.68–0.97 arcsec for GMOS-N in the measured subset.

Figure 1. Gemini South GMOS imaging in uu, gg, rr, ii, ZZ columns. August 28 uu shares the August 27 grigri row and carries its own date. Panels give date, summed integration, compass, scale, Sun direction, and positive heliocentric-velocity direction. North is up and east left; absent filters are blank. September 5, 8, and 14 rr use 0.48-arcsec valid-pixel bins with bilinear rendering. September 6 rr uses a mask-normalized 0.94-arcsec-FWHM Gaussian average to suppress mask holes, reducing display resolution. Photometry and profiles use unsmoothed data. Exposure coverage is spatially nonuniform. Figure 1. Continued.

GMOS grigri calibration uses PS1 stars evaluated at g−r=0.6g-r=0.6. Comet-specific passband corrections are unmeasured; uu and ZZ also have sparse-standard or catalog-transformation limitations. Native calibrations are retained electronically; accepted measurements are transformed to approximate SDSS AB equivalents below. Calibration or aperture-quality failures exclude March and affected February images.

Saturated-flat flags and repeated resampling produced patches in the September 5, 6, 8, and 14 GMOS-S rr displays. Figure 1 and 8 use aligned instrumental stacks, with valid-pixel averages in 3×33 \times3 bins (0.48 arcsec) on September 5, 8, and 14. Bins with fewer than three valid pixels remain black. September 6 uses a mask-normalized Gaussian average with σ=0.40\sigma=0.40 arcsec (FWHM 0.94 arcsec) and at least 5% valid kernel weight. These bilinearly rendered displays have nonuniform coverage; profile measurements and Figure 7 use unsmoothed data.

GMOS intensity images, dust-model backgrounds, and February MOSFIRE images use a median background from 20–29 arcsec and an inverse-hyperbolic-sine stretch. The upper level is Iupper=max⁡(1.05Ipeak,8σsky)I_{\mathrm{upper}}=\max(1.05I_{\mathrm{peak}},8\sigma_{\mathrm{sky}}), with IpeakI_{\mathrm{peak}} measured within 3 arcsec and σsky\sigma_{\mathrm{sky}} the background scatter; softening is max⁡(2σsky,0.005Iupper)\max(2\sigma_{\mathrm{sky}},0.005I_{\mathrm{upper}}). Figure 7 uses a fixed ratio scale with unity in white.

The LRIS imaging setup resembles those used for 1998 KY26 and Neptunian Trojans [15, 16]. LPipe [91] reductions were registered and summed. May data use approximate PS1 gg, ii AB and Cousins RR Vega calibrations. Photometry uses a 1.7765-arcsec radius (10,000 km) and 4–6-arcsec sky annulus. Blank apertures receive identical sky subtraction and exclude masked or uncovered pixels. Sky-variance, saturation-mask, and exposure maps support quality checks, although covariance is incomplete. Adopting the larger of the blank-aperture scatter and saved magnitude error gives native gg, RR, ii uncertainties of 0.121, 0.133, and 0.197 mag. Shared color-term propagation gives SDSS gg, rr, ii errors of 0.140, 0.116, and 0.197 mag, excluding the unmeasured comet-specific correction.

Spectroscopic reduction and calibration—GMOS solar-analog ratios use air wavelengths and normalization over 545–555 nm. We adopt the 40-pixel full-width extraction, testing alternative apertures and sky regions. Gas calculations use the full rectangular extraction dimensions.

LRIS blue/red pairs provide all-exposure, coma, and coma–tail spectra. We convert native FνF_\nu to FλF_\lambda before division by the CALSPEC synthetic solar spectrum [11]. The 320–560- and 560–1000-nm arms meet at 560 nm. Robust linear continua fitted to 1-nm median bins over 545–559 and 561–575 nm, excluding 556–558 nm and immediate edges, give an all-extraction red-arm multiplier kR=1.1915±0.0287k_R = 1.1915 \pm0.0287. Its uncertainty comes from residual bootstraps in contiguous 2-nm blocks; 10-, 15-, and 20-nm fitting neighborhoods give 1.1771–1.2160. We multiply the red arm by kRk_R and normalize both solar-divided arms over 545–555 nm. Original and adjusted fluxes retain the shared scale uncertainty. Air and vacuum solar templates are tested separately because the native wavelength convention is undocumented. Absolute calibration and wavelength-dependent slit losses remain sources of uncertainty in the arm match.

LRIS used 400/3400 and 400/8500 dispersers. We adopt a 0.7-arcsec slit for production rates, although reduced headers do not verify its width. J.C.’s reduction notes identify three tail-aligned exposures in coma–tail and two parallactic-angle exposures in coma, combined into the all-exposure spectrum. A fourth tail-aligned exposure was rejected for possible stellar contamination. Numerical slit position angles are unavailable. G191B2B provides flux calibration; the observed solar twin HD150469 was not divided from the delivered spectra. The 70-pixel extraction-boundary separation corresponds to 9.45 arcsec at the assumed binned blue scale of 0.135 arcsec pixel−1^{-1}. Unknown binning motivates a doubled-height sensitivity test.

Infrared observations

MOSFIRE observations on 2025 July 11 UTC used Keck engineering program 2025A E498 (PI: Josh Walawender); 2026 February 6 UTC used 2026A N045 (PI: Ian Wong).4 Flamingos-2 observations on 2025 September 3 and September 5–6 used GS-2025B-FT-206, and GNIRS on December 1–2 used GN-2025B-FT-110 (both PI: Bryce T. Bolin).

MOSFIRE broadband setups resemble the YY, JJ observations of 2024 YR4_4 [17] and YY, JJ, HH, KK observations of 2024 PT5_5 [9]. July data cover YY, JJ, HH, KsK_s, and February JJ, KsK_s. Complete-aperture individual exposures provide July Y−JY-J through sidereal calibration transfer. Flamingos-2 obtained JH and HK spectra; GNIRS obtained two cross-dispersed blocks on each night. July MOSFIRE geometry uses an exposure-weighted midpoint; February uses the first retained MJD-OBS because no stack midpoint is available.

NIR exposure start times and airmasses are identified in Table 1. GNIRS used the 32/mmSB-G5533 grating, XD filter, 0.68-arcsec slit, and 0.1517-arcsec pixels. Flamingos-2 used JH-G5801 and HK-G5802 prisms with an 8-pixel (1.43-arcsec) slit. LRIS, Flamingos-2, and GNIRS lack independent seeing measurements; image-quality percentile categories are retained electronically.

MOSFIRE imaging reduction and calibration—MOSFIRE stacks are shown in ADU s−1^{-1} (Figure 18). July JJ, HH, KsK_s each contain 27 exposures totaling 707.0, 785.6, and 785.6 s; the YY display combines 21 of 27 exposures (611.0 s). Masks exclude nonfinite science or variance, nonzero quality flags, and coverage below three exposures. The 12-pixel YY target aperture has 78.4% valid coverage; missing regions are shown in dark gray and excluded from photometry.

Figure 18. MOSFIRE YY, JJ, HH, KsK_s imaging, north up and east left. Labels give date, summed integration, directions, and scale; Sun/heliocentric-velocity vectors are orange/blue. Each instrumental stack has an independent grayscale stretch. July YY combines 21 exposures totaling 611 s. Dark gray marks nonfinite science/variance values, quality flags, or coverage below three exposures; resampling requires valid contributing pixels. This uncalibrated stack has incomplete inner-coma coverage and nonuniform exposure time. July Y−JY-J uses separate, complete-aperture exposures and sidereal calibration transfer. February Y/HY/H cells are blank; green circles locate the July comet.

Seven common unsaturated stars initially gave inconsistent July Y−JY-J colors (−0.17 to +0.62 mag). We transfer calibration from the JJ-band A0-like reference star, unsaturated in 8.72874-s sidereal YY exposures but saturated in 29.0958-s comet exposures. Three bridge stars provide five accepted transfer pairs across two comet exposures. Same-frame bridge/anchor ratios cancel transparency without predicted bridge-star colors.

Sidereal images are offset-subtracted and flat-fielded using the existing YY flat, bad-pixel mask, astrometry, and 26,000-DN threshold. Local background planes use 25–35-pixel sky annuli; stellar aperture radii are 8 pixels, with 12-pixel growth checks. Complete comet apertures have a 12-pixel radius (2.1576 arcsec) and 60–80-pixel sky annulus. We multiply the science-frame comet/bridge ratio by the sidereal bridge/anchor ratio, divide by the comet/anchor JJ ratio, and apply (Y−J)a=0.2663(Y-J)_{\mathrm{a}} = 0.2663 mag. We combine the three per-bridge medians without missing-pixel or nonlinear corrections. The incomplete YY mosaic is used only for display.

For JJ, HH, KsK_{\mathrm{s}}, ephemeris-centered apertures have 12-pixel radii (2.1576 arcsec at 0.1798 arcsec pixel−1^{-1}); 8, 16, and 20 pixels (1.4384, 2.8768, and 3.5960 arcsec) test aperture sensitivity. Fractional weights use 5×55 \times5 subpixels. A robust sky plane uses a source-masked 60–80-pixel annulus, with alternative annuli testing background sensitivity. Pixels require zero quality flags and at least three exposures; nested apertures are correlated.

February zero points average predictions for two stars with solar-like Pan-STARRS colors, using the CALSPEC solar spectrum R. C. Bohlin et al. [11] scaled to each star’s rr AB magnitude. Their unknown infrared spectral mismatch is excluded from quoted uncertainties. July calibration fits a reddened Castelli–Kurucz atmosphere F. Castelli & R. L. Kurucz [31] to one A0-like star’s Pan-STARRS grizygrizy photometry with a 0.03-mag error floor. The adopted model has Teff=9000T_{\mathrm{eff}} = 9000 K, log⁡g=4.5\log g = 4.5, solar metallicity, and AV=2.60A_V = 2.60 mag under the J. A. Cardelli et al. [30] law (RV=3.1R_V = 3.1); model variations test sensitivity. Stellar apertures use 8-pixel radii without total-flux correction. We reject a blended G-like candidate. The adopted star’s raw JJ peak, 25,326 DN, approaches the 26,000-DN rejection threshold.

Synthetic photometry uses manufacturer MOSFIRE filter curves and CALSPEC solar and Vega spectra. Missing atmospheric and full-system responses limit calibration to approximate native MOSFIRE colors. Solar values are (Y−J,J−H,H−Ks,J−Ks)⊙=(0.2382,0.3002,0.0561,0.3563)(Y-J, J-H, H-K_{\mathrm{s}}, J-K_{\mathrm{s}})_{\odot} = (0.2382, 0.3002, 0.0561, 0.3563) mag. Following D. Morate et al. [83], Rb/RJ=100.4[(J−b)comet−(J−b)⊙]R_b/R_J = 10^{0.4[(J-b)_{\mathrm{comet}}-(J-b)_{\odot}]}; reported colors include the solar term, and JJ-normalized reflectances have correlated errors.

Direct 2MASS calibration failed quality checks: February zero points scatter by 0.57 mag in JJ and 0.73 mag in KsK_{\mathrm{s}}, with unstable aperture growth, incomplete core coverage, and structured residuals. July candidates lack a consistent astrometric translation. We adopt the model-dependent calibrations above. Field-source minor-axis FWHM spans 0.50–1.44 arcsec in July and 1.15–1.42 arcsec in February JJ; no satisfactory similar-brightness February KsK_{\mathrm{s}} star survives. Band-by-band profiles are given in § 3.2.1.

Flamingos-2 and GNIRS spectral reduction—Figure 19 shows Flamingos-2 JH alongside the retained H spectrum from HK and groups GNIRS blocks by night. F2 integrations are 1020 s in HK on September 3 and 340 s in JH on September 5/6. GNIRS blocks contain 1708 and 1220 s on December 1 and 1464 s each on December 2 (2928 s per night). F2 cross-setting colors are excluded because settings were observed on different nights. GNIRS orders are normalized to order 4, using each nonoverlapping segment once.

F2 and nightly combined GNIRS spectra, with relative reflectance plotted against wavelength for Flamingos-2 and GNIRS observations on December 1 and 2, and illustrative ice models

Figure 19. F2. F2 and nightly combined GNIRS spectra. The upper panel shows F2 JH alongside the retained H segment of HK from a different night, excluding JH below 1.0 μm\mu\mathrm{m} and HK outside 1.45–1.80 μm\mu\mathrm{m}. GNIRS blocks are combined separately for December 1 and 2. Errors use local scatter and a block-repeatability floor; gray bands mark excluded atmospheric intervals. Illustrative ice models assign 10% of the scattered light at 1.20 μm\mu\mathrm{m} to grains of radius 1, 10, or 100 μm\mu\mathrm{m}. Each model’s continuum level and slope are fitted to the spectrum; black crosses mark the 1.20-μm\mu\mathrm{m} anchors.

We use a 40-pixel F2 extraction radius (7.2 arcsec at the nominal pixel scale of 0.18 arcsec pixel−1^{-1}); radii of 20, 60, and 80 pixels provide correlated extraction checks. The primary extraction spans 14.4 arcsec along the slit. Reduction notes identify the second columns as solar-divided spectra, normalized here over 1.57–1.63 μ\mum. Signed third columns are excluded from uncertainty estimates. We discard JH samples below 1.0 μ\mum and retain only HK’s 1.45–1.80-μ\mum H segment, removing its spurious short-wavelength extension and red upturn. The cuts eliminate F2 KsK_{\mathrm{s}} coverage and its 2-μ\mum ice diagnostic. Deep atmospheric intervals remain masked.

Empirical errors use robust scatter about local linear trends in 0.01-μ\mum bins, divided by the square root of an effective sample count accounting for positive autocorrelation and at least three native pixels per independent element. Smooth calibration errors remain unquantified. GNIRS nightly bins are inverse-variance averages with a half-block-separation error floor; individual blocks test repeatability and set a separate color-error floor.

UTC date/timeFacility/instrumentFilter/gratingtexpt_{\mathrm{exp}}XXrhr_hΔ\Deltaα\alphaδE\delta_Et−Tpt-T_p
2025-07-11 07:10K1/MOSFIREY∗Y^*6111.624.1723.2175.4-0.85-110.2
2025-07-11 07:09K1/MOSFIREJJ7071.594.1723.2175.4-0.85-110.2
2025-07-11 07:18K1/MOSFIREHH7861.554.1723.2175.4-0.85-110.2
2025-07-11 07:27K1/MOSFIREKsK_{\mathrm{s}}7861.504.1723.2175.4-0.85-110.2
2025-08-27 23:58GS/GMOS-Sgg601.172.6372.58922.30.40-62.5
2025-08-27 23:56GS/GMOS-Srr601.172.6372.58922.30.40-62.5

Table 1. Individual imaging filters and spectroscopic settings.

UTC date/timeFacility/instrumentFilter/gratingtexpt_{\mathrm{exp}}XXrhr_hΔ\Deltaα\alphaδE\delta_Et−Tpt-T_p
2025-08-27 23:54GS/GMOS-Si601.162.6372.58922.30.40-62.5
2025-08-28 00:02GS/GMOS-Su2401.182.6372.58922.30.40-62.5
2025-08-28 00:23GS/GMOS-SB480+/5409001.242.6362.58922.30.40-62.5
2025-08-28 00:42GS/GMOS-SB480+/5509001.322.6362.58922.30.40-62.5
2025-09-03 23:30{sup}†`GS/F2HK grism10201.262.4262.56723.10.62-55.5
2025-09-05 23:59{sup}†`GS/F2JH grism3401.412.3672.56223.20.68-53.5
2025-09-05 23:39GS/GMOS-Su1001.302.3672.56223.20.68-53.5
2025-09-05 23:37GS/GMOS-Sg1001.302.3672.56223.20.68-53.5
2025-09-05 23:36GS/GMOS-Sr1001.292.3672.56223.20.68-53.5
2025-09-05 23:35GS/GMOS-Si1001.292.3672.56223.20.68-53.5
2025-09-05 23:34GS/GMOS-SZ1001.282.3672.56223.20.68-53.5
2025-09-06 23:32GS/GMOS-Sg251.302.3382.55923.20.71-52.5
2025-09-06 23:30GS/GMOS-Sr251.292.3382.55923.20.71-52.5
2025-09-06 23:27GS/GMOS-Si251.282.3382.55923.20.71-52.5
2025-09-06 23:24GS/GMOS-SZ251.272.3382.55923.20.71-52.5
2025-09-08 23:35GS/GMOS-Su1001.372.2802.55523.20.77-50.5
2025-09-08 23:34GS/GMOS-Sg1001.362.2802.55523.20.77-50.5
2025-09-08 23:33GS/GMOS-Sr1001.362.2802.55523.20.77-50.5
2025-09-08 23:32GS/GMOS-Si1001.352.2802.55523.20.77-50.5
2025-09-08 23:31GS/GMOS-SZ1001.352.2802.55523.20.77-50.5
2025-09-14 23:38GS/GMOS-Sg501.622.1112.54222.60.95-44.5
2025-09-14 23:36GS/GMOS-Sr601.602.1112.54222.60.95-44.5
2025-09-15 00:04GS/GMOS-SB480+/54012001.862.1102.54222.60.95-44.5
2025-09-15 00:28GS/GMOS-SB480+/55012002.182.1102.54222.60.95-44.5
2025-11-26 15:46GN/GMOS-Ng361.351.7021.96230.22.3828.2
2025-11-26 15:47GN/GMOS-Nr241.351.7021.96230.22.3828.2
2025-11-26 15:48GN/GMOS-Ni141.341.7021.96230.22.3828.2
2025-11-26 15:45GN/GMOS-NZ161.351.7021.96230.22.3828.2
2025-11-26 14:33GN/GMOS-NB480+/540601.911.7001.96330.22.3828.1
2025-11-26 14:39GN/GMOS-NB480+/550601.831.7011.96230.22.3828.1
2025-12-01 14:28{sup}†`GN/GNIRS32/mm, B117081.461.8151.90530.62.4033.1
2025-12-01 15:12{sup}†`GN/GNIRS32/mm, B212201.281.8161.90530.62.4033.1
2025-12-02 14:28{sup}†`GN/GNIRS32/mm, B114641.431.8401.89530.62.4034.1
2025-12-02 15:03{sup}†`GN/GNIRS32/mm, B214641.271.8401.89530.62.4034.1
2025-12-19 12:08GN/GMOS-Ng101.462.2951.79824.12.1651.0
2025-12-22 12:02GN/GMOS-Ng401.342.3821.80222.12.0754.0
2025-12-22 12:03GN/GMOS-Nr401.342.3821.80222.12.0754.0
2025-12-22 12:04GN/GMOS-Ni401.332.3821.80222.12.0754.0
2025-12-22 12:05GN/GMOS-NZ401.332.3821.80222.12.0754.0
2025-12-22 12:29GN/GMOS-NB480+/5401801.232.3831.80222.12.0754.0
2025-12-22 12:38GN/GMOS-NB480+/5501801.202.3831.80222.12.0754.0
2026-01-20 09:53GN/GMOS-Ng501.033.2792.2971.40.7882.9
2026-01-20 09:55GN/GMOS-Nr501.023.2792.2971.40.7882.9
2026-01-20 09:57GN/GMOS-Ni501.023.2792.2981.40.7882.9
2026-01-20 10:01GN/GMOS-NZ1001.023.2792.2981.40.7882.9
2026-01-21 08:03GN/GMOS-NB480+/5404501.253.3092.3261.00.7483.8
2026-01-21 08:17GN/GMOS-NB480+/5504501.203.3092.3261.00.7483.9

Table 1 (continued).

UTC date/timeFacility/instrumentFilter/gratingtexpt_{\mathrm{exp}}XXrhr_hΔ\Deltaα\alphaδE\delta_Et−Tpt-T_p
2026-02-05 05:49GN/GMOS-Ng501.403.7942.8685.90.1798.7
2026-02-05 05:51GN/GMOS-Nr501.393.7942.8685.90.1798.7
2026-02-05 05:53GN/GMOS-Ni501.383.7942.8685.90.1798.8
2026-02-05 05:56GN/GMOS-NZ1001.363.7942.8685.90.1798.8
2026-02-06 –K1/MOSFIREJ521.083.8332.9176.20.1499.9
2026-02-06 –K1/MOSFIREKs1401.073.8332.9176.20.1499.9
2026-02-07 07:42GN/GMOS-Ng501.043.8622.9546.50.11100.8
2026-02-07 07:44GN/GMOS-Nr501.033.8622.9546.50.11100.8
2026-02-07 07:46GN/GMOS-Ni501.033.8622.9546.50.11100.8
2026-02-07 07:49GN/GMOS-NZ1001.033.8632.9546.50.11100.8
2026-02-17 09:59GN/GMOS-Ng721.164.1963.3978.8-0.16110.9
2026-02-17 10:07GN/GMOS-Nr1441.194.1963.3988.8-0.16110.9
2026-02-17 10:04GN/GMOS-Ni721.184.1963.3988.8-0.16110.9
2026-02-17 10:08GN/GMOS-NZ1441.194.1963.3988.8-0.16110.9
2026-02-17 10:31GN/GMOS-NB480+/5405001.274.1973.3998.8-0.16110.9
2026-02-17 10:45GN/GMOS-NB480+/5505001.334.1973.3998.8-0.16111.0
2026-03-27 08:26GN/GMOS-Ng1201.565.4655.29310.5-0.62148.9
2026-03-27 08:29GN/GMOS-Nr1201.595.4655.29310.5-0.62148.9
2026-03-27 08:32GN/GMOS-Ni1201.615.4655.29310.5-0.62148.9
2026-03-27 08:37GN/GMOS-NZ2401.655.4665.29410.5-0.62148.9
2026-03-27 07:30GN/GMOS-NB480+/54018001.265.4645.29110.5-0.62148.8
2026-03-27 08:06GN/GMOS-NB480+/55018001.435.4655.29210.5-0.62148.8
2026-05-15 06:16K1/LRISG1380–7.1187.7616.0-0.62197.8
2026-05-15 06:16K1/LRISR900–7.1187.7616.0-0.62197.8
2026-05-15 06:35K1/LRISI720–7.1197.7626.0-0.62197.8
2025-09-01 00:00‡K1/LRIS400/3400–2.002.5152.57522.80.52-58.5
2025-09-01 00:00‡K1/LRIS400/8500–2.102.5152.57522.80.52-58.5

Table 1 (continued, PDF p. 7).

Times are UTC exposure-weighted midpoints, except † (input start time) and ‡ (confirmed September 1 date; 00:00 supplies geometry because the midpoint is unavailable). ∗July Y is an uncalibrated stack with incomplete target coverage; airmass comes from its reference exposure and 611 s is the total input time, with nonuniform coverage. Exposure times are in seconds, distances in au, angles in degrees, and t−Tpt-T_p in days. δE\delta_E is the signed orbital-plane viewing angle ([14]). GS/GN and K1 denote Gemini South/North and Keck I. GMOS B480+ uses G5327 (South) or G5309 (North) with 1-arcsec slits. GNIRS uses 32/mmSB_G5533, orders 3–7, and a 0.68-arcsec slit; F2 uses JH-G5801 or HK-G5802 and an 8-pixel (1.43-arcsec) slit. LRIS uses 400/3400 blue and 400/8500 red dispersers. Reduced headers verify neither the adopted 0.7-arcsec LRIS slit nor binning; its 70-pixel extraction spans 9.45 arcsec assuming unbinned 0.135-arcsec pixels. The electronic exposure log identifies correlated calibrated alternatives, counted once.

Results

Optical results

Optical imaging overview

Figure 1 and 2 show the north-up, east-left GMOS-S and GMOS-N/LRIS sequences, with dates, filters, total exposures, scales, and projected directions marked. Orange denotes the Sun and blue the heliocentric velocity throughout the imaging figures. The sequence shows the coma broadening and the tail becoming more prominent. Measurements use unsmoothed arrays; § 2.1 describes the masks and display smoothing.

Figure 2. Gemini North GMOS and Keck LRIS imaging in gg, r/Rr/R, ii, ZZ columns, following Figure 1. Absent filters are blank and the LRIS row is centered. White circles locate the comet in March GMOS and May LRIS images; masked pixels are black.

Figure 2 (continued). Continued.

Common-aperture photometry and optical colors

We measure photometry in a 10,00010{,}000-km-radius aperture (20,00020{,}000-km diameter), subtending 1.78–7.67 arcsec (Table 2), with fractional pixel weights at its boundary. Two calibrated apertures lack less than 0.15% of their area; we estimate the missing flux from valid pixels at equal cometocentric radius. The uncertainty includes the larger of half this correction and its difference from a diametrically opposed-pixel estimate. Larger gaps remain uncorrected. Tables list valid areas and quality flags.

Sky subtraction removes some extended coma, so the photometry measures aperture excess above the local background. We test sky annuli, centering, neighboring stars, masks, and coverage. Statistical errors use the diagonal variance; calibration and background uncertainties are recorded separately. Colors use same-night measurements in a common photometric system and physical aperture: u−gu-g, g−rg-r, r−ir-i, g−ig-i, i−zi-z, J−KsJ-K_s, Y−JY-J, J−HJ-H, and H−KsH-K_s where available. We distinguish KsK_s from KK.

We transform visible photometry to approximate SDSS AB equivalents [114]. For x=(g−r)P1x=(g-r)_{\mathrm{P1}}, gSDSS−gP1=0.013+0.145x+0.019x2g_{\mathrm{SDSS}}-g_{\mathrm{P1}}=0.013+0.145x+0.019x^2, rSDSS−rP1=−0.001+0.004x+0.007x2r_{\mathrm{SDSS}}-r_{\mathrm{P1}}=-0.001+0.004x+0.007x^2, iSDSS−iP1=−0.005+0.011x+0.010x2i_{\mathrm{SDSS}}-i_{\mathrm{P1}}=-0.005+0.011x+0.010x^2, and zSDSS−zP1=0.013−0.039x−0.012x2z_{\mathrm{SDSS}}-z_{\mathrm{P1}}=0.013-0.039x-0.012x^2. Monte Carlo errors include magnitude uncertainties, stellar residuals of 0.008, 0.004, 0.004, and 0.010 mag, respectively, and shared-band covariance. Without simultaneous g−rg-r, we adopt x=0.60±0.15x=0.60\pm0.15 for this correction. SDSS uu is unchanged. The electronic tables retain native fluxes and passbands.

Applying stellar transformations to an emission-bearing comet introduces an unquantified systematic error. MPC-scaled rr is an SDSS-like proxy with a 0.20-mag error floor and is excluded from colors. Independently calibrated spectra and gas rates retain their flux scales.

Table 2 contains 29 campaign measurements on 12 dates, including two MPC-normalized rr estimates, and 27 aperture-normalized Liverpool measurements from 15 published epochs [45]. The imaging log accounts for all 17 dates and 57 independent filter stacks. Missing-area fractions in November ZZ and December rr are 0.138% and 0.056%, with recovered flux fractions of 0.069% and 0.140%. Other stacks lack zero points or contain large masks or bright-star artifacts. The August 28 and September 8 uu measurements give A(0)fρ=610A(0)f\rho=610 and 1271 cm, subject to gas contamination and sparse-standard calibration. § 3.2 presents MOSFIRE photometry.

Matched grigri photometry is available on August 27, February 7, and February 17. Figure 5 includes all accepted gg, rr, RR measurements; Figure 6 omits the two MPC-normalized rr points from display. Both include errors, some smaller than the symbols.

Figure 5. Brightness and Halley–Marcus phase-corrected activity versus time from perihelion; the upper axis gives heliocentric distance. Top: approximate SDSS rr and g−0.732g-0.732 in a 10,000-km-radius aperture, with constant-A(0∘)fρA(0^\circ)f\rho curves. Blue triangles mark Liverpool rr (13 epochs), and open green inverted triangles mark shifted gg (14 epochs) ([45]); open diamonds mark MPC-scaled points. Liverpool magnitudes retain their published zero points, with the small aperture adjustment described in the text. Bottom: red-band fits exclude gas-bearing gg. Solid curves and shading show our fits and formal 1σ1\sigma errors on the mean; the residual scatter is larger. Comparison slopes are b=−1.14±0.16b=-1.14\pm0.16 inbound and −1.58±0.09-1.58\pm0.09 outbound for broad GG ([45], blue/red dashed); −1.8±0.3-1.8\pm0.3 ([66], black dotted); −3.9-3.9 to −1.1-1.1 across 3.3 au ([115], purple dot-dashed); and −2.5±0.3-2.5\pm0.3 outbound ([58], teal dash-dot). These laws are normalized to our corresponding branch at 3 au and cover the published intervals, including Hui’s December–January HST data in 200–1000-km-radius apertures. Aperture and phase-law differences limit comparisons to curve shape. Shading applies only to our fits. The right axis gives geometry-normalized rr brightness; dotted vertical lines mark perihelion. Some errors are smaller than the symbols.

Figure 6. Scattering cross section in a 10,000-km-radius aperture, for albedo 0.10, the Halley–Marcus phase law, and approximate SDSS photometry. Triangles show Liverpool data [45]; stars show discovery photometry [12]. The two MPC-normalized GMOS rr estimates are omitted from the display and retained in the numerical analysis. Liverpool values use the common aperture and phase correction, with no fitted normalization offsets. Shifted gg is a gas-bearing rr proxy; fits use r/Rr/R. The dotted vertical line marks perihelion; gray shading marks the observing gap. Blue/red trends cover −150≤t−Tp≤−50-150 \le t-T_{p} \le-50 days inbound and +100≤t−Tp≤+214.9+100 \le t-T_{p} \le+214.9 days outbound. Bands give formal 1σ1\sigma errors on the fitted mean, scaled by residual scatter. The late fit uses February 7, February 17, and May 15 GMOS/LRIS data; dashed segments extrapolate beyond the fitted epochs. The inbound zero, UT 2025 June 6 (t−Tp=−144.8±1.5t-T_{p}=-144.8\pm1.5 days), follows the TESS and ZTF prediscovery detections [76] (Q. Ye et al. 2025). The outbound zero, UT 2026 June 1 (t−Tp=+214.9±6.1t-T_{p}=+214.9\pm6.1 days), extrapolates the linear decline. Some errors are smaller than the symbols.

UTC dateInstrumentBandt−Tpt-T_{\mathrm{p}} (d)θρ\theta_{\rho} (arcsec)MagnitudeA(0)fρA(0)f\rho (cm)
2026-02-07GMOS-Nz100.84.6716.222 ± 0.032724 ± 21
2026-02-07GMOS-Ng100.84.6717.176 ± 0.036522 ± 17
2026-02-07GMOS-Ni100.84.6716.296 ± 0.030688 ± 19
2026-02-07GMOS-Nr100.84.6716.512 ± 0.030631 ± 17
2026-02-17GMOS-Nz110.94.0616.955 ± 0.032625 ± 18
2026-02-17GMOS-Ng110.94.0618.079 ± 0.036386 ± 13
2026-02-17GMOS-Ni110.94.0617.068 ± 0.030573 ± 16
2026-02-17GMOS-Nr110.94.0617.302 ± 0.030517 ± 14
2026-05-15LRISr197.81.7822.024 ± 0.11691 ± 10
2026-05-15LRISg197.81.7822.876 ± 0.14063 ± 8
2026-05-15LRISi197.81.7821.753 ± 0.197104 ± 19
2025-07-17LiverpoolL^{L}r-103.54.5117.119 ± 0.030428 ± 12
2025-08-02LiverpoolL^{L}g-87.64.9417.225 ± 0.271458 ± 114
2025-08-02LiverpoolL^{L}r-87.64.9416.595 ± 0.020536 ± 10
2025-08-13LiverpoolL^{L}g-76.65.1616.629 ± 0.020657 ± 12
2025-08-13LiverpoolL^{L}r-76.65.1616.109 ± 0.010694 ± 6
2025-08-14LiverpoolL^{L}g-75.65.1816.685 ± 0.091612 ± 51
2025-08-14LiverpoolL^{L}r-75.65.1816.055 ± 0.060717 ± 39
2025-08-17LiverpoolL^{L}g-72.65.2216.479 ± 0.030699 ± 19
2025-08-17LiverpoolL^{L}r-72.65.2215.909 ± 0.010773 ± 7
2025-08-27LiverpoolL^{L}g-62.65.3216.054 ± 0.020845 ± 16
2025-08-27LiverpoolL^{L}r-62.65.3215.553 ± 0.080877 ± 65
2025-09-03LiverpoolL^{L}g-55.65.3615.837 ± 0.010877 ± 8
2025-09-03LiverpoolL^{L}r-55.65.3615.277 ± 0.010961 ± 9
2025-12-06LiverpoolL^{L}g37.77.4112.734 ± 0.0305806 ± 161
2025-12-07LiverpoolL^{L}g38.77.4513.038 ± 0.0204456 ± 82
2025-12-19LiverpoolL^{L}g50.77.6612.803 ± 0.0306426 ± 177
2025-12-19LiverpoolL^{L}r50.77.6612.433 ± 0.0205910 ± 109
2026-01-11LiverpoolL^{L}g73.66.7314.099 ± 0.0302617 ± 73
2026-01-11LiverpoolL^{L}r73.66.7313.729 ± 0.0202410 ± 44
2026-01-20LiverpoolL^{L}g82.66.0214.785 ± 0.0301684 ± 47
2026-01-20LiverpoolL^{L}r82.66.0214.205 ± 0.0301880 ± 52
2026-01-28LiverpoolL^{L}g91.45.3415.905 ± 0.030962 ± 26
2026-01-28LiverpoolL^{L}r91.45.3415.255 ± 0.0201146 ± 21
2026-02-08LiverpoolL^{L}g102.54.5516.605 ± 0.050969 ± 45
2026-02-08LiverpoolL^{L}r102.54.5516.075 ± 0.0301034 ± 29
2026-02-21LiverpoolL^{L}g115.43.8217.506 ± 0.080809 ± 60
2026-02-21LiverpoolL^{L}r115.43.8217.196 ± 0.040704 ± 26

Table 2 (continued).

Magnitudes are approximate SDSS AB equivalents; native values and transformations are retained electronically. Tonry et al. [114] stellar transformations leave cometary emission-band differences uncorrected. Without simultaneous g−rg-r, PS1 0.60±0.150.60 \pm0.15 supplies the passband correction. LRIS RR uses measured g−rg-r. c.: small missing-area correction. MM: MPC-scaled SDSS-like rr, with a 0.20-mag systematic floor, excluded from colors. LL: published Liverpool Sloan photometry, adjusted by less than 0.008 mag to the common aperture, without further passband or fitted zero-point shifts. Gas can contribute to gg-band activity.

Discovery photometry gives SDSS g−r=0.84±0.05g-r=0.84 \pm0.05, r−i=0.16±0.03r-i=0.16 \pm0.03, and i−z=−0.02±0.07i-z=-0.02 \pm0.07 ([12]); the August Gemini report gives g−r=0.68±0.01g-r=0.68 \pm0.01 and r−i=0.32±0.01r-i=0.32 \pm0.01 ([22]). Our three matched GMOS epochs, transformed as above (Figure 3), give weighted means g−r=0.732±0.029g-r=0.732 \pm0.029 and r−i=0.237±0.025r-i=0.237 \pm0.025 mag. Constant-color fits yield χ2=2.85\chi^2=2.85 and 0.60 for two degrees of freedom (p=0.24p=0.24 and 0.74), indicating no significant change. T. Santana-Ros et al. [105] reported increasing activity and reddening during July 2–29; their longer homogeneous campaign found stable reddish colors T. Santana-Ros et al. [106]. Coverage, passband, and aperture differences complicate direct comparison.

Figure 3. Optical colors: g−rg-r versus r−ir-i (left) and r−zr-z (right). Blue/red/green asteroid groups follow Z. Ivezić et al. [61]; comet and outer-Solar-System samples follow M. Solontoi et al. [112] and E. O. Ofek [88]. The 1I point is purple with yellow errors; 2I has a black core with red outline and errors, using the values in B. T. Bolin et al. [12]. Left: arithmetic means of the July 3 and 6 discovery colors, g−r=0.835±0.034g-r = 0.835 \pm0.034 and r−i=0.160±0.025r-i = 0.160 \pm0.025 mag, assuming independent epoch errors and excluding shared calibration uncertainty. Right: the same July 3–6 g−rg-r mean paired with July 6-only r−z=0.160±0.054r-z = 0.160 \pm0.054 mag. Dated GMOS points use the approximate stellar PS1-to-SDSS transformation (J. L. Tonry et al. [114]); ellipses include shared-band covariance. Three epochs have matched calibrated grigri colors, two also with ZZ.

HATPI found no significant short-timescale variability after August 6 J. D. Hartman et al. [53]. Its broad-GG brightening index differs from our red-band, phase-corrected AfρAf\rho slope in passband, phase treatment, and aperture weighting.

Matched i−zi-z measurements on 2025 September 5 and 8 and 2026 February 7 and 17 give an inverse-variance weighted mean 0.109 ± 0.022 mag and χ2=0.88\chi^2 = 0.88 for three degrees of freedom. Errors assume independent epochs and exclude shared comet-specific passband errors. September uses the assumed g−rg-r transformation color because matched rr photometry is unavailable; February uses measured g−rg-r.

The approximate SDSS colors on August 27, February 7, and February 17 are g−r=0.755±0.050g-r = 0.755 \pm0.050, 0.665 ± 0.050, and 0.777 ± 0.050, and r−i=0.262±0.043r-i = 0.262 \pm0.043, 0.216 ± 0.043, and 0.234 ± 0.043. February also gives r−z=0.290±0.043r-z = 0.290 \pm0.043 and 0.347 ± 0.043. Calibration and aperture exclusions limit the matched sample Figure 4. The July 15 PS1 colors g−r=0.603±0.031g-r = 0.603 \pm0.031 and r−i=0.210±0.031r-i = 0.210 \pm0.031 J. Beniyama [8] transform to 0.706 ± 0.037 and 0.209 ± 0.031 mag. Post-perihelion averages of g−r=0.52±0.09g-r = 0.52 \pm0.09 and r−i=0.13±0.08r-i = 0.13 \pm0.08 G. Ahuja et al. [3] are bluer, although aperture and passband differences limit a compositional interpretation.

Figure 4. Figure 4. Colors versus t−Tpt-T_p, using Tp=JD 2460977.995263 TDBT_p = \mathrm{JD}\ 2460977.995263\ \mathrm{TDB} from JPL Horizons solution 54 [42], converted to UTC. GMOS/LRIS colors are approximate SDSS equivalents; P200/APO retain published SDSS values. July PS1 colors [8] use the same Tonry transformation. The purple band spans the observing interval of [45], with weighted mean g−r=0.58g-r = 0.58 mag and reported scatter 0.08 mag; no r−ir-i mean is available. All matched epochs and errors are shown.

Rubin commissioning photometry gives g−r=0.657±0.013g-r = 0.657 \pm0.013 and r−i=0.235±0.018r-i = 0.235 \pm0.018 mag C. O. Chandler et al. [32], with similar r−ir-i and bluer g−rg-r than our means. UH88/SNIFS spectra at rh=4.4r_h = 4.4–2.5 au give mostly stable synthetic PS1 colors g−r=0.69g-r = 0.69–0.75 and r−i=0.26r-i = 0.26–0.30 mag W. B. Hoogendam et al. [56]. These results support a persistently red continuum across differing passbands and extraction apertures.

Figure 3 includes 6,738 distinct SDSS Moving Object Catalog asteroids Z. Ivezić et al. [61] with 2<a<3.52 < a < 3.5 au, q>1.67q > 1.67 au, and errors ≤0.05\le0.05 mag in g,r,i,zg,r,i,z. Using a∗=0.89(g−r)+0.45(r−i)−0.57a^* = 0.89(g-r) + 0.45(r-i) - 0.57, we define C-like colors by a∗<0a^* < 0, V-like colors by a∗≥0a^* \ge0 and i−z<−0.15i-z < -0.15, and S-like colors by the remaining a∗≥0a^* \ge0 objects. These photometric groups approximate taxonomy. We also plot 31 active-comet observations M. Solontoi et al. [112] and usable median colors from E. O. Ofek [88], retaining their published apertures and magnitude estimators.

We adopt the ISO colors compiled by B. T. Bolin et al. [12]: (g−r,r−i,r−z)=(0.84,0.31,0.41)(g-r,r-i,r-z) = (0.84, 0.31, 0.41) with marginal errors (0.05, 0.09, 0.10) mag for 11 K. J. Meech et al. [80], and (0.63,0.20,−0.03)(0.63, 0.20, -0.03) with (0.05, 0.05, 0.059) mag for 21 B. T. Bolin et al. [19]; covariances are unavailable. The discovery g−rg-r and r−ir-i average July 3 and their r−zr-z uses July 6 alone. Figure 4 includes the campaign-average g−r=0.58±0.08g-r = 0.58 \pm0.08 of A. Graykowski et al. [45], with the reported dispersion as its uncertainty; that study provides no mean r−ir-i.

Dust activity and scattering cross section

For a calibrated bandpass flux ratio, the conventional aperture activity parameter M. F. A’Hearn et al. [2] is

Afρ(α)=4rh2Δcm2ρcm10−0.4(mb−m⊙,b),(1)Af\rho(\alpha) = \frac{4r_h^2\Delta_{\rm cm}^2}{\rho_{\rm cm}}10^{-0.4(m_b-m_{\odot,b})}, \tag*{(1)}

Here rhr_h is heliocentric distance in au, Δcm\Delta_{\rm cm} is observer distance, and ρcm\rho_{\rm cm} is aperture radius, both in cm. Comet and solar magnitudes share a passband and magnitude system. We adopt solar AB magnitudes at 1 au from Table 3, column 6, of C. N. A. Willmer [118]: SDSS m⊙,u=−25.18m_{\odot,u} = -25.18, m⊙,g=−26.47m_{\odot,g} = -26.47, m⊙,r=−26.93m_{\odot,r} = -26.93, m⊙,i=−27.05m_{\odot,i} = -27.05, and m⊙,z=−27.07m_{\odot,z} = -27.07 mag. The Halley–Marcus dust phase function D. G. Schleicher et al. [109]; D. G. Schleicher [107], normalized to ΦHM(0)=1\Phi_{\rm HM}(0) = 1, gives

A(0∘)fρ=Afρ(α)/ΦHM(α).(2)A(0^\circ)f\rho= Af\rho(\alpha)/\Phi_{\rm HM}(\alpha). \tag*{(2)}

Molecular emission contributes to gg-band activity; dust estimates use rr, including the May Cousins RR photometry transformed to approximate SDSS rr. Native measurements remain in the electronic table.

Figure 5 follows the LD2_2 presentation B. T. Bolin et al. [14], with constant-A(0∘)fρA(0^\circ)f\rho curves evaluated at the observing geometry, geometry-normalized brightness on the second ordinate, and heliocentric distance on the upper abscissa. We plot g−0.732g-0.732 beside rr using the mean GMOS color; the shift retains gas contamination. P200/APO B. T. Bolin et al. [12] and Liverpool IO:O Sloan photometry A. Graykowski et al. [45] extend the sample. The 2-m Liverpool Telescope contributes 13 rr and 14 gg epochs from 2025 July 17 to 2026 February 21. We omit its December 6 and 7 rr measurements and retain gg. Published magnitudes and geometry are adopted, with t−Tpt-T_p recomputed from UTC dates; broad-GG measurements are excluded from the rr sample.

Following A. Graykowski et al. [45], we fit phase-corrected red-band activity. At fixed physical aperture, their law m(0)=H0+5log⁡10Δ+2.5nlog⁡10rhm(0) = H_0 + 5\log_{10}\Delta+ 2.5n\log_{10}r_h and Equation (1) imply A(0∘)fρ∝rh2−nA(0^\circ)f\rho\propto r_h^{2-n}, or b=2−nb = 2-n. Their broad-GG indices give b=−1.14±0.16b = -1.14 \pm0.16 before the September brightening and −1.58±0.09-1.58 \pm0.09 outbound. We fit

A(0∘)fρ=A3(rh3 au)b,(3)A(0^\circ)f\rho= A_3\left(\frac{r_h}{3\,\mathrm{au}}\right)^b, \tag*{(3)}

where A3A_3 is activity at 3 au. Separate inbound and outbound fits use red-band measurements, including Liverpool, in log⁡10A(0∘)fρ\log_{10} A(0^\circ)f\rho with σlog⁡A=0.4σm\sigma_{\log A} = 0.4\sigma_m. Covariances are multiplied by max⁡(1,χ2/ν)\max(1,\chi^2/\nu) for excess scatter. Our inbound series ends on September 14, before the brightening reported by A. Graykowski et al. [45]; their broad-GG results provide slope comparisons.

Published P200/APO SDSS and Liverpool Sloan magnitudes receive no further passband transformation. We normalize angular apertures to 10,000 km assuming steady-coma F∝ρF \propto\rho: m10000=mρ−2.5log⁡10(10000/ρkm)m_{10000} = m_\rho- 2.5\log_{10}(10000/\rho_{\rm km}). Liverpool’s apertures already approximate 10,000 km (A. Graykowski et al. [45]); tabulated angular radii and rounded distances give 9955–10,065 km and corrections of −0.005-0.005 to +0.007+0.007 mag. Their zero points are retained, with no additional camera or seeing correction. All facilities use the same Halley–Marcus phase law. For LRIS, we solve RC−rP1=−0.137−0.108x−0.029x2R_C-r_{P1} = -0.137 - 0.108x - 0.029x^2 (J. L. Tonry et al. [114]) with measured gP1−RCg_{P1}-R_C, then transform rP1r_{P1} to SDSS. Errors include the 0.015-mag RCR_C and tabulated rr transformation residuals, with the comet-specific uncertainty noted above. Sensitivity tests exclude LRIS or allow a separate Liverpool normalization.

Figure 5 compares growth laws normalized to our A3A_3 at 3 au. At 3 au, fixed 10,000-km RR photometry gives n=3.8±0.3n = 3.8 \pm0.3, hence b=−1.8±0.3b = -1.8 \pm0.3, for a linear phase coefficient 0.04±0.020.04 \pm0.02 mag deg−1^{-1} (D. Jewitt & J. Luu [66]). J. L. Tonry et al. [115] find cross-section growth changing from rh−3.9r_h^{-3.9} to rh−1.1r_h^{-1.1} near 3.3 au in fixed angular apertures with a G=0.15G = 0.15

phase function. Their purple dot-dashed law and the black dotted Jewitt–Luu curve span the published coverage. The common normalization compares growth shapes across differing apertures, passbands, and phase laws.

The outbound HSTHST index n=4.5±0.3n = 4.5 \pm0.3 gives b=−2.5±0.3b = -2.5 \pm0.3 [58], shallower than our −3.34±0.43-3.34 \pm0.43 by 1.6σ1.6\sigma. Its teal dash-dot curve is normalized to outbound A3A_3 over December 12–January 22 (rh=2.108r_h = 2.108–3.3483.348 au). HSTHST F350LP photometry uses 200–1000-km apertures and a linear-exponential phase correction.

We calculate an effective scattering cross section from a phase-corrected absolute magnitude,

Hb=mb−5log⁡10(rhΔ)+2.5log⁡10ΦHM(α),(4)H_b = m_b - 5\log_{10}(r_h\Delta) + 2.5\log_{10}\Phi_{\mathrm{HM}}(\alpha), \tag*{(4)}
Cb=π(aukm)2pb100.4(m⊙,b−Hb).(5)C_b = \frac{\pi(\mathrm{au}_{\mathrm{km}})^2}{p_b}10^{0.4(m_{\odot,b}-H_b)}. \tag*{(5)}

Here bb denotes the band, mbm_b is apparent magnitude, and HbH_b is its value at rh=Δ=1r_h = \Delta= 1 au and zero phase angle. Distances are in au, α\alpha is the Sun–comet–observer angle, and ΦHM(0)=1\Phi_{\mathrm{HM}}(0) = 1. Cross section CbC_b is in km2^2, pbp_b is dust geometric albedo, aukm=149597870.7\mathrm{au}_{\mathrm{km}} = 149597870.7 km, and m⊙,bm_{\odot,b} is solar apparent magnitude at 1 au [118]. The 2I/Borisov expression [19] becomes CV≃1.5×106pV−110−0.4HVC_V \simeq1.5 \times10^6p_V^{-1}10^{-0.4H_V} km2^2 in VV; each band requires its corresponding solar magnitude. We adopt p=0.10p = 0.10, with C∝p−1C \propto p^{-1} (Figure 6).

Cross-section growth, maximum, and fading.—Following B. T. Bolin et al. [19], we fit Cr=a+b(t−tref)C_r = a + b(t - t_{\mathrm{ref}}) to inbound growth and late outbound fading, where t=tUTC−Tpt = t_{\mathrm{UTC}} - T_p in days and treft_{\mathrm{ref}} is the weighted mean epoch. The extrapolated zero is t0=tref−a/bt_0 = t_{\mathrm{ref}} - a/b. We weight by photometric errors and inflate covariance by max⁡(1,χ2/ν)\max(1, \chi^2/\nu) for residual scatter, with ν\nu degrees of freedom. Errors assume the adopted linear model, phase law, albedo, and calibration.

The 11 inbound measurements give dCr/dt=8.15±0.27dC_r/dt = 8.15 \pm0.27 km2^2 day−1^{-1}, χ2/ν=12.5\chi^2/\nu= 12.5, and a zero at t=−144.8±1.5t = -144.8 \pm1.5 days (UT 2025 June 6; Figure 6). Excluding Liverpool gives 6.29±0.286.29 \pm0.28 km2^2 day−1^{-1} and a 2025 May 29 zero. TESS detections during May 7–June 2 at rh=6.35r_h = 6.35–5.46 au show evidence for weak activity [76]; clear ZTF nightly-stack detections begin May 15 (Q. Ye et al. 2025). The start of TESS coverage and first clear ZTF detection precede our zero by about 30 and 22 days, respectively. Q. Ye et al. (2025) inferred sustained outflow by rh∼9r_h \sim9 au under their grain-speed and residence-time assumptions. Our zero therefore describes the later growth interval; activity began earlier. Applying the 2I/Borisov onset-extrapolation method [65] depends on the interval sampled.

A linear fit to all 12 outbound points gives χ2/ν≃385\chi^2/\nu\simeq385. The 2026 February 7, February 17, and May 15 GMOS/LRIS points give dCr/dt=−4.10±0.35dC_r/dt = -4.10 \pm0.35 km2^2 day−1^{-1} and a zero at t=+214.9±6.1t = +214.9 \pm6.1 days (UT 2026 June 1), with χ2/ν=7.5\chi^2/\nu= 7.5 for one degree of freedom. Figure 6 extends this trend about 17 days beyond the last image. Including the two late Liverpool epochs gives −4.75±1.06-4.75 \pm1.06 km2^2 day−1^{-1} and a zero near 2026 May 30 (t=+212.5±16.8t = +212.5 \pm16.8 days), with greater scatter. Continued linear fading would remove the fitted cross section by late May to early June. The sparse baseline and residual scatter limit this extrapolation: aperture cross section depends on dust production and escape, so the coma could persist longer if fading slows or production resumes.

Equal-epoch means in seven-day bins anchored on perihelion, interpolated with a shape-preserving cubic, peak at Cr≃3.15×103C_r \simeq3.15 \times10^3 km2^2 at t=+52.4t = +52.4 days. The largest individual measurement is Liverpool’s December 19 value, 4.64×1034.64 \times10^3 km2^2 at +50.7+50.7 days. Tests with 3- and 14-day bins and half-bin shifts give peaks at +50.7+50.7–52.452.4 days and (3.15(3.15–4.64)×1034.64) \times10^3 km2^2. The 72.7-day gap from −44.5-44.5 to +28.2+28.2 days permits an earlier unobserved peak. Interpolated Cr(Tp)C_r(T_p) is 2.47×1032.47 \times10^3 km2^2 for the cubic and 2.26×1032.26 \times10^3 km2^2 for a straight line between the bounding epochs.

Heliocentric dependence of phase-corrected dust activity

The 11 inbound measurements at rh=4.43r_h = 4.43–2.11 au give A3=680±21A_3 = 680 \pm21 cm and b=−2.10±0.14b = -2.10 \pm0.14; 12 outbound measurements at 1.70–7.12 au give A3=1940±189A_3 = 1940 \pm189 cm and b=−3.34±0.43b = -3.34 \pm0.43. Outbound normalization is 2.85±0.292.85 \pm0.29 times higher. Before covariance inflation, χ2/ν=47.0\chi^2/\nu= 47.0 and 154.3, respectively. Confidence bands describe the fitted mean; variability, aperture assumptions, and inter-facility calibration contribute additional scatter.

The May LRIS point extends the baseline to t−Tp=+197.8t - T_p = +197.8 days; omitting it gives outbound b=−3.33±0.47b = -3.33 \pm0.47. Excluding Liverpool gives b=−1.83±0.05b = -1.83 \pm0.05 inbound and −2.56±0.12-2.56 \pm0.12 outbound; Liverpool alone gives −1.54±0.15-1.54 \pm0.15 and −3.44±0.19-3.44 \pm0.19. Free Liverpool offsets give −1.64±0.11-1.64 \pm0.11 and −3.10±0.18-3.10 \pm0.18, with Liverpool higher by 0.122 and 0.263 dex, respectively, and substantial residual scatter. These sensitivity-test offsets are not applied to the photometry or adopted fits. The electronic table provides measurements, covariances, and facility/MPC/LRIS exclusion tests.

Our outbound slope is steeper than the broad-GG value b=−1.58±0.09b = -1.58 \pm0.09 [45], with gas and facility differences complicating interpretation. The May point alone provides no evidence for another break. Since

A(0∘)fρA(0^\circ)f\rho measures albedo-weighted cross section per aperture radius, grain residence times, sizes, and phase behavior affect its evolution. The higher outbound normalization and decline trace dust persistence and escape; mass loss also requires a grain and velocity model.

The HST phase model [58] has linear slope 0.026±0.0060.026 \pm0.006 mag deg−1^{-1}, opposition enhancement 0.19±0.010.19 \pm0.01 mag, and e-folding width 3±1∘3 \pm1^\circ. Phase behavior and orbital-plane projection make January especially geometry-sensitive. The HST radius 1.3±0.21.3 \pm0.2 km sets our shell-depth scale and is consistent with the earlier 2.8-km upper bound [63].

At 10,000 km, B. Liu et al. [74] find cross-section indices −1.86±0.27-1.86 \pm0.27 inbound and −2.42±0.06-2.42 \pm0.06 outbound, using our activity-index convention with a linear phase correction. Their shallower outbound slope reinforces the sensitivity to sampling, phase correction, and relative calibration.

Coma morphology and dust trajectories

We divide gg and rr images by their azimuthal-median radial profiles [104, 103], using 0.32-arcsec annuli and detector masks. For four affected September rr stacks, both bands use matched 0.48-arcsec native-area bins and annuli, excluding insufficient coverage. Figure 7 omits March and uses no spatial filter. Red, blue, and white denote excess, deficit, and unity relative to the median, with north up and east left. These ratios locate asymmetries but do not separate gas and dust. Profiles use the original masked 0.16-arcsec arrays. Centering tests shift one pixel in each coordinate and compare angular medians at 2–6 and 6–10 arcsec, excluding residual stars and incomplete coverage.

We generate a Finson–Probstein reference grid [39] by propagating particles under solar gravity reduced by radiation pressure,

r¨d=−(1−β)GM⊙rd∣rd∣3,(6)\ddot{\mathbf{r}}_{\mathrm{d}} = -(1-\beta)GM_{\odot}\frac{\mathbf{r}_{\mathrm{d}}}{\lvert\mathbf{r}_{\mathrm{d}}\rvert^3}, \tag*{(6)}

Here β\beta is the radiation-pressure/solar-gravity ratio. Particles begin at the comet’s position and velocity, forming a zero-ejection-speed grid; synchrones share release time and syndynes share β\beta. Figure 8 projects trajectories onto the sky and labels the signed orbital-plane angle δE\delta_E at each midpoint. The grid spans β=0.001\beta= 0.001–11 and ages 5–240 days. For compact spheres, aμm≃0.574Qpr/[ρd(g cm−3)β]a_{\mu\mathrm{m}} \simeq0.574Q_{\mathrm{pr}}/[\rho_{\mathrm{d}}(\mathrm{g\,cm}^{-3})\beta] [29], or 0.6–600 μ\mum for Qpr=1Q_{\mathrm{pr}} = 1 and ρd=1\rho_{\mathrm{d}} = 1 g cm−3^{-3}. Sizes also depend on ejection velocity and release history. Inbound syndynes overlap within 0.5–1.3 degrees at 10 arcsec, preventing unique size or age estimates.

Molecular emission can broaden gg relative to dust-dominated rr [24]. We normalize azimuthal-median profiles at 0.8–1.6 arcsec and compare their 3–6-arcsec means. The g/rg/r shape ratio is 1.01 in August and 1.05, 1.23, 1.06, and 1.10 on September 5, 6, 8, and 14. From November 26 through February 17 it lies within 0.4–3.2% of unity. PSF and background differences prevent converting these ratios to gas fractions.

On August 27, a tail at position angle 105° reaches the 6–10-arcsec annulus, while the opposite enhancement near 285° is confined to the inner coma. This compact sunward feature weakens through September as the tail strengthens (Figure 1 and 2).

By November 26 the coma is broader, developing an asymmetric envelope in December–January and a persistent projected tail in February. The rr profile at 3–6 arcsec, normalized at 0.8–1.6 arcsec, rises from 0.10 on August 27 to 0.26–0.28 in November–February. These angular profiles sample changing physical radii and lack PSF matching. Similar post-perihelion gg and rr profiles show that the envelope includes red continuum. Stars and limited surface-brightness sensitivity hinder tracing the outer coma in March and May.

D. Jewitt & J. Liu (2025) observed a July sunward fan developing a prominent antisolar tail by mid-August, attributing the delay to approximately 100-μ\mum grains moving at 5 m s−1^{-1} under their onset and dust assumptions. Our sequence is consistent with continued release and radiation-pressure redistribution, although changing geometry projects different populations and directions. The antisolar position angle moves from approximately 103° inbound to 292–294° in November–December and 94° in February. Foreshortening is strongest on January 20 (α=1.4∘\alpha= 1.4^\circ), and the perihelion gap prevents dating a sudden morphological transition.

Shifting the August 27 center by 0.16 arcsec preserves both inner maxima in gg and rr; the weaker feature moves one 10-degree bin. It fades in September and is compatible with the sunward structure near 280° reported by T. Santana-Ros et al. [106]; the stronger side follows the tail. A distinct sunward source remains tentative, as do source identifications for the structures seen by G. Ahuja et al. [3]. Residual stars also generate angular maxima, especially in February.

Azimuthal-median division of paired GMOS $g/r$ images

Figure 7. Azimuthal-median division of paired GMOS g/rg/r images, with absent bands blank and March excluded. Orange labels give rhr_h, Δ\Delta, α\alpha, and δE\delta_E (Table 1). Compass/Sun arrows and scale bars are orange; heliocentric-velocity arrows are blue, with black outlines and label backgrounds. North is up and east left. The ratio scale shows deficits in blue, excesses in red, and unity in white; values beyond 0.5–1.5 saturate and masked pixels remain blank. The four September pairs use matched 0.48-arcsec native-area bins and radial annuli; other dates use 0.32-arcsec annuli on a 0.16-arcsec grid. No additional filter is applied. Detector masks, stellar residuals, and centering sensitivity limit jet identification.

Continuation of Figure 7 showing Keck and Gemini monitoring of 3I/ATLAS

Figure 7 (continued).

Syndynes and synchrones on GMOS $r$ images

Figure 8. Syndynes and synchrones on GMOS rr images, excluding March and LRIS. Legends give radiation-pressure parameter β\beta and release age; geometry uses imaging midpoints. Sun arrows are orange and heliocentric-velocity arrows blue. Trajectories assume zero ejection speed and include solar gravity and radiation pressure. Overlapping inbound curves limit grain-size and release-time constraints; no size distribution or production history is fitted. September images use the display treatments in Figure 1, including the September 6 mask-normalized average. Orange labels give rhr_h, Δ\Delta, phase angle α\alpha, and signed orbital-plane viewing angle δE\delta_E (Table 1).

Dust morphology and syndyne models across observing epochs

Figure 8 (continued).

Persistent extended dust is compatible with sustained activity. Recovering the release history requires fitting surface brightness across epochs, as for P/2016 J1 [84] and 2I/Borisov [18]; overlapping syndynes alone do not constrain production rates.

Dust ejection speeds and mass loss

**Dust speeds from transverse widths and the sunward plume—**At small orbital-plane angles, tail widths constrain dust motion normal to the orbit ([64, 14]). Nine GMOS rr epochs provide initial profiles at ∣δE∣<1∘|\delta_E| < 1^\circ. In the unsmoothed, masked 0.16-arcsec arrays, we fit a Gaussian plus linear background to 0.32-arcsec transverse bins in 2-arcsec-wide strips at antisolar offsets of 3–21 arcsec. Initial cuts require >70%> 70\% valid coverage per bin, at least 40 valid bins, amplitude significance >5> 5, fractional width error <0.3< 0.3, and a centroid within 3.8 arcsec of the axis.

We then require fully unmasked footprints, reduced χ2≤5\chi^2 \le5, and no compact contaminants. This excludes affected September and February 7 profiles, strongly foreshortened January profiles, and the innermost February 5 cut with a suspected star. The 13 retained profiles comprise seven at 9–21 arcsec on August 27, two at 5–7 arcsec on February 5, and four at 3–9 arcsec on February 17. These determine the plotted speeds and mass interpolation; the electronic log lists rejected profiles.

For transverse FWHM θW\theta_W, antisolar offset θl\theta_l, and sunward half-maximum offset θD\theta_D in arcsec, the projected lengths at geocentric distance Δ\Delta in meters are

Wsky=ΔθW206265 arcsec rad−1,lsky=Δθl206265 arcsec rad−1,Dsky=ΔθD206265 arcsec rad−1.(7)\begin{aligned} W_{\mathrm{sky}} &= \Delta\frac{\theta_W}{206265\ \mathrm{arcsec\ rad}^{-1}},\\ l_{\mathrm{sky}} &= \Delta\frac{\theta_l}{206265\ \mathrm{arcsec\ rad}^{-1}},\\ D_{\mathrm{sky}} &= \Delta\frac{\theta_D}{206265\ \mathrm{arcsec\ rad}^{-1}}. \tag*{(7)} \end{aligned}

Here 206265 is arcsec per radian and Δ=1.495978707×1011(Δ/au)\Delta= 1.495978707 \times10^{11}(\Delta/\mathrm{au}) m; WskyW_{\mathrm{sky}} is a full width and DskyD_{\mathrm{sky}} a half-width. At grain age τ\tau, the half-width speed is v⊥≃Wsky/(2τP⊥)v_\perp\simeq W_{\mathrm{sky}/(2\tau P_\perp)}, where P⊥=∣n^⋅c^∣P_\perp= |\hat{\mathbf{n}} \cdot\hat{\mathbf{c}}| projects the orbital normal onto the sky-plane cut direction. We derive the normal from archived heliocentric positions and velocities. Constant local radiation pressure gives τ=[2lsky/(βg⊙sin⁡α)]1/2\tau= [2l_{\mathrm{sky}}/(\beta g_\odot\sin\alpha)]^{1/2} and

v⊥=U⊥βWsky2P⊥(βg⊙sin⁡α2lsky)1/2,(8)v_\perp= U_\perp\sqrt{\beta}\frac{W_{\mathrm{sky}}}{2P_\perp}\left(\frac{\beta g_\odot\sin\alpha}{2l_{\mathrm{sky}}}\right)^{1/2}, \tag*{(8)}

Here α\alpha is phase angle and g⊙=0.00593(rh/au)−2 m s−2g_\odot= 0.00593(r_h/\mathrm{au})^{-2}\ \mathrm{m\ s}^{-2} is solar gravitational acceleration. The factor of two converts FWHM to half-width; sin⁡α\sin\alpha deprojects distance for a radial tail. Figure 9 reports U⊥U_\perp; speed requires β\beta. For spherical grains,

Figure 9. GMOS rr transverse-width measurement on 2025 August 27. Left: masked image, north up and east left, with accepted 2-arcsec strips; cyan marks the example cut. Sun/heliocentric-velocity arrows are orange/blue. Upper right: profile after linear-background subtraction, normalized to the Gaussian amplitude; magenta marks FWHM. Errors use sky sampling scaled by max⁡(1,χν2)\sqrt{\max(1,\chi_\nu^2)}. Lower right: accepted widths versus projected antisolar distance (D. Jewitt et al. [64]); the dashed curve uses the epoch-median U⊥U_\perp. Widths include PSF broadening; formal errors exclude projection, grain-age, and PSF systematics.

β=0.574Qpr[ρd/(1000 kg m−3)][a/(1 μm)],(9)\beta= \frac{0.574Q_{\mathrm{pr}}}{[\rho_d/(1000\ \mathrm{kg\ m}^{-3})][a/(1\ \mu\mathrm{m})]}, \tag*{(9)}

where aa is grain radius, ρd\rho_d density, and QprQ_{\mathrm{pr}} radiation-pressure efficiency ([29]). Adopting ρd=1000 kg m−3\rho_d = 1000\ \mathrm{kg\ m}^{-3} and Qpr=1Q_{\mathrm{pr}} = 1 gives β=0.00396\beta= 0.00396 at an illustrative radius of 145 μm\mu\mathrm{m}. Retained epochs have P⊥>0.999P_\perp> 0.999, although release ages remain ambiguous for curved or old trajectories. Median U⊥U_\perp is 19.5, 14.5, and 18.0 m s−1^{-1} on August 27, February 5, and February 17. Profile-to-profile ranges are 15.7–21.7, 14.2–14.7, and 17.5–19.2 m s−1^{-1}; corresponding 145-μ\mum median speeds are 1.23, 0.91, and 1.13 m s−1^{-1}.

For the 2025 August 27 GMOS-S rr cut at 15 arcsec (Figure 9), FWHM 8.66 ±\pm 0.46 arcsec gives Wsky=1.626×107W_{\mathrm{sky}} = 1.626 \times10^7 m. With lsky=2.817×107l_{\mathrm{sky}} = 2.817 \times10^7 m, rh=2.637r_h = 2.637 au, α=22.27∘\alpha= 22.27^\circ, and P⊥=0.99982P_\perp= 0.99982, this gives v⊥=19.48βv_\perp= 19.48\sqrt{\beta} m s−1^{-1}, or 1.23 m s−1^{-1} at β=0.00396\beta= 0.00396.

HSTHST gives v⊥≃7.5βv_\perp\simeq7.5\sqrt{\beta} m s−1^{-1} on July 21 with the same half-width convention ([63]), lower than our medians by factors of 2.6, 1.9, and 2.4. Applying our β=0.574/aμm\beta= 0.574/a_{\mu\mathrm{m}} rather than their 1/aμm1/a_{\mu\mathrm{m}} gives an HSTHST speed of 0.47 m s−1^{-1} at 145 μm\mu\mathrm{m}. Offsets, seeing, coma contributions, and grain ages differ, limiting an acceleration interpretation. Our zero-speed grid spans 0.1–0.4 arcsec in-plane at these offsets; subtracting this extent in quadrature would require a brightness-weighted dust model.

We apply the sunward-coma method ([69]), using HWHM as in [12]. We extract 0.64-arcsec strips around a subpixel photocenter, testing 0.32- and 0.96-arcsec widths and ±\pm0.08-arcsec shifts in each coordinate. Unmasked footprints and peaks exceeding ten times the outer-sky scatter retain ten of twelve rr epochs: September 6 has a masked core and March a weak, irregular profile. August 28 uu and December 19 gg remain diagnostic only. Clean sunward cores on September 5, 8, 14 and February 7 are usable despite contaminated outer-tail profiles. Unlike the discovery analysis, these stacks lack a validated seeing correction.

Radiation pressure gives vsun2=2βg⊙Dv_{\mathrm{sun}}^2=2\beta g_\odot D for turnaround distance DD (D. C. Jewitt & K. J. Meech [69], Equation (4)). Using projected HWHM gives a projected-extent speed; a Sun-aligned trajectory gives D=Dsky/sin⁡αD=D_{\mathrm{sky}}/\sin\alpha. Thus,

vsun,sky=Sskyβ=2βg⊙Dsky,vsun,∥=S∥β=2βg⊙Dskysin⁡α.(10)\begin{aligned} v_{\mathrm{sun,sky}}&=S_{\mathrm{sky}}\sqrt{\beta}=\sqrt{2\beta g_\odot D_{\mathrm{sky}}},\\ v_{\mathrm{sun,\parallel}}&=S_{\parallel}\sqrt{\beta}=\sqrt{\frac{2\beta g_\odot D_{\mathrm{sky}}}{\sin\alpha}}. \tag*{(10)} \end{aligned}

The coefficients SskyS_{\mathrm{sky}} and S∥S_{\parallel} span 33.5–70.5 and 70.3–105.3 m s−1^{-1} (Figure 10); January’s α=1.4∘\alpha=1.4^\circ geometry is excluded from the axial estimate. At 145 μ\mum, the speeds are 2.1–4.4 and 4.4–6.6 m s−1^{-1}. The axial model equates HWHM with turnaround distance, although no intrinsic edge is detected there. Trailed or multiply imaged stars prevent per-epoch PSF deconvolution. Open symbols therefore show observed-width estimates with strip-width and centering sensitivities. PSF broadening raises the inferred speed within the adopted model.

GMOS r sunward-width measurement

Figure 10. GMOS rr sunward-width measurement on 2025 August 27 ([69]). Left: north-up, east-left image with the 0.64-arcsec extraction strip (orange), photocenter (cyan cross), and half-maximum location (cyan line); masked pixels are gray. Right: profile normalized to its near-center peak, with the half-maximum crossing in cyan. Statistical 1σ1\sigma bars use 10,000 Monte Carlo realizations of the diagonal variance map, recalculating strip medians and peak normalization. Most bars are smaller than symbols; the peak is fixed to unity. Errors exclude interpixel covariance, unrepresented sky errors, seeing, and geometry. Shading spans strip-width and centering sensitivity. SskyS_{\rm sky} and S∥S_{\parallel} enter v=Sβv=S\sqrt{\beta} in Equation (10). The HWHM includes seeing and coma broadening; no sharp turnaround edge is identified.

The August 27 GMOS-S rr profile (Figure 10) has HWHM 0.646 arcsec, or 1214 km. Errors propagate diagonal variance through median extraction and peak normalization in 10,000 Monte Carlo realizations, excluding resampling covariance and sky systematics. For Dsky=1.214×106D_{\mathrm{sky}}=1.214\times10^6 m, Ssky=45.5S_{\mathrm{sky}}=45.5 and S∥=73.9S_{\parallel}=73.9 m s−1^{-1}, giving 2.86 and 4.65 m s−1^{-1} at β=0.00396\beta=0.00396.

The discovery relation ve=2Dtβg1/rhv_e=\sqrt{2D_t\beta g_1/r_h} agrees with D. C. Jewitt & K. J. Meech [69] for acceleration g1g_1 at 1 au, rhr_h in au, and DtD_t in meters. Its 1850-km extent, g1=0.006g_1=0.006 m s−2^{-2}, and rh=4.43r_h=4.43 au give 33.6β33.6\sqrt{\beta} m s−1^{-1}, correcting the published 1.1β1.1\sqrt{\beta} m s−1^{-1}. This uses the published extent without further deprojection. HSTHST’s sunward coefficient, 50 m s−1^{-1} (D. Jewitt et al. [63]), is lower than our axial values by factors of 1.4–2.1. Epoch, resolution, profile, and projection differences limit this comparison. The directional contrast is consistent with anisotropic dayside emission, motivating separate mass estimates.

Comparison at fixed v/βv/\sqrt{\beta} avoids mixing size–β\beta prescriptions. Our illustrative 145-μ\mum conversion (β=0.00396\beta=0.00396) gives sunward speeds of 3.1 m s−1^{-1} for HSTHST and 6.3 m s−1^{-1} for Tianwen-1 (D. Jewitt et al. [63]; X. Ren et al. [100]).

Tianwen-1 images from September 30–October 03 give v=100βv=100\sqrt{\beta} m s−1^{-1} (X. Ren et al. [100]), within our axial coefficient range 70.3–105.3 m s−1^{-1}. Their β=10−3\beta=10^{-3}–10−210^{-2} gives our axial speed range 2.1–10.5 m s−1^{-1}, overlapping their 3–10 m s−1^{-1}, with differing epochs, geometry, and extent definitions. The approximately 5-m s−1^{-1} estimate for 100-μ\mum grains (D. Jewitt & J. L. Luu [66]) combines the HSTHST normalization with delayed tail development and β=1/aμm\beta=1/a_{\mu\mathrm{m}}; it shares the HSTHST velocity constraint.

B. Liu et al. [74] obtain an inbound sunward coefficient 190 ±\pm 3 m s−1^{-1} from a deprojected 10%-of-peak boundary, beyond our HWHM. Their January 18 perpendicular coefficient 79 ±\pm 6 m s−1^{-1} uses projected tail distance. Our radialtail correction gives (79±6)sin⁡2.5∘=16.5±1.3 m s−1(79 \pm6)\sqrt{\sin2.5^\circ}=16.5\pm1.3\ {\rm m\,s}^{-1}, within our range. PSF, geometry, and their β=1/aμm\beta=1/a_{\mu{\rm m}} prescription remain different.

T. Santana-Ros et al. [105] report 0.012–1.2 m s−1^{-1}, overlapping our perpendicular 0.9–1.2 m s−1^{-1} at 145 μ\mum, but provide no size-to-speed mapping for a common-radius comparison. The morphology model of F. Moreno et al. [85] uses v=500βrh−1/2cos⁡z m s−1v=500\sqrt{\beta}r_h^{-1/2}\cos z\ {\rm m\,s}^{-1}, with release distance rhr_h in au and solar zenith angle zz. Uniform averaging over the illuminated hemisphere gives v/β=250rh−1/2 m s−1v/\sqrt{\beta}=250r_h^{-1/2}\ {\rm m\,s}^{-1}, or 125–177 m s−1^{-1} at 4–2 au. This terminal-speed prescription exceeds our directional scales and uses grains from 10 μ\mum to an upper cutoff of 1 or 10 cm, density 800 kg m−3^{-3}, and a different release geometry.

Dust mass loss and equivalent surface erosion—We convert phase-corrected rr-band activity to dust mass loss using Equation (5) of B. T. Bolin et al. [12], generalized to size-dependent speed,

M˙d=4πρd3pr⟨av⟩A(0)fρ,⟨av⟩=∫a3v(a,t)nap(a) da∫a2nap(a) da.(11)\dot{M}_{d}=\frac{4\pi\rho_{d}}{3p_{r}}\langle av\rangle A(0)f_{\rho},\qquad\langle av\rangle=\frac{\int a^{3}v(a,t)n_{\rm ap}(a)\,da}{\int a^{2}n_{\rm ap}(a)\,da}. \tag*{(11)}

Here ρd=1000 kg m−3\rho_d=1000\ {\rm kg\,m}^{-3} is grain density, pr=0.10p_r=0.10 is dust geometric albedo, aa is radius, v(a,t)v(a,t) is radial speed, and nap(a)n_{\rm ap}(a) is the aperture grain-number distribution. All quantities use SI units, including A(0)fρA(0)f_\rho in meters. The conversion assumes optically thin, approximately steady outflow; gas, nucleus light, fragmentation, and changing residence times introduce systematic errors.

We adopt nap∝a−3n_{\rm ap}\propto a^{-3} over amin⁡=1 μma_{\min}=1\ \mu{\rm m} to amax⁡=1 mma_{\max}=1\ {\rm mm}, encompassing the 10–100-μ\mum grains favored by MIRI ([7]). The upper cutoff is unconstrained by our images. The area-weighted radius is (amax⁡−amin⁡)/ln⁡(amax⁡/amin⁡)=145 μm(a_{\max}-a_{\min})/\ln(a_{\max}/a_{\min})=145\ \mu{\rm m}. With β=B/a\beta=B/a, B=5.74×10−7 mB=5.74\times10^{-7}\ {\rm m}, and v(a,t)=C(t)βv(a,t)=C(t)\sqrt{\beta},

⟨av⟩=2C(t)B[amax⁡−amin⁡]ln⁡(amax⁡/amin⁡).(12)\langle av\rangle=\frac{2C(t)\sqrt{B}\left[\sqrt{a_{\max}}-\sqrt{a_{\min}}\right]}{\ln(a_{\max}/a_{\min})}. \tag*{(12)}

For isotropic Gaussian velocities, C=2U⊥/πln⁡2=1.355U⊥C=2U_{\perp}/\sqrt{\pi\ln2}=1.355U_{\perp} converts one-dimensional HWHM to mean three-dimensional speed. The Sun-aligned prescription uses C=S∥C=S_{\parallel}, giving a separate mass estimate.

The 23 accepted red-band measurements span 2025 July 3–2026 May 15 (t−Tp=−118.18t-T_p=-118.18 to +197.80+197.80 days), including P200, APO, Liverpool, and transformed LRIS data. We interpolate A(0)fρA(0)f_\rho and each velocity coefficient

Figure 11. Dust-speed constraints from transverse widths and sunward profiles ([69]). Filled orange circles show three artifact-screened perpendicular coefficients, with bars spanning retained cuts. Open green squares show ten projected sunward HWHM coefficients; open teal triangles (“GMOS sunward extent: Sun-aligned model”) show nine estimates that treat HWHM as the axial turnaround distance and correct for phase-angle projection. January’s near-opposition geometry permits only a projected estimate. September 6, March, and uu-only/gg-only epochs provide no sunward dust constraint. Sunward bars span strip-width and centering sensitivity, excluding unknown PSF corrections; no sharp turnaround edge is identified. Dashed lines show interpolation for the separate mass budgets. Diamonds give HST estimates ([63]); the black cross evaluates the discovery-paper width in SI units ([12]). The right axis uses β=0.574/aμm\beta= 0.574/a_{\mu\mathrm{m}} at 145 μm145\,\mu\mathrm{m}. Perpendicular and axial estimates constrain different velocity components; projected sunward values retain geometric dependence.

independently in time and integrate their product,

Md=∫t1t2M˙d(t) dt,(13)M_d = \int_{t_1}^{t_2} \dot{M}_d(t)\,\mathrm{d}t, \tag*{(13)}

with time in seconds. Photometry is not extrapolated; velocity coefficients are held at the nearest endpoint outside their measured baseline, excluding HST as an anchor. Figure 12 shows rates and cumulative masses.

Dust mass-loss rates and cumulative dust mass

Figure 12. Dust mass-loss rates from red-band A(0)fρA(0)f\rho (top) and cumulative mass (bottom), for 1-µm–1-mm grains, aperture grain-number index −3-3, ρd=1000\rho_d = 1000 kg m−3^{-3}, and pr=0.10p_r = 0.10. Purple uses three artifact-screened transverse-width coefficients with an isotropic Gaussian velocity model; shading spans retained-profile sensitivity. Green uses nine sunward HWHM coefficients with a Sun-aligned turnaround model and phase-angle correction, excluding January’s nearly end-on geometry. Width coefficients are linearly interpolated and held constant beyond the endpoint epochs; activity is interpolated independently. Errors propagate photometric uncertainty only and may be smaller than symbols. The open star and whiskers show the midpoint and full MIRI thermal-rate range for a 1-mm cutoff ([7]). Gray shading marks the perihelion gap; dashed rates bridge gaps longer than 30 days. Dotted cumulative curves test an alternative gap interpolation of [A(0)fρrh2][A(0)f\rho r_h^2], divided by daily rh2r_h^2. This preserves the bounding observations while allowing a solar-flux-shaped enhancement, without bounding the total mass. The budgets remain conditional on the two velocity models.

The transverse model gives Md=2.71×109M_d = 2.71 \times10^9 kg, rates of 6.2–360 kg s−1^{-1}, and effective speeds ⟨av⟩/aeff=0.91\langle av\rangle/a_{\mathrm{eff}} = 0.91–1.23 m s−1^{-1}. Substituting profile extrema gives (2.42(2.42–2.91)×1092.91) \times10^9 kg. Holding the February or August coefficient across the 161-day velocity gap gives 2.40 or 3.07×1093.07 \times10^9 kg. The nine axial coefficients give 1.05×10101.05 \times10^{10} kg. Directional and PSF assumptions dominate profile-repeatability errors.

Across the 72.7-day perihelion gap, interpolating [A(0)fρ]rh2[A(0)f\rho]r_h^2 and dividing by daily rh2r_h^2 preserves the endpoints while allowing a solar-flux-shaped enhancement. It gives 3.16×1093.16 \times10^9 and 1.23×10101.23 \times10^{10} kg for the transverse and sunward models; an unobserved outburst remains unconstrained. Transverse aperture crossing times of 94–127 days weaken the steady-flow approximation. If a r−3r^{-3} describes source production instead of the aperture distribution, residence-time weighting raises the estimates by 1.84.

The cross-section derivative gives net aperture dust accumulation: Map=(4/3)ρdaeffCrM_{\mathrm{ap}} = (4/3)\rho_d a_{\mathrm{eff}} C_r, with CrC_r in m2^2 and aeff=145a_{\mathrm{eff}} = 145 μm. The inbound slope gives dMap/dt=18.2±0.6\mathrm{d}M_{\mathrm{ap}}/\mathrm{d}t = 18.2 \pm0.6 kg s−1^{-1} and 1.58×1081.58 \times10^8 kg accumulated from the fitted zero to September 14. The seven-day maximum contains 6.08×1086.08 \times10^8 kg. Continued grain escape makes total emission exceed the retained inventory.

Extending the seven-day profile to its zeros on 2025 June 6 and 2026 June 1 gives 2.79×1092.79 \times10^9 and 1.08×10101.08 \times10^{10} kg, equivalent to 0.26- and 1.02-m shells. The same smoothed profile over July–May gives 2.76×1092.76 \times10^9 and 1.07×10101.07 \times10^{10} kg. Smoothing and perihelion interpolation slightly raise these values over the principal unsmoothed budgets; extending the endpoints adds little. Earlier activity and later cessation remain unconstrained.

We express the dust budget as an equivalent globally removed shell,

ΔR=R0−(R03−3Md4πρn)1/3,(14)\Delta R = R_0 - \left(R_0^3 - \frac{3M_d}{4\pi\rho_n}\right)^{1/3}, \tag*{(14)}

Here R0=1.3±0.2R_0 = 1.3 \pm0.2 km is the HST post-perihelion radius for nucleus albedo pV=0.04p_V = 0.04 ([58]), and ρn=500\rho_n = 500 kg m−3^{-3} is assumed nucleus density. Transverse and axial models give global dust-equivalent depths of 0.26 and 0.99 m, excluding volatile loss. A smaller active area implies deeper local erosion.

The larger MIRI estimates ([7]) use different velocity, grain-size, and activity-history assumptions. Their size-independent 50 m s−1^{-1} exceeds our effective speeds of 0.91–1.23 m s−1^{-1} (transverse) and 3.26–4.89 m s−1^{-1} (axial), raising conversion factors by 41–55 and 10–15 at fixed grain distribution. On December 15, their 1-mm-cutoff rates of 1820–2770 kg s−1^{-1} exceed our interpolated 341 and 1230 kg s−1^{-1} by factors of 5.3–8.1 and 1.5–2.3.

For comparison, adopting 50 m s−1^{-1} with our grain distribution, density, albedo, and piecewise-linear activity gives 1.26×10111.26 \times10^{11} kg over July 3–May 15 and a 12.02-m shell. Our principal budgets use image-based speeds. The common-speed model gives 1.68×1041.68 \times10^4 kg s−1^{-1} on December 15, still 6.1–9.2 times the MIRI 1-mm-cutoff rates. MIRI measures thermal ϵfρ\epsilon f\rho within approximately 3700 km, whereas our optical aperture is 10,000 km. Emissivity, albedo, temperature, and radial grain distribution also affect the comparison ([7]).

Their December-normalized rh−2r_h^{-2} integration spans June 1–December 27, including perihelion; ours interpolates July–May. Their 2.5–56.5-m shells span radii of 1.1–1.5 km, temperature assumptions, and 1-mm–1-cm upper grain cutoffs; the centimeter extension raises constant-speed mass conversion by 7.5 ([7]). Equation (14) gives 0.255 and 0.989 m for our nominal masses, confirming that mass conversion drives the discrepancy. Both estimates omit volatile loss and local recession. Projection, PSF broadening, anisotropy, and residence times limit our absolute mass estimates.

Optical spectra and molecular and atomic production

Figures 13 and 14 show six GMOS epochs and the September 1 LRIS spectra, excluding March. September and November GMOS fluxes use MPC normalizations. We display the noisy February 17 spectrum only at λ≥400\lambda\ge400 nm, but retain the complete extraction for gas measurements. Its CN limit excludes the unknown blue-response systematic.

Visible reflectance spectra in date order for six GMOS epochs and the September 1 LRIS spectrum

Figure 13. Visible reflectance spectra in date order, combining both GMOS central-wavelength settings. Six GMOS epochs and September 1 LRIS are shown; March is excluded. LRIS uses synthetic-solar division and the red-arm continuity correction. Purple/orange curves show coma/coma–tail with display offsets of −0.10/+0.10; gray shows the unshifted, correlated all-exposure spectrum. Reference lines mark CN, C3_3, C2_2, and Ni I. The noisy February 17 region below 400 nm is omitted, retaining C3_3.

We derive CN and C2_2 production with a forward Haser model. The steady, spherical daughter density is projected along the line of sight and convolved with a normalized Gaussian PSF PP of assumed FWHM 0.8 arcsec. For GMOS, the extraction operator E\mathcal{E} integrates the source rectangle and subtracts the linear sky fit; combined spectra use the 550-nm geometry and equally weighted nod positions. For LRIS, it integrates a centered 0.7×9.450.7 \times9.45-arcsec rectangle

Figure 14. Visible flux spectra in date order for six GMOS epochs and LRIS. August uses a measured standard response and imaging; December–February use model responses anchored to imaging. September/November use MPC aperture photometry transferred through GMOS image profiles. LRIS uses G191B2B and a single red-arm continuity factor. Translucent LRIS coma and orange coma–tail curves have offsets of −0.25 and +0.25 in plotted units (10−1610^{-16} erg cm−2^{-2} s−1^{-1} Å−1^{-1}); gas measurements use unshifted spectra. The noisy February 17 region below 400 nm is omitted, retaining C3_3.

without a sky correction because the sky footprint is unavailable. The effective residence time and production rate are

τj,eff=E[(ΣjQtrial)∗P],Qj=4πΔcm2Fjgjτj,eff(15)\tau_{j,\mathrm{eff}}=\mathcal{E}\left[\left(\frac{\Sigma_j}{Q_{\mathrm{trial}}}\right)*P\right],\qquad Q_j=\frac{4\pi\Delta_{\mathrm{cm}}^2F_j}{g_j\tau_{j,\mathrm{eff}}} \tag*{(15)}

where Σj\Sigma_j is column density, FjF_j continuum-subtracted band flux, gjg_j fluorescence efficiency, and Δcm\Delta_{\mathrm{cm}} observer distance in cm. Physical-area integration gives τj,eff\tau_{j,\mathrm{eff}} in seconds, independent of the trial rate QtrialQ_{\mathrm{trial}}. We adopt v=0.85(rh/1 au)−1/2 km s−1v=0.85(r_h/1\,\mathrm{au})^{-1/2}\,\mathrm{km}\,\mathrm{s}^{-1}. Independent rectangular integration agrees with sbpy (M. Mommert et al. [82]) for matched parameters.

CN and C2_2 parent/daughter lengths at 1 au are (1.3×104,2.1×105)(1.3\times10^4,2.1\times10^5) and (2.2×104,6.6×104)(2.2\times10^4,6.6\times10^4) km (M. F. A’Hearn et al. [1]), scaled by rh2r_h^2. CN fluorescence is interpolated in rhr_h and heliocentric radial velocity using the Lowell implementation of D. G. Schleicher [108]; C2_2 uses 4.5×10−13rh−24.5\times10^{-13}r_h^{-2} erg s−1^{-1} molecule−1^{-1}. C3_3 follows the December ATel prescription: g3=3.8×10−13rh−2g_3=3.8\times10^{-13}r_h^{-2} erg s−1^{-1} molecule−1^{-1}, consistent with A. L. Cochran et al. [33], parent/daughter lengths (3100,145000)rh2(3100,145000)r_h^2 km, and v3=0.5v_3=0.5 km s−1^{-1}. We integrate its Haser column over the source rectangle with sbpy, without PSF or sky corrections, using each epoch’s Horizons geometry (Table 3).

UTC dateInstrument/extractionrhr_hΔ\Deltar˙h\dot{r}_hgCN/10−13g_{\mathrm{CN}}/10^{-13}vCN,C2v_{\mathrm{CN,C2}}w×hw\times h
2025-08-28GMOS2.6362.589-52.890.4410.5241.0×6.4001.0\times6.400
2025-09-01LRIS coma2.5152.575-52.050.5060.5360.7×9.4500.7\times9.450
2025-09-01LRIS coma-tail2.5152.575-52.050.5060.5360.7×9.4500.7\times9.450
2025-09-14/15GMOS2.1102.542-47.670.8320.5851.0×6.4001.0\times6.400
2025-11-26GMOS1.7011.96237.861.3100.6521.0×6.4561.0\times6.456
2025-12-22GMOS2.3831.80250.920.6070.5511.0×6.4561.0\times6.456
2026-01-21GMOS3.3092.32655.800.4200.4671.0×6.4561.0\times6.456
2026-02-17GMOS4.1973.39957.420.2710.4151.0×6.4561.0\times6.456

Table 3. Epoch-specific geometry and gas-conversion inputs.

Distances are in au, r˙h\dot{r}_h and speeds in km s−1^{-1}, dimensions in arcsec, and gCNg_{\mathrm{CN}} in erg s−1^{-1} molecule−1^{-1}. CN/C2_2 use v=0.85rh−1/2v=0.85r_h^{-1/2}, a 0.8-arcsec Gaussian PSF, GMOS sky transfer, and one-au parent/daughter lengths (13000,210000)(13000,210000) and (22000,66000)(22000,66000) km. C3_3 uses v=0.5v=0.5 km s−1^{-1}, a centered rectangle, g=3.8×10−13rh−2g=3.8\times10^{-13}r_h^{-2}, and lengths (3100,145000)rh2(3100,145000)r_h^2 km, without PSF/sky correction. LRIS assumes a 0.7-arcsec slit and 9.45-arcsec extraction height for unbinned pixels, with geometry at 2025 September 1 00:00 UTC. December uses rh=2.38269281r_h=2.38269281 au (2.383 au here); the ATel example used 2.368 au.

We integrate CN and C2_2 solar-continuum residuals in 0.5-nm bins over 383.0–390.5 and 486–518.5 nm. A broadened solar spectrum with a local linear color term is fitted over 380–382.5 and 390.8–393 nm for CN, and 475–486 and 519–527 nm for C2_2. Errors include rebinning covariance, continuum scatter, and the shared flux scale. C3_3 uses

F3=[⟨Fλ⟩397.5<λ<415−median⁡(Fλ|379<λ<384 or 393<λ<397)](175 A˚),(16)F_3=\left[\langle F_\lambda\rangle_{397.5<\lambda<415}-\operatorname{median}\left(F_\lambda\middle|_{379<\lambda<384\ \mathrm{or}\ 393<\lambda<397}\right)\right](175\,\text{\AA}), \tag*{(16)}

with wavelengths in nm and FλF_\lambda in erg cm−2^{-2} s−1^{-1} Å−1^{-1}, before solar division. We combine the 540- and 550-nm arrays with inverse-variance weights in 0.25-nm bins, preserving September/November’s native 0.5-nm information and covariance and omitting rejected bins. All epochs cover the full band; August covers 81% of the 379–384-nm continuum window, starting near 380 nm, and all of 393–397 nm. Extraction heights are 6.400 arcsec for GMOS-S and 6.456 arcsec for GMOS-N, with 1-arcsec slits; LRIS uses 0.7 × 9.45 arcsec.

We calibrate the August spectrum with the spectrophotometric standard EG 274 (M. Hamuy et al. [51, 50]) and imaging. The December–February spectra use the reduced PHOENIX-based response (T.-O. Husser et al. [59]), normalized to imaging. September 14/15 and November 26 use MPC SrS_r photometry: station J47 reports median r=14.70r=14.70 within 6.2 arcsec on September 14, and station 213 reports r=12.60r=12.60 within 6.5 arcsec on November 26. GMOS image flux ratios transfer these circular-aperture magnitudes to the spectroscopic rectangles, giving slit-equivalent magnitudes 16.200 and 14.620. We normalize the solar-restored spectra through the PS1 rr response, with report scatter and a 0.20-mag calibration floor. The MPC observations precede the spectra by approximately 4 and 10 hr. Published gas rates are excluded from calibration, and the blue response remains uncertain.

We propagate C3_3 noise through Equation (16) in 4096 realizations, including combination weights and September–November rebin covariance. The imaging or MPC normalization enters once as a correlated gray-scale term.

Residual reduction covariance and the LRIS absolute-scale error are incompletely known. Formal errors exclude continuum choice, response shape, fluorescence, source geometry, and sky-subtraction systematics. We classify excesses above 3σ3\sigma as formal detections; otherwise we report max⁡(0,Q^)+3σQ\max(0,\widehat{Q})+3\sigma_Q, retaining signed residuals electronically. Dust color and solar absorption affect C3_3 attribution, while local continuum fits can absorb molecular wings. We test this with the broadened CALSPEC Kurucz spectrum sun_mod_001.fits ([11]) and a linear color term. Extended/red-only fits define the December/January ranges; the February limit spans all tested continua. These tests cover four imaging-scaled and two MPC-scaled GMOS epochs, both LRIS orientations, and their correlated all-exposure combination. February’s 379–397-nm continuum retains the blue-response limitation.

For Ni I, we jointly fit 347–356 and 358–366 nm with a common template amplitude, solar continuum, and transitions from the corrected multi-level fluorescence model ([28, 27]), evaluated at the LRIS geometry and radial velocity. Following D. Hutsemekers et al. [60], we adopt a short source, long-lived atoms, and projected column ∝1/ρ\propto1/\rho, with speed 0.5 km s−1^{-1} in the 0.7 × 9.45-arcsec rectangle. For rectangular half-sizes aa, bb, the residence time is [aasinh(b/a)+basinh(a/b)]/v[a\mathop{\rm asinh}(b/a)+b\mathop{\rm asinh}(a/b)]/v, without a PSF/sky correction. Gaussian FWHMs of 0.50, 0.65, and 0.80 nm test broadening. Blends and the unknown source scale limit interpretation. Table 4 reports this joint fit without assigning a single-band flux.

UTC dateInstrument/extractiont−Tpt-T_p (d)SpeciesFF (10−1410^{-14} erg cm−2^{-2} s−1^{-1})QQ (s−1^{-1})
2025-08-28GMOS−62.5CN1.287±0.0471.287 \pm0.047(1.85±0.07)×1025(1.85 \pm0.07) \times10^{25}
2025-08-28GMOS−62.5C3_31.868±0.1111.868 \pm0.111(6.04±0.36)×1024(6.04 \pm0.36) \times10^{24}
2025-08-28GMOS−62.5C2_2<0.244<0.244<3.93×1024<3.93 \times10^{24}
2025-09-01LRIS coma−58.5CN0.597±0.0190.597 \pm0.019(6.51±0.21)×1024(6.51 \pm0.21) \times10^{24}
2025-09-01LRIS coma−58.5C3_31.560±0.0431.560 \pm0.043(4.64±0.13)×1024(4.64 \pm0.13) \times10^{24}
2025-09-01LRIS coma−58.5C2_2<0.102<0.102<1.28×1024<1.28 \times10^{24}
2025-09-01LRIS coma−58.5Ni I…n\ldots^{n}(2.18±0.04)×1023(2.18 \pm0.04) \times10^{23}
2025-09-01LRIS coma-tail−58.5CN0.730±0.0310.730 \pm0.031(7.96±0.33)×1024(7.96 \pm0.33) \times10^{24}
2025-09-01LRIS coma-tail−58.5C3_32.201±0.0422.201 \pm0.042(6.55±0.13)×1024(6.55 \pm0.13) \times10^{24}
2025-09-01LRIS coma-tail−58.5C2_2<0.149<0.149<1.88×1024<1.88 \times10^{24}
2025-09-01LRIS coma-tail−58.5Ni I…n\ldots^{n}(2.48±0.04)×1023(2.48 \pm0.04) \times10^{23}
2025-09-14/15GMOS−44.5CN6.320±1.2176.320 \pm1.217(3.65±0.70)×1025(3.65 \pm0.70) \times10^{25}
2025-09-14/15GMOS−44.5C3_35.100±0.9855.100 \pm0.985(7.79±1.50)×1024(7.79 \pm1.50) \times10^{24}
2025-09-14/15GMOS−44.5C2_21.213±0.2411.213 \pm0.241(9.37±1.86)×1024(9.37 \pm1.86) \times10^{24}
2025-11-26GMOS28.1CN14.076±3.24514.076 \pm3.245(3.81±0.88)×1025(3.81 \pm0.88) \times10^{25}
2025-11-26GMOS28.1C3_316.554±3.18816.554 \pm3.188(1.13±0.22)×1025(1.13 \pm0.22) \times10^{25}
2025-11-26GMOS28.1C2_24.992±1.0534.992 \pm1.053(1.83±0.39)×1025(1.83 \pm0.39) \times10^{25}
2025-12-22GMOS54.0CN1.554±0.1791.554 \pm0.179(1.36±0.16)×1025(1.36 \pm0.16) \times10^{25}
2025-12-22GMOS54.0C3_32.8032.803–3.2773.277(5.73(5.73–6.70)×10246.70) \times10^{24}
2025-12-22GMOS54.0C2_2<0.977<0.977<1.08×1025<1.08 \times10^{25}
2026-01-21GMOS83.9CN<0.222<0.222<4.48×1024<4.48 \times10^{24}
2026-01-21GMOS83.9C3_30.5890.589–0.6510.651(4.00(4.00–4.42)×10244.42) \times10^{24}
2026-01-21GMOS83.9C2_2<0.247<0.247<8.47×1024<8.47 \times10^{24}
2026-02-17GMOS111.0CN<0.175<0.175<9.50×1024<9.50 \times10^{24}
2026-02-17GMOS111.0C3_3<0.333<0.333<5.71×1024<5.71 \times10^{24}
2026-02-17GMOS111.0C2_2<0.063<0.063<6.15×1024<6.15 \times10^{24}

Table 4. Integrated gas fluxes and production rates.

Pre-December C3_3 fluxes use the ATel estimator: mean flux density at 397.5–415 nm minus the pooled median at 379–384 and 393–397 nm, multiplied by 175 Å. C3_3 rates use g=3.8×10−13rh−2g=3.8 \times10^{-13}r_h^{-2} erg s−1^{-1} molecule−1^{-1}, parent/daughter lengths (3100,145000)rh2(3100,145000)r_h^2 km, v=0.5v=0.5 km s−1^{-1}, and the extraction rectangle. CN/C2_2 use the forward PSF/sky model with v=0.85rh−1/2v=0.85r_h^{-1/2} km s−1^{-1}. Rectangles are 1 × 6.400 arcsec for GMOS-S, 1 × 6.456 arcsec for GMOS-N, and 0.7 × 9.45 arcsec for LRIS. September/November GMOS scales use MPC photometry; December–February use model responses. Errors include noise and quantified gray-scale uncertainty, excluding C3_3 continuum and chromatic-response systematics. LRIS absolute-scale error is unquantified. December/January C3_3 ranges span extended/red-only solar-continuum results without statistical coverage. The February limit is the maximum max⁡(0,Fj)+3σj\max(0,F_j)+3\sigma_j over three continua, including noise and gray-scale error. Other limits use max⁡(0,Q^)+3σQ\max(0,\widehat{Q})+3\sigma_Q or its flux equivalent. Late C3_3 remains sensitive to continuum/response calibration; the estimator integrates the complex without separating every transition. n^{n}Ni I uses a common-amplitude multi-line fit and fixed-speed atomic model. Electronic tables retain signed residuals and the correlated LRIS all-exposure C3_3 estimate.

CN production is (1.85±0.07)×1025(1.85 \pm0.07) \times10^{25}, (3.65±0.70)×1025(3.65 \pm0.70) \times10^{25}, (3.81±0.88)×1025(3.81 \pm0.88) \times10^{25}, and (1.36±0.16)×1025(1.36 \pm0.16) \times10^{25} s−1^{-1} in August, September, November, and December. September and November also show C2_2. On November 26, CN and C2_2 have signal-to-noise ratios 7.4 and 10.2 before the common scale error. The C3_3 excess, (1.655±0.319)×10−13(1.655 \pm0.319) \times10^{-13} erg cm−2^{-2} s−1^{-1} including that error, gives Q(C3)=(1.13±0.22)×1025 s−1Q(\mathrm{C}_3)=(1.13\pm0.22)\times10^{25}\ \mathrm{s}^{-1}. MPC scatter, aperture transfer, and the 0.20-mag floor give an 18.7% normalization uncertainty. December’s weaker 516.5-nm bandhead and integrated C2\mathrm{C}_2 residual consistent with zero yield an upper limit.

Keck/KCWI detected CN, C2\mathrm{C}_2, C3\mathrm{C}_3, Ni, and Fe on November 16 (W. B. Hoogendam et al. [57]); K. Medler et al. [79] detected CN and Ni before and after perihelion and weak outbound C3\mathrm{C}_3. These identifications support November carbon-chain emission, although rate comparisons depend on epoch, extraction, calibration, and source model.

The LRIS coma spectrum gives Q(CN)=(6.51±0.21)×1024Q(\mathrm{CN})=(6.51\pm0.21)\times10^{24}, Q(C3)=(4.64±0.13)×1024Q(\mathrm{C}_3)=(4.64\pm0.13)\times10^{24}, Q(C2)<1.28×1024 s−1Q(\mathrm{C}_2)<1.28\times10^{24}\ \mathrm{s}^{-1}, and Q(Ni I)=(2.18±0.04)×1023 atoms s−1Q(\mathrm{Ni\,I})=(2.18\pm0.04)\times10^{23}\ \mathrm{atoms\ s}^{-1}. Coma–tail values are (7.96±0.33)×1024(7.96\pm0.33)\times10^{24}, (6.55±0.13)×1024(6.55\pm0.13)\times10^{24}, <1.88×1024 s−1<1.88\times10^{24}\ \mathrm{s}^{-1}, and (2.48±0.04)×1023 atoms s−1(2.48\pm0.04)\times10^{23}\ \mathrm{atoms\ s}^{-1}, respectively. Both use the 0.7-arcsec slit and the species-specific models above. Errors exclude the unknown LRIS absolute scale, blends, and source geometry.

The December ATel CN rate, 6.49×1026 s−16.49\times10^{26}\ \mathrm{s}^{-1} (B. T. Bolin et al. [25]), contains an unresolved conversion inconsistency. Its written gCN(1 au)=3.7×10−13g_{\mathrm{CN}}(1\ \mathrm{au})=3.7\times10^{-13} applied to the original Jy arrays gives 1.47×1025 s−11.47\times10^{25}\ \mathrm{s}^{-1}, close to our rate. The saved output corresponds instead to gCN(1 au)=8.4×10−15g_{\mathrm{CN}}(1\ \mathrm{au})=8.4\times10^{-15} with other inputs fixed. Figure 16 retains the published value; the discrepancy has no secure physical interpretation. November ATel CN and C2\mathrm{C}_2 rates exceed ours by factors of 18.9 and 23.3 (B. T. Bolin et al. [24]).

Figure 16. Gas-production rates compared with ATels and G. Ahuja et al. [3]. December/January C3_3 brackets span solar-continuum choices, (5.73(5.73–6.70)×10246.70) \times10^{24} and (4.00(4.00–4.42)×10244.42) \times10^{24} s−1^{-1}, without statistical coverage. The February limit is <5.71×1024< 5.71 \times10^{24} s−1^{-1} across three continua (Figure 15). Our C3_3 rates use the extraction-matched Haser model with v=0.5v = 0.5 km s−1^{-1}; pre-December fluxes use the blue-baseline estimator. CN/C2_2 use the forward PSF/sky model and Ni the atomic model. Purple marks LRIS; open diamonds identify coma–tail. Literature values retain their conventions. Our calibration is independent of those rates; ranges and limits exclude unquantified calibration/model systematics.

G. Ahuja et al. [3] report December 20 Q(CN)=(1.855±0.232)×1025 s−1Q(\mathrm{CN})=(1.855\pm0.232)\times10^{25}\ \mathrm{s}^{-1} at 1 km s−11\ \mathrm{km\ s}^{-1}. Scaling only speed to our December 22 value, 0.551 km s−10.551\ \mathrm{km\ s}^{-1}, gives (1.02±0.13)×1025 s−1(1.02\pm0.13)\times10^{25}\ \mathrm{s}^{-1}, a factor of 1.33 below our rate. Their unscaled November 27 CN and C2\mathrm{C}_2 rates are (5.54±0.051)×1025(5.54\pm0.051)\times10^{25} and (5.62±0.329)×1025 s−1(5.62\pm0.329)\times10^{25}\ \mathrm{s}^{-1}. Table 5 and Figure 16 retain published rates; epoch, aperture, and fluorescence differences remain. Our LRIS coma CN rate is close to the August 21 value 1024.81±0.01 s−110^{24.81\pm0.01}\ \mathrm{s}^{-1} (R. Rahatgaonkar et al. [97]), eleven days earlier and with different model assumptions.

UTC datet−Tpt-T_p (d)rhr_h (au)Specieslog⁡Q\log Qσlog⁡Q\sigma_{\log Q}Source
2025-08-21-69.02.84CN24.8100.010Rahatgaonkar et al. [97]
2025-08-21-69.02.84NiI23.3000.070Rahatgaonkar et al. [97]
2025-11-2628.01.70C2\mathrm{C}_226.6300.004ATel 17561
2025-11-2628.01.70C3\mathrm{C}_325.2550.005ATel 17561
2025-11-2628.01.70CN26.8590.013ATel 17561
2025-12-1547.02.18NiI23.8900.030Hutsemekers et al. [60]
2025-12-2153.02.35NiI23.7900.040Hutsemekers et al. [60]
2025-12-2254.02.38C2\mathrm{C}_226.3730.004ATel 17719
2025-12-2254.02.38C3\mathrm{C}_324.8440.009ATel 17719
2025-12-2254.02.38CN26.8120.013ATel 17719
2025-12-2658.02.50NiI23.6600.040Hutsemekers et al. [60]
2026-01-1174.02.99NiI23.2600.050Hutsemekers et al. [60]
2026-01-1982.03.25NiI23.0600.060Hutsemekers et al. [60]
2026-01-2184.03.31C2\mathrm{C}_224.7670.004ATel 17841
2026-01-2184.03.31C3\mathrm{C}_323.7630.005ATel 17841
2026-01-2184.03.31CN24.6570.005ATel 17841
2026-01-2790.03.51NiI22.8800.070Hutsemekers et al. [60]
2026-02-07101.03.87NiI22.4400.100Hutsemekers et al. [60]
2025-11-1618.01.51CN25.9070.021Ahuja et al. [3]
2025-11-1618.01.51C2\mathrm{C}_225.8450.015Ahuja et al. [3]
2025-11-1618.01.51C3\mathrm{C}_325.0290.080Ahuja et al. [3]
2025-11-1719.01.52CN25.9540.018Ahuja et al. [3]
2025-11-1719.01.52C2\mathrm{C}_225.8020.020Ahuja et al. [3]
2025-11-1719.01.52C3\mathrm{C}_325.0320.092Ahuja et al. [3]
2025-11-2729.01.72CN25.7440.004Ahuja et al. [3]
2025-11-2729.01.72C2\mathrm{C}_225.7500.025Ahuja et al. [3]
2025-11-2729.01.72C3\mathrm{C}_324.8320.049Ahuja et al. [3]

Table 5. Literature production rates used for comparison.

UTC datet−Tpt-T_{p} (d)rhr_{\mathrm{h}} (au)Specieslog⁡Q\log Qσlog⁡Q\sigma_{\log Q}Source
2025-11-2830.01.74CN25.7690.007Ahuja et al. 2026
2025-11-2830.01.74C225.7440.040Ahuja et al. 2026
2025-11-2830.01.74C324.7930.085Ahuja et al. 2026
2025-12-0436.01.89CN25.6140.021Ahuja et al. 2026
2025-12-0436.01.89C225.4050.091Ahuja et al. 2026
2025-12-0436.01.89C324.5120.190Ahuja et al. 2026
2025-12-0941.02.02CN25.5750.106Ahuja et al. 2026
2025-12-0941.02.02C225.3590.086Ahuja et al. 2026
2025-12-1244.02.10CN25.3480.027Ahuja et al. 2026
2025-12-1244.02.10C225.3130.062Ahuja et al. 2026
2025-12-1244.02.10C324.6380.240Ahuja et al. 2026
2025-12-1345.02.12CN25.3340.055Ahuja et al. 2026
2025-12-1345.02.12C225.3360.095Ahuja et al. 2026
2025-12-1345.02.12C324.7810.191Ahuja et al. 2026
2025-12-1749.02.24CN25.2320.079Ahuja et al. 2026
2025-12-1749.02.24C225.2290.249Ahuja et al. 2026
2025-12-1749.02.24C324.7000.265Ahuja et al. 2026
2025-12-2052.02.32CN25.2680.054Ahuja et al. 2026
2025-12-2052.02.32C225.0270.127Ahuja et al. 2026
2025-12-2052.02.32C324.4810.391Ahuja et al. 2026
2026-01-2083.03.28CN25.1300.189Ahuja et al. 2026
2026-01-2083.03.28C225.0040.393Ahuja et al. 2026

Table 5 (continued).

Rates retain the published fluorescence, outflow speed, scale lengths, apertures, and band coverage, in molecules s−1\mathrm{s}^{-1} (atoms s−1\mathrm{s}^{-1} for Ni). Geometry and t−Tpt-T_{p} use the common orbit and source date. The December 20 value is recalculated from that date, correcting the inconsistent TFP in G. Ahuja et al. [3], Table 3.

ATel-prescription C3 rates are (6.04±0.36)×1024 s−1(6.04\mathbin{\pm}0.36) \times10^{24}\ \mathrm{s}^{-1} on August 28 and (7.79±1.50)×1024 s−1(7.79\mathbin{\pm}1.50) \times10^{24}\ \mathrm{s}^{-1} on September 14/15. The correlated LRIS all-exposure value, (5.72±0.13)×1024 s−1(5.72 \mathbin{\pm} 0.13) \times10^{24}\ \mathrm{s}^{-1}, lies between the two orientations and is retained electronically. Their difference includes orientation, time, and calibration effects.

On December 22, the uncorrected estimator gives F3=(7.53±0.34)×10−14 erg cm−2s−1F_{3}=(7.53\mathbin{\pm}0.34)\times10^{-14}\ \mathrm{erg\ cm}^{-2}\mathrm{s}^{-1}, or (1.54±0.07)×1025 s−1(1.54\mathbin{\pm}0.07)\times10^{25}\ \mathrm{s}^{-1} for the 6.456 × 1-arcsec extraction. The ATel used a different spectrum and a 130-arcsec rectangle, reproducing 6.98×1024 s−16.98\times10^{24}\ \mathrm{s}^{-1}. The spectra differ in continuum shape and by a factor of 2.34 in band excess, motivating the solar-continuum tests below.

Our November C3 rate is a factor of 1.59 below the ATel value 1.80×1025 s−11.80\times10^{25}\ \mathrm{s}^{-1} [24] and 1.66 times above the November 27 value (6.79±0.77)×1024 s−1(6.79\mathbin{\pm}0.77)\times10^{24}\ \mathrm{s}^{-1} [3]; scaling their speed from 1 to 0.5 km s−1\mathrm{s}^{-1} increases the latter ratio to 3.33. January 21’s uncorrected estimator gives (1.67±0.09)×1025 s−1(1.67\mathbin{\pm}0.09)\times10^{25}\ \mathrm{s}^{-1}, 28.8 times the ATel 17841 value 5.79×1023 s−15.79\times10^{23}\ \mathrm{s}^{-1}. December–February use model-response calibrations, with January anchored to imaging approximately 22 hr earlier. C3 response-shape envelopes are approximately 19%, 15%, and 15%, respectively; continuum placement and chromatic calibration limit these comparisons.

Post-perihelion C3 continuum and aperture controls.—We apply Equation (16) to gas-free solar continua to estimate contamination of the C3 excess. A template convolved to 0.65 nm FWHM is fitted in amplitude and linear color to each GMOS setting. Anchors are 391–394 and 415–420 nm (local), 380–382, 440–445, and 475–480 nm (extended), or 440–445 and 475–480 nm (red only). Models use the data’s native-bin combination weights. Anchors were chosen before comparing published rates: local fits risk absorbing molecular wings, whereas distant anchors extrapolate continuum color.

Gas-free controls give apparent C3 fluxes of (6.18(6.18–6.30)×10−14 erg cm−2s−16.30)\times10^{-14}\ \mathrm{erg\ cm}^{-2}\mathrm{s}^{-1} in December and (2.11(2.11–2.30)×10−14 erg cm−2s−12.30)\times10^{-14}\ \mathrm{erg\ cm}^{-2}\mathrm{s}^{-1} in January, accounting for 82–84% and 86–94% of the estimator signals. Subtracting the extended/red-only solar fits gives 397.5–415-nm residual integrals of (2.80(2.80–3.28)×10−143.28)\times10^{-14} and (5.89(5.89–6.51)×10−15 erg cm−2s−16.51)\times10^{-15}\ \mathrm{erg\ cm}^{-2}\mathrm{s}^{-1}. The 6.456×16.456 \times1-arcsec extraction, epoch geometry, and fixed-speed Haser model then give Q(C3)=(5.73Q(\mathrm{C}_3)=(5.73–6.70)×10246.70)\times10^{24} molecules s−1\mathrm{s}^{-1} on 2025 December 22 and (4.00(4.00–4.42)×10244.42)\times10^{24} molecules s−1\mathrm{s}^{-1} on 2026 January 21. December approaches the ATel value 6.98×10246.98\times10^{24} molecules s−1\mathrm{s}^{-1}; January remains 6.96.9–7.67.6 times higher. The adopted ranges describe continuum-choice sensitivity, with additional statistical and calibration uncertainty. Local fits give much smaller residuals, especially in January, and reduced χ2≃3\chi^2\simeq3–88 indicates mismatch beyond statistical noise.

For fixed flux, the actual aperture increases QQ by factors of 5.08, 5.94, and 6.15 relative to a 130×1130\times1-arcsec rectangle in December, January, and February. Increasing extraction height from 3.228 to 12.912 arcsec changes the aperture-corrected estimator rate from 2.03 to 1.18×1025 s−11.18\times10^{25}\ \mathrm{s}^{-1} in December and 2.51 to 1.01×1025 s−11.01\times10^{25}\ \mathrm{s}^{-1} in January. This exceeds formal errors and may reflect dust contamination, seeing, sky subtraction, or chromatic calibration. At the adopted height, alternative sky regions change the estimates by 0.3% and 2.0%; central-wavelength settings differ by factors of 1.12 and 1.27. These effects explain only part of January’s discrepancy.

February 17’s residual changes sign with the continuum anchors. We adopt the envelope F3,lim⁡=max⁡j[max⁡(0,F3,j)+3σj]F_{3,\lim}=\max_j[\max(0,F_{3,j})+3\sigma_j] over the local, extended, and red-only fits, where σj2=σstat,j2+(fgrayF3,j)2\sigma_j^2=\sigma_{\mathrm{stat},j}^2+(f_{\mathrm{gray}}F_{3,j})^2 includes continuum-fit and integration noise and shared gray-scale fraction 0.030794. The red-only fit sets F3,lim⁡=3.33×10−15 erg cm−2s−1F_{3,\lim}=3.33\times10^{-15}\ \mathrm{erg\ cm}^{-2}\mathrm{s}^{-1} and Q(C3)<5.71×1024Q(\mathrm{C}_3)<5.71\times10^{24} molecules s−1\mathrm{s}^{-1}. This 3σ3\sigma envelope covers the tested continua but excludes unknown response-shape and physical-model errors. Figure 15 and Figure 16 and Table 4 use this limit and the December/January ranges; electronic tables retain uncorrected estimates.

Figure 15. Molecular fluxes (left) and production rates (right). C3_3 uses an extraction-matched rectangular Haser model with v=0.5v=0.5 km s−1^{-1}; pre-December fluxes follow Equation (16). December/January brackets span the extended/red-only solar-continuum results, (5.73(5.73–6.70)×10246.70) \times10^{24} and (4.00(4.00–4.42)×10244.42) \times10^{24} s−1^{-1}, and represent continuum-choice ranges with no measured midpoint or statistical coverage. The February limit, Q<5.71×1024Q < 5.71 \times10^{24} s−1^{-1}, is the largest residual plus 3σ3\sigma over three continua. It includes noise and gray-scale error, excluding unknown chromatic-response and model errors. CN/C2_2 use the forward PSF/sky model. GMOS is blue and LRIS purple; squares show LRIS coma and open diamonds coma–tail. Open blue diamonds identify MPC-scaled spectra. Thick/thin bars on other points show statistical/gray-scale-inclusive errors; upper-limit arrows follow instrument colors. Signed fluxes are retained where applicable. March is excluded.

G. S. H. Paek et al. [90] resolved an extended inbound CN coma at rh<3r_h<3 au and increasing Q(CN)Q(\mathrm{CN}) relative to AfρAf\rho, supporting a gas contribution to blue coma growth. Their CN-sensitive 400-nm band samples different emission from C2\mathrm{C}_2 in our broadband gg images.

In July 2025, JCMT gave 3σ3\sigma limits of Q(HCN)<1.7×1024 s−1Q(\mathrm{HCN})<1.7\times10^{24}\ \mathrm{s}^{-1} at rh=4.01r_h=4.01–3.973.97 au and Q(CO)<1.1×1027 s−1Q(\mathrm{CO})<1.1\times10^{27}\ \mathrm{s}^{-1} at 3.943.94–3.843.84 au (J. T. Hinkle et al. [54]). Including later observations gave Q(HCN)∝rhnQ(\mathrm{HCN})\propto r_h^n with n=12.7−2.5+6.9n=12.7^{+6.9}_{-2.5} describing parent-molecule activation before our daughter-species detections. The epochs preclude a simultaneous HCN/CN ratio, and this gas-production index has a different physical meaning from the dust Afρ(0)Af\rho(0) index.

Gas-production ratios

Table 6 and Figure 17 compare C3\mathrm{C}_3, CN, C2\mathrm{C}_2, and Ni. The common gray scale cancels in same-spectrum ratios. Without cross-band statistical covariance for C3\mathrm{C}_3, we remove that scale term and report the maximum first-order standard deviation allowed by the marginal errors: σx/y/∣y∣+∣x∣σy/y2\sigma_{x/y}/|y|+|x|\sigma_y/y^2 for x/yx/y. Continuum-choice and response-shape errors are excluded; December’s continuum range is separate. C2\mathrm{C}_2/CN and Ni I/CN propagate band errors without the shared scale. We divide upper-limit numerators by detected denominators and omit ratios with undetected denominators. Ratios retain the distinct C3\mathrm{C}_3, CN/C2\mathrm{C}_2, and atomic source models.

UTC dateInstrument/extractiont−Tpt - T_p (d)RatioValue
2025-08-28GMOS-62.5C2/CNC_2/\mathrm{CN}<0.213< 0.213
2025-08-28GMOS-62.5C3/CNC_3/\mathrm{CN}0.327±0.0240.327 \pm0.024
2025-08-28GMOS-62.5C3/C2C_3/C_2–
2025-09-01LRIS coma-58.5C2/CNC_2/\mathrm{CN}<0.198< 0.198
2025-09-01LRIS coma-58.5C3/CNC_3/\mathrm{CN}0.713±0.0420.713 \pm0.042
2025-09-01LRIS coma-58.5C3/C2C_3/C_2–
2025-09-01LRIS coma-tail-58.5C2/CNC_2/\mathrm{CN}<0.239< 0.239
2025-09-01LRIS coma-tail-58.5C3/CNC_3/\mathrm{CN}0.823±0.0500.823 \pm0.050
2025-09-01LRIS coma-tail-58.5C3/C2C_3/C_2–
2025-09-14/15GMOS-44.5C2/CNC_2/\mathrm{CN}0.257±0.0140.257 \pm0.014
2025-09-14/15GMOS-44.5C3/CNC_3/\mathrm{CN}0.214±0.0090.214 \pm0.009
2025-09-14/15GMOS-44.5C3/C2C_3/C_20.832±0.0620.832 \pm0.062
2025-11-26GMOS28.1C2/CNC_2/\mathrm{CN}0.481±0.0800.481 \pm0.080
2025-11-26GMOS28.1C3/CNC_3/\mathrm{CN}0.296±0.0540.296 \pm0.054
2025-11-26GMOS28.1C3/C2C_3/C_20.616±0.0890.616 \pm0.089
2025-12-22GMOS54.0C2/CNC_2/\mathrm{CN}<0.840< 0.840
2025-12-22GMOS54.0C3/CNC_3/\mathrm{CN}0.4220.422–0.4930.493
2025-12-22GMOS54.0C3/C2C_3/C_2–
2026-01-21GMOS83.9C2/CNC_2/\mathrm{CN}–
2026-01-21GMOS83.9C3/CNC_3/\mathrm{CN}–
2026-01-21GMOS83.9C3/C2C_3/C_2–
2026-02-17GMOS111.0C2/CNC_2/\mathrm{CN}–
2026-02-17GMOS111.0C3/CNC_3/\mathrm{CN}–
2026-02-17GMOS111.0C3/C2C_3/C_2–
2025-09-01LRIS coma-58.5Ni I/CN\mathrm{Ni\,I}/\mathrm{CN}0.034±0.0010.034 \pm0.001
2025-09-01LRIS coma-tail-58.5Ni I/CN\mathrm{Ni\,I}/\mathrm{CN}0.031±0.0010.031 \pm0.001

Table 6. Gas-production ratios.

Figure 17. Gas-production ratios. The December C3C_3/CN bracket, 0.422–0.493, divides the two-continuum C3C_3 range by nominal CN; it excludes CN uncertainty and has no statistical coverage. Colors and symbols identify species pairs; purple edges mark LRIS and open centers coma–tail. Open comparison points retain published prescriptions. Shared gray normalization cancels. Other C3C_3 bars give the maximum first-order statistical error over unknown cross-band covariance, excluding continuum and chromatic-response systematics. C2C_2/CN and Ni I/CN errors propagate band uncertainties. Arrows mark upper limits; undetected denominators are omitted. Species-dependent source models and December–February C3C_3 continuum sensitivity limit abundance comparisons.

Q(C2)/Q(CN)Q(\mathrm{C}_2)/Q(\mathrm{CN}) rises from 0.257±0.0140.257\pm0.014 on 2025 September 14/15 to 0.481±0.0800.481\pm0.080 on November 26. LRIS Ni I/CN is 0.0336±0.00120.0336\pm0.0012 for coma and 0.0311±0.00140.0311\pm0.0014 for coma-tail. Its Ni rates, (2.18(2.18–2.48)×10232.48)\times10^{23} atoms s−1\mathrm{s}^{-1}, are comparable to the August 21 value 1023.30±0.0710^{23.30\pm0.07} (R. Rahatgaonkar et al. [97]), subject to epoch, fluorescence, and source-model differences. C3\mathrm{C}_3/CN is 0.327±0.0240.327\pm0.024, 0.214±0.0090.214\pm0.009, and 0.296±0.0540.296\pm0.054 in August, September, and November. December’s 0.4220.422–0.4930.493 range divides the two solar-continuum C3\mathrm{C}_3 estimates by nominal CN production, excluding CN uncertainty. LRIS gives 0.713±0.0420.713\pm0.042 for coma and 0.823±0.0500.823\pm0.050 for coma-tail. Continuum and source-model dependence limit compositional interpretation.

The January 21 ATel reports Q(CN)=(4.54±0.05)×1024Q(\mathrm{CN})=(4.54\pm0.05)\times10^{24}, Q(C2)=(5.85±0.05)×1024Q(\mathrm{C}_2)=(5.85\pm0.05)\times10^{24}, and Q(C3)=(5.79±0.06)×1023 s−1Q(\mathrm{C}_3)=(5.79\pm0.06)\times10^{23}\ \mathrm{s}^{-1} (B. T. Bolin et al. [26]). Our CN limit, 4.48×1024 s−14.48\times10^{24}\ \mathrm{s}^{-1}, is close to their value; our C2\mathrm{C}_2 limit, 8.47×1024 s−18.47\times10^{24}\ \mathrm{s}^{-1}, includes theirs. Published ATel ratios are plotted separately. The September ATel identifies emission without rates (B. T. Bolin et al. [23]).

On August 24, W. B. Hoogendam et al. [55] measured emission-profile ee-folding radii of 593.7±14.8593.7\pm14.8 km for Ni and 841.0±15.4841.0\pm15.4 km for CN; the Ni coma is more concentrated. Post-perihelion profiles are more extended (W. B. Hoogendam et al. [57]). These empirical radii describe the spatial dependence of Ni I/CN and its aperture correction; they are distinct from Haser survival lengths.

C3C_3 uses the ATel fixed-speed rectangular model; CN/C2C_2 use the forward PSF/sky model. Shared gray normalization cancels. C3C_3 ratio errors maximize the first-order standard deviation over unknown cross-band covariance: σx/∣y∣+∣x∣σy/y2\sigma_x/|y| + |x|\sigma_y/y^2 for x/yx/y, after removing the shared gray term. They exclude continuum and chromatic-response errors. C2/CNC_2/\mathrm{CN} and Ni I/CN\mathrm{Ni\,I}/\mathrm{CN} errors propagate band uncertainties; undetected denominators are omitted. Species-dependent models preclude a depletion classification. Gas-free continuum controls explain most of the December/January C3C_3 signal; February bounds the tested continua. The December C3/CNC_3/\mathrm{CN} range uses nominal CN, excludes its error, and has no statistical coverage.

Infrared results

MOSFIRE coma profiles and aperture measurements

Elliptical Gaussian fits within 3 arcsec give July MOSFIRE comet minor-axis FWHMs of 1.31, 1.27, and 1.79 arcsec in JJ, HH, KsK_s; geometric-mean widths are 1.42, 1.36, and 2.03 arcsec. Comparison sources fall within 1.5 arcsec of a 2MASS position, have A/B photometric quality without relevant artifact flags, and aperture fluxes within one magnitude of the comet. Their minor-axis FWHMs are 0.58–0.61 arcsec in JJ (two sources), 0.50 arcsec in HH (one), and 1.44 arcsec in KsK_s (one). Trailing and structured residuals limit their use as PSF estimates.

February minor-axis FWHMs are 1.87 and 1.79 arcsec in JJ and KsK_s, with geometric means of 1.93 and 1.94 arcsec. Two similarly bright JJ stars have widths of 1.15–1.42 arcsec, supporting an extended comet core. No KsK_s star passes the same selection. Tracking and background residuals affect these profile comparisons independently of the color calibration.

Additional February apertures of 1.590 and 0.430 arcsec provide instrumental-count checks. The smaller aperture matches the physical area of the GNIRS rectangle (0.68 × 2 arcsec at Δ=1.905\Delta= 1.905 au), with a circular radius of 909 km. The primary F2 extraction (1.43 × 14.4 arcsec at Δ=2.562\Delta= 2.562 au) instead requires an equal-area circular radius of approximately 4760 km, or 2.25 arcsec in February. The 0.430-arcsec aperture requires a cross-filter PSF correction within the approximately 1.8-arcsec core, and coma asymmetry changes the weighting between rectangles and circles. Figure 20 uses calibrated MOSFIRE radii of 1.4384–3.5960 arcsec, with aperture and epoch differences retained in the spectroscopic comparison.

Near-infrared spectra, colors, and variability

We restore the solar spectrum F⊙(λ)F_{\odot}(\lambda) to reflectance R(λ)R(\lambda) for synthetic photometry, assuming solar-equivalent continua for the comparison stars. The photon-weighted signal through transmission Tb(λ)T_b(\lambda) is

Ib∝∫R(λ)F⊙(λ)Tb(λ)λ dλ.(17)I_b \propto\int R(\lambda)F_{\odot}(\lambda)T_b(\lambda)\lambda\,\mathrm{d}\lambda. \tag*{(17)}

We integrate valid samples against the appropriate AB or Vega reference. Incomplete coverage makes broadband colors model-dependent.

The July sidereal transfer gives Y−J=0.466Y-J=0.466 mag at 2.16-arcsec radius. The three bridge-star estimates (0.413, 0.520, and 0.466 mag) have 0.053-mag sample scatter. Maximum shifts with aperture, background, and anchor spectrum are 0.082, 0.069, and 0.078 mag; their quadrature sum with the full bridge scatter gives the 0.143-mag sensitivity scale in Table 8 and Figure 20 and Figure 21. Correlations, spatial and transparency transfer, tracking-mode PSFs, contamination, and anchor mismatch remain incompletely quantified. A 25–35-pixel JJ-anchor background gives 0.501 mag. The nominal color is 0.227 mag redder than the native-filter solar value of 0.238 mag, without establishing a significant difference.

Figure 20. NIR imaging and partial-passband spectral colors by aperture. July MOSFIRE Y−JY-J uses a 2.16-arcsec radius and a 0.143-mag sidereal-transfer sensitivity bar without formal 1σ\sigma coverage. F2 extraction radii of 20/40/60/80 pixels span 3.6/7.2/10.8/14.4 arcsec at 0.18 arcsec pixel−1^{-1} along the slit. Symbols/colors distinguish GNIRS 1/2/3-arcsec/native extractions; filled squares mark primary choices. Nightly GNIRS colors combine block-relative fluxes; errors use the larger of propagated scatter and half the block color difference. F2 errors use local scatter. Extractions are correlated; horizontal offsets separate symbols. F2 cuts preclude H−KsH-K_s and J−KsJ-K_s. Gray dashed segments show solar colors over the primary spectral intervals. MOSFIRE JHKsJHK_s radii are 1.44/2.16 arcsec in July and 1.44/2.16/2.88/3.60 arcsec in February, with 2.16 arcsec nominal. February errors assume solar infrared reference-star spectra and exclude shared spectral mismatch. July open-symbol bars combine formal flux errors and the full inter-aperture difference in quadrature: 0.052, 0.150, and 0.202 mag for J−HJ-H, H−KsH-K_s, and J−KsJ-K_s. These sensitivities lack established Gaussian coverage and exclude shared contamination/calibration biases. Passband and physical-aperture differences limit temporal comparisons. Some errors are smaller than symbols.

SampleY−JY-JJ−HJ-HH−KsH-K_sJ−KsJ-K_s
3I MOSFIRE, July 11, 1.44″—0.187±0.0520.187 \pm0.0520.304±0.1500.304 \pm0.1500.491±0.2020.491 \pm0.202
3I MOSFIRE, July 11, 2.16″0.466±0.143∗0.466 \pm0.143^{*}0.239±0.0520.239 \pm0.0520.454±0.1500.454 \pm0.1500.693±0.2020.693 \pm0.202
3I MOSFIRE, February 6———0.303±0.0510.303 \pm0.051
Sun, native MOSFIRE synthetic0.2380.2380.3000.3000.0560.0560.3560.356
Sun, UKIRT [118]0.270.27———
3I F2, September 5/6—0.241±0.0050.241 \pm0.005——
3I GNIRS, December 1—0.270±0.0270.270 \pm0.027−0.050±0.020-0.050 \pm0.0200.219±0.0080.219 \pm0.008
3I GNIRS, December 2—0.277±0.0530.277 \pm0.053−0.067±0.026-0.067 \pm0.0260.210±0.0270.210 \pm0.027
Sun, full 2MASS synthetic—0.3320.3320.0430.0430.3740.374
MBOSS (2012): median ±\pm sample dispersion [number of objects]
SampleY−JY-JJ−HJ-HH−KH-KJ−KJ-K
Gray Centaurs—0.39±0.090.39 \pm0.09 [11]0.16±0.090.16 \pm0.09 [9]0.55±0.120.55 \pm0.12 [9]
Red Centaurs—0.39±0.060.39 \pm0.06 [6]−0.06±0.06-0.06 \pm0.06 [6]0.33±0.070.33 \pm0.07 [6]
TNOs—0.40±0.210.40 \pm0.21 [119]0.05±0.110.05 \pm0.11 [39]0.43±0.190.43 \pm0.19 [39]
Jupiter Trojans—0.45±0.060.45 \pm0.06 [12]0.13±0.040.13 \pm0.04 [12]0.59±0.090.59 \pm0.09 [12]
49P/Arend–Rigaux nucleus—0.360.36 [1]0.140.14 [1]0.500.50 [1]
Comet comae: weighted mean ±\pm published uncertainty [number of comets]
1986 coma sample—0.46±0.070.46 \pm0.07 [14]0.10±0.080.10 \pm0.08 [13]—
1988 coma sample—0.41±0.060.41 \pm0.06 [9]0.14±0.050.14 \pm0.05 [8]—
Individual active comet: published colors and errors
C/2005 E2 (McNaught)—0.48±0.010.48 \pm0.010.01±0.010.01 \pm0.010.49±0.010.49 \pm0.01
Other comet compilations: J−HJ-H median [reported observations]
Active short-period comets—0.420.42 [12]——
Inactive short-period comets—0.410.41 [2]——
Active long-period comets—0.480.48 [3]——
Inactive long-period comets—0.370.37 [1]——

Table 8. NIR colors and contextual population comparisons

Figure 21. Solar-relative NIR colors of 3I/ATLAS and comparison populations: (a) Δ(J−H)\Delta(J-H) versus Δ(H−K)\Delta(H-K) or Δ(H−Ks)\Delta(H-K_s); (b) Δ(Y−J)\Delta(Y-J) versus Δ(J−Ks)\Delta(J-K_s). The Sun is at the origin. Faint points show 1886 classified MOVIS asteroids [94, 93]; crosses mark spectral-template means. Open diamonds show July 11 MOSFIRE at 2.16-arcsec radius; open squares/triangles show December 1/2 GNIRS. Filled symbols in (a) show MBOSS gray/red Centaurs, TNOs, Jupiter Trojans, and the 49P nucleus [48], plus the 1986/1988 comet-coma samples [68]. Table 8 gives per-color counts; group means can use different objects in each color. Each dataset uses its adopted solar colors, including the MBOSS convention [49]. Historical KK, MOSFIRE KsK_s, partial-passband GNIRS, and VISTA colors retain response differences. July bars measure empirical calibration/aperture sensitivity without established Gaussian coverage; GNIRS bars measure scatter/repeatability. Marginal bars omit shared-band covariance. Stellar contamination and calibration uncertainty limit the July–December comparison. Only epochs with both plotted colors are shown.

In the same aperture, July colors are J−H=0.239±0.052J-H=0.239\pm0.052, H−Ks=0.454±0.150H-K_s=0.454\pm0.150, and J−Ks=0.693±0.202J-K_s=0.693\pm0.202 mag. Radii of 1.4384 and 2.1576 arcsec (8 and 12 pixels; 12 nominal) give colors of 0.187–0.239, 0.304–0.454, and 0.491–0.693 mag, respectively. At either radius, adopted errors combine the absolute aperture difference with the formal flux error in quadrature. The latter is only 0.003–0.004 mag, from σm=(2.5/ln⁡10)σF/F\sigma_m=(2.5/\ln10)\sigma_F/F and independent statistical errors between bands. The plotted empirical errors measure aperture sensitivity, with unestablished Gaussian 1σ1\sigma coverage and unconstrained common contamination and calibration bias. Adding single-band calibration sensitivities would require their covariance. Larger-aperture measurements remain in the electronic table. Stellar residuals and differing physical apertures limit a July–February color-evolution test.

Table 7 and Figure 20 give Vega colors over the accepted 2MASS passband intervals. F2 JH gives J−H=0.241±0.005J-H=0.241\pm0.005 mag from spectral scatter; the 0.198–0.274-mag range across correlated extractions dominates its uncertainty. The trimmed HK spectrum lacks H−KsH-K_s and J−KsJ-K_s coverage and was obtained on a different date, preventing direct broadband combination with JH.

SpectrumColorValue (mag)σscatter\sigma_{\mathrm{scatter}}Repeatability floorExtraction range
F2 Sep 5/6J-H0.2410.0050.198–0.274
GNIRS Dec 1H-Ks-0.0500.0030.020−0.115–−0.005
GNIRS Dec 1J-H0.2700.0020.0270.184–0.381
GNIRS Dec 1J-Ks0.2190.0020.0080.179–0.266
GNIRS Dec 2H-Ks-0.0670.0050.026−0.101–−0.027
GNIRS Dec 2J-H0.2770.0040.0530.189–0.360
GNIRS Dec 2J-Ks0.2100.0040.0270.162–0.265

Table 7. Synthetic NIR colors on accepted portions of the 2MASS passbands.

GNIRS partial colors (J−HJ-H, H−KsH-K_s, J−KsJ-K_s) are (0.270, −0.050-0.050, 0.219) mag on December 1 and (0.277, −0.067-0.067, 0.210) mag on December 2. Uncertainty scales including half the block separation are (0.027, 0.020, 0.008) and (0.053, 0.026, 0.027) mag. We combine relative band fluxes with common exposure weights, preserving J−H+H−Ks=J−KsJ-H+H-K_s=J-K_s. Coverage differs slightly between blocks and samples only part of each passband.

February MOSFIRE gives J−Ks=0.303±0.051J-K_s=0.303\pm0.051 mag and RKs/RJ=0.952±0.045R_{K_s}/R_J=0.952\pm0.045 at 2.1576-arcsec radius, consistent with a near-solar continuum. The error combines measurement, reference-star scatter, aperture, and sky terms under the assumed solar infrared reference spectra; shared spectral mismatch remains unquantified. The color changes from 0.305 to 0.280 mag over radii of 1.4384–3.5960 arcsec.

GNIRS J−HJ-H differs by 0.054 and 0.106 mag between blocks on December 1 and 2, exceeding the 0.007-mag difference between nightly means. Similar H−KsH-K_s offsets exceed local scatter errors. Calibration, chromatic slit losses, and extraction differences prevent resolving physical variability. Table 7 and Figure 20 compare the correlated extractions: F2 extraction radii of 20, 40, 60, and 80 pixels (3.6, 7.2, 10.8, and 14.4 arcsec at 0.18 arcsec pixel−1^{-1})5, and GNIRS 1-, 2-, 3-arcsec and native extractions. Symbols distinguish these choices, with filled squares for primary spectral and February imaging measurements and open symbols for July.

Vega references and solar restoration use the accepted spectral intervals, so these colors differ from full-filter 2MASS/MOSFIRE colors. GNIRS nightly values combine exposure-weighted relative band fluxes; errors use the larger of propagated scatter and half the block color difference. Extraction ranges measure sensitivity across correlated aperture reductions. Standard-star, telluric, chromatic slit-loss, and missing-coverage systematics remain unquantified. F2 HK is restricted to 1.45–1.80 μ\mum and supplies no KsK_s color.

Comparison with solar colors and small-body populations

Matched spectral intervals and Vega references give solar J−H=0.324J-H=0.324 mag for F2 and 0.331/0.326 mag for December 1/2 GNIRS. Comet excesses are −0.082-0.082, −0.061-0.061, and −0.049-0.049 mag, respectively; GNIRS excesses are −0.094/−0.111-0.094/-0.111 mag in H−KsH-K_s and −0.155/−0.159-0.155/-0.159 mag in J−KsJ-K_s. These intervals show a neutral to weakly blue continuum within comparison-star and response uncertainties. Full 2MASS solar colors are (J−H,H−Ks,J−Ks)=(0.332,0.043,0.374)(J-H,H-K_s,J-K_s)=(0.332,0.043,0.374) mag; the interval-matched values account for incomplete coverage.

Table 8 gives unweighted medians and sample standard deviations from the object-averaged MBOSS catalog ([48])6. Each object contributes once per available color. We divide gray and red Centaurs at B−R=(B−V)+(V−R)=1.6B-R=(B-V)+(V-R)=1.6 mag and combine resonant, classical, scattered, and detached TNOs, including large icy objects. We derive J−KJ-K for objects with both J−HJ-H and H−KH-K, whose epoch sets may differ. Sample sizes vary by color, and dispersions include measurement noise and population diversity.

MBOSS J−HJ-H medians are 0.39 mag for both Centaur groups, 0.40 mag for TNOs, and 0.45 mag for Jupiter Trojans; H−KH-K and J−KJ-K medians span −0.06-0.06–0.16 and 0.33–0.59 mag. July MOSFIRE H−KsH-K_s is redder than these medians, December GNIRS lies near the red-Centaur value, and February MOSFIRE J−KsJ-K_s falls within the TNO dispersion.

UTC dateInstrumentBandt−Tpt-T_p (d)θρ\theta_\rho (arcsec)MagnitudeA(0)fρA(0)f\rho (cm)
2025-08-27GMOS-Sg-62.55.3316.512 ± 0.036552 ± 18
2025-08-27GMOS-Si-62.55.3315.495 ± 0.030826 ± 23
2025-08-27GMOS-Sr-62.55.3315.758 ± 0.030725 ± 20
2025-08-28GMOS-Su-62.55.3317.694 ± 0.067610 ± 38
2025-09-05GMOS-Sz-53.55.3814.984 ± 0.0331045 ± 31
2025-09-05GMOS-Sg-53.55.3816.102 ± 0.040649 ± 24
2025-09-05GMOS-Si-53.55.3815.109 ± 0.031949 ± 27
2025-09-08GMOS-Sz-50.55.4014.831 ± 0.0331111 ± 33
2025-09-08GMOS-Sg-50.55.4015.913 ± 0.040712 ± 26
2025-09-08GMOS-Si-50.55.4014.956 ± 0.0301010 ± 28
2025-09-08GMOS-Su-50.55.4016.575 ± 0.0911271 ± 106
2025-09-14GMOS-SrMr^M-44.55.4214.775 ± 0.2081115 ± 213
2025-11-26GMOS-Nzcz^c28.27.0312.026 ± 0.0335552 ± 167
2025-11-26GMOS-NrMr^M28.27.0312.523 ± 0.2023998 ± 745
2025-12-22GMOS-Nrcr^c54.07.6513.583 ± 0.0302114 ± 59
2026-01-20GMOS-Nz82.96.0014.663 ± 0.0331090 ± 33
2026-01-20GMOS-Nr82.96.0015.011 ± 0.030900 ± 25
2026-02-05GMOS-Nz98.84.8116.057 ± 0.033749 ± 23

Table 2. Photometry and Halley–Marcus phase-corrected activity in 10,000-km-radius apertures.

Filter variants in MBOSS and partial passbands in our spectra limit these comparisons. Its only comet nucleus with both J−HJ-H and H−KH-K is 49P/Arend–Rigaux.

For active comae, D. Jewitt & K. J. Meech [68] summarizes weighted mean J−HJ-H of 0.46±0.070.46 \pm0.07 and 0.41±0.060.41 \pm0.06 mag and H−KH-K of 0.10±0.080.10 \pm0.08 and 0.14±0.050.14 \pm0.05 mag for the 14- and 9-comet samples of D. Jewitt & K. J. Meech [67, 68]. The errors include systematic contributions. P/Giclas lacks H−KH-K in the first sample, and Bradfield is excluded from the second H−KH-K mean because of possible thermal contamination. M. S. Hanner et al. [52] contributes measurements to the 1986 sample. C/2005 E2 (McNaught) has J−H=0.48±0.01J-H=0.48 \pm0.01, H−K=0.01±0.01H-K=0.01 \pm0.01, and J−K=0.49±0.01J-K=0.49 \pm0.01 mag on 2005 September 26–27 (E. Picazzio et al. [92], their Table 1). Table 8 also retains the active- and inactive-comet compilations of J. de León et al. (2020). C-, P-, and D-type asteroid colors overlap in 2MASS ([113] (T. Grav & M. J. Holman 2004)). Comparisons between free coma grains and surface regoliths depend on grain size, porosity, phase angle, gas, and ice.

Colors are in magnitudes. F2/GNIRS use accepted passband intervals; MOSFIRE uses approximate native-filter Vega calibration. The asterisk marks the July Y−JY-J sensitivity scale; July JHKsJH K_s errors include aperture sensitivity. These empirical scales have unestablished Gaussian coverage and exclude common contamination or calibration bias. February errors exclude shared reference-star mismatch. MBOSS uses heterogeneous KK filters without transformation to KsK_s, with J−KJ-K derived for paired objects and errors describing sample dispersion. For 49P, catalog measurement errors are 0.141 mag in each of J−HJ-H and H−KH-K; missing shared-band covariance prevents a J−KJ-K error estimate. MBOSS has no YY or long-period-comet NIR entries. The 1986 and 1988 coma means (D. Jewitt & K. J. Meech [67, 68]) are quoted from D. Jewitt & K. J. Meech [68], with errors on weighted means including systematics. Different J−HJ-H and H−KH-K samples and weights prevent deriving a mean J−KJ-K. McNaught colors and errors follow Table 1 of E. Picazzio et al. [92]; the averaging aperture is unspecified. The final four rows give medians and observation counts from J. de León et al. (2020), drawing on M. S. Hanner et al. [52]; M. Popescu et al. [94]; O. R. Hainaut et al. [48]; M. V. Sykes et al. [113]; E. Picazzio et al. [92]; no dispersions are supplied. Dashes denote unavailable colors. Solar colors use Vega references, with UKIRT (Y−JY-J)⊙=0.27_\odot= 0.27 mag from C. N. A. Willmer [118]. Filter systems and scattering states limit taxonomic comparisons.

Figure 21 compares solar-relative colors with VISTA asteroids [94, 93]: July 11 MOSFIRE and December 1/2 GNIRS in J−HJ-H versus H−KsH-K_s, and July MOSFIRE in Y−JY-J versus J−KsJ-K_s. Both panels use 1886 classified objects with all three catalog colors (Y−JY-J, J−KsJ-K_s, H−KsH-K_s) finite and 0<σ≤0.100 < \sigma\le0.10 mag; J−H=(J−Ks)−(H−Ks)J-H = (J-K_s) - (H-K_s). VISTA solar colors are (Y−J,J−H,H−Ks,J−Ks)⊙=(0.219,0.254,0.082,0.336)(Y-J, J-H, H-K_s, J-K_s)_\odot= (0.219, 0.254, 0.082, 0.336) mag [93]. Crosses mark mean synthetic colors of A, B, C, D, K, L, O, Q, R, S, T, V, and X templates, with one object each for O and R.

The left panel adds the MBOSS medians, 49P nucleus, and Jewitt–Meech coma samples, subtracting solar (J−H,H−K)(J-H, H-K) of (0.23, 0.06) mag for MBOSS [49], (0.32, 0.07) mag for D. Jewitt & K. J. Meech [67], and (0.31, 0.06) mag for D. Jewitt & K. J. Meech [68]. Bars show MBOSS sample dispersions, Jewitt–Meech errors on means, and 49P measurement errors. Group summaries may sample different objects in each color. McNaught lacks matched solar colors and appears only in Table 8.

July points use the 2.16-arcsec aperture and empirical uncertainties, including the 0.143-mag Y−JY-J scale. Relative to matched solar colors, July H−KsH-K_s is redder than December, all three epochs have slightly blue J−HJ-H, and July Y−JY-J and J−KsJ-K_s are redder than solar. Contamination, calibration transfer, and differing extractions limit evidence for intrinsic evolution. Solar subtraction leaves differences among the MOSFIRE, partial GNIRS, VISTA, and historical K/KsK/K_s responses. Marginal error bars omit shared-HH covariance on the left and shared-JJ covariance on the right.

Water-ice models and absorption constraints

Absorption depth is D=1−⟨R⟩/RcD = 1 - \langle R\rangle/R_c relative to continuum RcR_c. Atmospheric gaps near 1.41.4 and 1.9 μm1.9\ \mu\mathrm{m} make the ice-band depths continuum-sensitive, so we also fit pure-ice templates.

B. Yang et al. [119] found no ice bands in IRTF/SpeX and Gemini/GNIRS spectra of 2I/Borisov, limiting ice to approximately 10% of the coma cross section under their mixture model. For 3I/ATLAS, B. Yang et al. [120] modeled July IRTF/SpeX 2.0-μm2.0\text{-}\mu\mathrm{m} absorption with approximately 70% Tagish Lake material and 30% 10-μm10\text{-}\mu\mathrm{m} ice grains; the 1.5-μm1.5\text{-}\mu\mathrm{m} band was undetected. Their order-of-magnitude areal fraction differs from our fitted ice contribution to scattered light at 1.20 μm1.20\ \mu\mathrm{m}.

Following B. Yang et al. [119], we compute 150 K crystalline-ice spectra for radii a=1,10,100 μma = 1, 10, 100\ \mu\mathrm{m} from the optical constants of R. M. E. Mastrapa et al. [78]; R. M. E. Mastrapa [77]. The Hapke equivalent-slab model includes Fresnel surface terms and a mean absorption path equal to the grain diameter. We also calculate homogeneous-sphere Mie efficiencies and single-scattering albedo w=Qsca/Qextw = Q_{\mathrm{sca}}/Q_{\mathrm{ext}} (S. Protopapa et al. [95]). Both use an isotropic geometric-albedo mapping and 4.7-nm FWHM smoothing, omitting porosity, dirty-ice mixtures, and a measured ice phase function. Figure 19 shows 10% mixtures with continua fitted over 1.151.15–1.301.30 and 1.701.70–1.78 μm1.78\ \mu\mathrm{m}, normalized at 1.20 μm1.20\ \mu\mathrm{m}.

We fit a linear or quadratic continuum times a mixture of smooth and normalized ice spectra, replacing the amorphous-carbon component of B. Yang et al. [119] with a smooth continuum. The nonnegative fraction measures ice-scattered light at 1.20 μm1.20\ \mu\mathrm{m}; mass or area fractions require additional grain assumptions. Profiling over continuum coefficients gives Δχ2=9\Delta\chi^2 = 9 ceilings after scaling scatter errors by continuum residuals. The ranges span both scattering models and continuum orders, excluding unknown smooth calibration errors. Because the ice fraction can approach zero, nondetection leaves grain radius unconstrained.

Figure 19 combines F2 settings and nightly GNIRS blocks. The F2 cutoff removes usable 2-μm2\text{-}\mu\mathrm{m} and KsK_s coverage; retained JH constrains the short-wavelength ice-band shape with strong continuum dependence. Neither instrument yields absorption robust to the tested continuum and calibration choices.

For 10-μm10\text{-}\mu\mathrm{m} grains, ceilings span 15–36% of the 1.20-μm1.20\text{-}\mu\mathrm{m} scattered light in F2 JH, 0.6–2.7% for GNIRS December 1, and 0.7–2.5% for December 2 (Table 9). Unquantified telluric and smooth response errors, together with different coverage and slit extractions, prevent an ice-depletion inference.

SpectrumGrain radius (μm\mu\mathrm{m})Ice-light ceiling range (%)
F2 JH135.7–100.0
F2 JH1015.2–36.0
F2 JH1009.7–20.8
GNIRS 2025120111.5–7.2
GNIRS 20251201100.6–2.7
GNIRS 202512011000.3–1.4
GNIRS 2025120211.8–7.2
GNIRS 20251202100.7–2.5
GNIRS 202512021000.3–1.2

Table 9. Ice-template sensitivity at 1.20 microns.

Ceilings give the nonnegative ice fraction of scattered light at 1.20 μm1.20\ \mu\mathrm{m}, profiled over linear/quadratic continua at Δχ2=9\Delta\chi^2 = 9. Ranges span Hapke-equivalent-slab and Mie-SSA models for 150 K crystalline ice; broad residuals inflate the scatter. Limits exclude unmeasured smooth response errors. Ice mass remains unconstrained, and grain size is degenerate with abundance.

Discussion and Conclusions

Activity, dust dynamics, and surface evolution

The evolution from a concentrated coma and compact sunward feature to an extended tail is consistent with accumulation and radiation-pressure sorting of dust released over many dates ([66]). Slowly moving grains can persist as production weakens. The synchrones and syndynes allow a range of release ages and grain sizes, while our cadence cannot resolve the periodic angle modulation of the high-latitude jet reported by M. Serra-Ricart et al. (2025). Similar continuum colors can accompany different spatial dust distributions, as also seen in C/2020 V2 (A. Voitko et al. [116]).

Our red-band activity measurements extend the baseline of A. Graykowski et al. [45] to 7.12 au. The steeper outbound decline persists when a separate Liverpool normalization is allowed, although the fitted indices change from −2.10±0.14-2.10 \pm0.14 and −3.34±0.43-3.34 \pm0.43 to −1.64±0.11-1.64 \pm0.11 and −3.10±0.18-3.10 \pm0.18 inbound and outbound, respectively. Delayed release, changing active areas, and retained grains can produce different branch normalizations. The perihelion gap prevents us from identifying when these changes occurred or whether a larger scattering area preceded the observed maximum on 2025 December 19, approximately 51 days after perihelion. The seven-day profile peaks near t=+52t = +52 days.

For fixed grain properties, the cross-section derivative measures dust replenishment minus escape and destruction. The aperture maximum occurs when these terms balance, so long crossing times can delay it relative to peak nucleus production. Thermal lag or changing illumination could also sustain outbound release. The subsequent fading shows that losses exceed replenishment, but its linear extrapolation cannot establish a unique cessation date. Phase-function, relative-calibration, and sampling uncertainties further affect the maximum’s amplitude.

At fixed β\beta, our perpendicular epoch medians exceed the July HSTHST value by factors of 1.9–2.6; our axial sunward estimates exceed the corresponding HSTHST estimate by factors of 1.4–2.1 (D. Jewitt et al. [63]). The sunward coefficients, 70.3–105.3 m s−1^{-1}, bracket Tianwen-1’s 100 m s−1^{-1} (X. Ren et al. [100]); at its inferred β=10−3\beta= 10^{-3}–10−210^{-2}, our speeds are 2.2–10.5 m s−1^{-1}, overlapping its 3–10 m s−1^{-1} estimate. The approximately 5-m s−1^{-1} speed adopted for 100-μ\mum grains by D. Jewitt & J. Luu [66] uses the HSTHST normalization with a different size conversion. Our illustrative radial-tail projection converts B. Liu et al. [74]’s perpendicular coefficient to 16.5 ±\pm 1.3 m s−1^{-1}, within our 14.5–19.5-m s−1^{-1} range; their sunward boundary extends beyond our half-maximum definition. Faster sunward motion is consistent with preferential dayside ejection, although differences in epoch, seeing, grain age, and width definition limit tests of temporal acceleration.

The transverse and sunward models give dust losses of 2.7×1092.7 \times10^{9} and 1.0×10101.0 \times10^{10} kg for 1-μ\mum–1-mm grains over the observed baseline. With the 1.3-km nucleus radius of M.-T. Hui et al. [58], these correspond to a global dust-equivalent shell approximately 0.3–1 m thick. Comparing these budgets with M. Belyakov et al. [7] requires accounting for the different grain-size and optical-property weighting of scattered and thermal emission, as well as velocity, albedo, and maximum grain size. Concentrated activity could excavate more deeply than the global equivalent; the active area, dust-to-ice ratio, and initial mantle thickness remain unknown.

Cosmic-ray processing has been proposed to explain 3I/ATLAS’s volatile and continuum properties (R. Maggiolo et al. [75]). Delayed methane-release models admit different combinations of initial dust content and irradiation exposure (A. Rawat & K. Acharyya [98]). Our erosion estimates alone cannot distinguish these histories without an independent constraint on processing depth.

Continuum colors, gas, and ice

The three GMOS epochs with matched grigri photometry have consistent colors despite substantial morphological evolution. Changes in spatial grain distribution therefore need not produce a measurable change in aperture-averaged color. Grain size, phase function, and gas emission all affect broadband colors. In particular, the C2_2 Swan bands at 486–518.5 nm, including the 516.5-nm bandhead, contribute to gg and can broaden its coma or reduce dust-jet contrast. Features shared by gg and rr support a dust contribution.

C2_2 production approximately doubles between 2025 September 14–15 and November 26 (Table 4), supporting a growing gas contribution across perihelion. The November spectrum precedes the largest measured red-band cross section by 23 days, leaving the timing of peak C2_2 production unresolved. Liverpool’s g−r=0.370±0.036g-r = 0.370 \pm0.036 mag on 2025 December 19 is bluer than the GMOS mean of 0.732 mag, indicating enhanced relative gg brightness. The December–February C2_2 limits allow declining production after November. Non-simultaneous spectra, other gas bands, dust color, and calibration differences prevent a unique interpretation of the December color change; rr can also contain gas emission. Quantifying the gas fraction requires a spatial gas model integrated through the imaging aperture and filter response. Our long-slit data support a qualitative comparison and identify gas contamination as a limitation of gg-band AfρAf\rho.

The rise in C2_2/CN is less sensitive to a shared flux-normalization error than either production rate alone. Relative spectral response, molecular lifetimes and distributions, illumination, and distributed grain sources still affect the ratio, while late C3_3 measurements depend strongly on continuum subtraction. Inbound CN with carbon-chain upper limits (M. Lazzarin et al. [72]) and prominent outbound CN, C3_3, and C2_2 with normal-comet mixing ratios (H. Kawakita et al. [71] document stronger carbon-chain emission outbound, as also reported by G. Ahuja et al. [3]. Their water-normalized abundances require conversion for comparison with our CN-normalized ratios. Spatial profiles (A. C. Mura et al. [87]) test the CN and metal scale lengths used in aperture corrections. LRIS Ni I provides a refractory-species measurement, but Ni I/CN requires distinct fluorescence and source models before interpretation as an elemental abundance.

The neutral to weakly blue NIR spectra agree qualitatively with the turnover near 1.1 µm reported by K. Medler et al. [79] for a continuum resembling 2I/Borisov. Curved grain reflectance can produce this behavior. February MOSFIRE colors are near-solar under the reference-star assumption; July Y−JY-J remains sensitive to stellar calibration and contamination. December nightly spectral differences fall below the block repeatability scale. Residual stars, unequal apertures, and passband differences limit color-evolution tests, and comparisons with inactive bodies must account for scattering by free coma grains rather than regoliths.

For 1010-µm pure-ice grains, GNIRS limits their contribution to roughly 1–3% of the scattered light at 1.20 µm under the tested continuum and scattering models; unknown smooth calibration errors are excluded. The early ice spectra of B. Yang et al. [120] and the outer-coma dirty-ice interpretation of B. Liu et al. [74] sample different epochs and regions. An icy population that contributes little to our inner-coma continuum remains allowed. MIRI silicate spectra constrain grain composition through thermal emission (M. Belyakov et al. [7]).

Comparison with 1I, 2I, and the interstellar population

The sustained, resolved activity of 3I/ATLAS resembles 2I/Borisov more closely than 1I/‘Oumuamua. Yet 1I showed nongravitational acceleration without a detected dust coma (M. Micheli et al. [81]), and 2I had unusually CO-rich gas ([10, 34]). Similar optical colors across these objects can accompany different volatile inventories, grain properties, and source geometries. The 1.3-km radius inferred for 3I (M.-T. Hui et al. [58]) exceeds 2I’s 0.5-km HST upper limit under their respective albedo assumptions (D. Jewitt et al. [62]); unknown active areas limit comparisons of production per unit surface area. Comae also complicate the rotation measurements obtainable from 1I’s large-amplitude, tumbling lightcurve ([80, 38]).

Three detections provide too small and selected a sample to establish the relative abundance of these properties. Active coma increase the discovery volume, while size, albedo, trajectory, cadence, perihelion distance, and observing coverage influence the detected sample. Following each encounter helps distinguish intrinsic differences among parent bodies from their responses to solar heating.

Our Keck and Gemini observations show changing morphology and gas production, including an increasing C2C_2/CN ratio, alongside stable optical colors. Image-based dust speeds imply losses of 10910^{9}–101010^{10} kg, equivalent to a global layer approximately 0.3–1 m thick under the adopted models. The NIR spectra place model-dependent limits on the ice contribution to scattered light, with no robust absorption detection.

Acknowledgments

The Liverpool Telescope is operated on the island of La Palma by Liverpool John Moores University in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofísica de Canarias with financial support from the UK Science and Technology Facilities Council, based on observations at the PATAG proposal AZ22A01 (PI: F. Romanov).

We thank Josh Walawender, Ian Wong, and Matthew Belyakov for their assistance with the MOSFIRE observations, reduction of the Gemini data, and observation planning.

D.F. conducted this research at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (80NM0018D0004).

This work has been supported by the Czech Science Foundation through grant 25-16789S (J. Hanuš).

This work was supported by a NASA Keck PI Data Award, administered by the NASA Exoplanet Science Institute.

Data presented herein were obtained at the W. M. Keck Observatory from telescope time allocated to the National Aeronautics and Space Administration through the agency’s scientific partnership with the California Institute of Technology and the University of California.

Some of the data presented herein were obtained at Keck Observatory, which is a private 501(c)3 non-profit organization operated as a scientific partnership among the California Institute of Technology, the University of California, and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. The raw Keck Observatory data presented herein, once public, can be accessed from the Keck Observatory Archive (KOA), which is operated by the W. M. Keck Observatory and the NASA Exoplanet Science Institute (NExScI), under contract with the National Aeronautics and Space Administration.

The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Mauna Kea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.

We thank Prof. Karen Meech, Dr. Elena Sabbi, the Gemini Observatory team, and the Shadow the Scientists (StS) team for supporting these observations and public observing sessions. Dr. Sabbi was PI of GS-2025B-DD-103, which provided the campaign’s first Gemini South DDT observations. StS is made possible through the Creating Equity in STEAM (CrEST) program at the University of California, Santa Cruz, and support from the Heising-Simons Foundation.

Based on observations obtained at the international Gemini Observatory, a program of NSF NOIRLab, which is managed by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the U.S. National Science Foundation on behalf of the Gemini Observatory partnership: the U.S. National Science Foundation (United States), National Research Council (Canada), Agencia Nacional de Investigación y Desarrollo (Chile), Ministerio de Ciencia, Tecnología e Innovación (Argentina), Ministério da Ciência, Tecnologia, Inovações e Comunicações (Brazil), and Korea Astronomy and Space Science Institute (Republic of Korea).

This research benefited from B. T. Bolin’s participation in the 2026 workshop “Exploring Planetary Systems in the Era of Time-domain Astronomy”, hosted by the Institute for Astronomy at the University of Hawai’i and supported by award 2106927 from the National Science Foundation’s Astronomy & Astrophysics Research Grants program.

We used ChatGPT (OpenAI) to assist with manuscript editing, development of analysis code and figure preparation. The authors are responsible for the scientific results and the final manuscript.

Facilities: Keck:I (LRIS, MOSFIRE), Gemini:North (GMOS, GNIRS), Gemini:South (GMOS, Flamingos-2), Liverpool:2m (IO:O)

Software: LPipe (D. A. Perley [91]), Astropy, NumPy, SciPy, Matplotlib, Photutils, sbpy (M. Mommert et al. [82]), JPL Horizons

References

  1. [1]A’Hearn, M. F., Millis, R. C., Schleicher, D. G., Osip, D. J., & Birch, P. V. 1995, Icarus, 118, 223, doi: 10.1006/icar.1995.1190DOI
  2. [2]A’Hearn, M. F., Schleicher, D. G., Millis, R. L., Feldman, P. D., & Thompson, D. T. 1984, AJ, 89, 579, doi: 10.1086/113552DOI
  3. [3]Ahuja, G., Ganesh, S., Aravind, K., et al. 2026, Monthly Notices of the Royal Astronomical Society, 552, stag1800, doi: 10.1093/mnras/stag1800DOI
  4. [4]Alvarez-Candal, A., Rizos, J. L., Lara, L. M., et al. 2025, Astronomy & Astrophysics, 700, L10, doi: 10.1051/0004-6361/202555338
  5. [5]Belyakov, M., Fremling, C., Graham, M. J., et al. 2025, arXiv e-prints. https://arxiv.org/abs/2507.11720
  6. [6]Belyakov, M., Wong, I., Bolin, B. T., et al. 2026a, ApJL, 1001, L11, doi: 10.3847/2041-8213/ae5700DOI
  7. [7]Belyakov, M., Wong, I., Lisse, C. M., et al. 2026b, ApJL, 1009, L41, doi: 10.3847/2041-8213/aea84eDOI
  8. [8]Beniyama, J. 2025, PASJ, 77, L71, doi: 10.1093/pasj/psaf097
  9. [9]Beniyama, J., Bolin, B. T., Sergeyev, A. V., et al. 2025, A&A, 700, A183, doi: 10.1051/0004-6361/202555633
  10. [10]Bodewits, D., Noonan, J. W., Feldman, P. D., et al. 2020, Nature Astronomy, 4, 867, doi: 10.1038/s41550-020-1095-2DOI
  11. [11]Bohlin, R. C., Gordon, K. D., & Tremblay, P.-E. 2014, PASP, 126, 711, doi: 10.1086/677655DOI
  12. [12]Bolin, B. T., Belyakov, M., Fremling, C., et al. 2025a, MNRAS Letters, 542, L139, doi: 10.1093/mnrasl/slaf078DOI
  13. [14]Bolin, B. T., Fernandez, Y. R., Lisse, C. M., et al. 2021, AJ, 161, 116, doi: 10.3847/1538-3881/ab94b
  14. [15]Bolin, B. T., Fremling, C., Belyakov, M., et al. 2025c, AJ, 169, 303, doi: 10.3847/1538-3881/adccbeDOI
  15. [16]Bolin, B. T., Fremling, C., Morbidelli, A., et al. 2023, MNRAS, 521, L29, doi: 10.1093/mnrasl/slad018DOI
  16. [17]Bolin, B. T., Hanuš, J., Denneau, L., et al. 2025d, ApJL, 984, L25, doi: 10.3847/2041-8213/adc910DOI
  17. [18]Bolin, B. T., & Lisse, C. M. 2020, MNRAS, 497, 4031, doi: 10.1093/mnras/staa2192DOI
  18. [19]Bolin, B. T., Lisse, C. M., Kasliwal, M. M., et al. 2020, AJ, 160, 26, doi: 10.3847/1538-3881/ab9305DOI
  19. [20]Bolin, B. T., Weaver, H. A., Fernandez, Y. R., et al. 2018, ApJL, 852, L2, doi: 10.3847/2041-8213/aaa0c9DOI
  20. [22]Bolin, B. T., Abron, L.-M., Belyakov, M., et al. 2025e, The Anti-Solar Tail of Comet 3I/ATLAS Seen in Deep Multi-band Imaging Taken at Gemini South,, The Astronomer’s Telegram, 17363 https://www.astronomerstelegram.org/?read=17363
  21. [23]Bolin, B. T., Wong, I., Lemaux, B., et al. 2025f, Gemini pre-perihelion detection of CN/C3/C2 gas emission and dust fans in interstellar comet 3I/ATLAS,, The Astronomer’s Telegram, 17503 https://www.astronomerstelegram.org/?read=17503
  22. [24]Bolin, B. T., Wong, I., Abron, L.-M., et al. 2025g, Gemini post-perihelion visible photometry and spectroscopy of interstellar comet 3I/ATLAS,, The Astronomer’s Telegram, 17561 https://www.astronomerstelegram.org/?read=17561
  23. [25]Bolin, B. T., Wong, I., Abron, L.-M., et al. 2026b, December 2025 Gemini spectroscopic and photometric constraints on the activity and gas contents of 3I/ATLAS,, The Astronomer’s Telegram, 17719 https://www.astronomerstelegram.org/?read=17719
  24. [26]Bolin, B. T., Wong, I., Abron, L.-M., et al. 2026c, 2026 January Gemini imaging and spectroscopic monitoring of 3I/ATLAS,, The Astronomer’s Telegram, 17841 https://www.astronomerstelegram.org/?read=17841
  25. [27]Bromley, S. 2026, Computed Fluorescence Efficiencies of Atomic Ni and Atomic Fe, Zenodo, doi: 10.5281/zenodo.20514283DOI
  26. [28]Bromley, S. J., Neff, B., Loch, S. D., et al. 2026, The Planetary Science Journal, 7, doi: 10.3847/PSJ/ae706d
  27. [29]Burns, J. A., Lamy, P. L., & Soter, S. 1979, Icarus, 40, 1, doi: 10.1016/0019-1035(79)90050-2DOI
  28. [30]Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, The Astrophysical Journal, 345, 245, doi: 10.1086/167900
  29. [31]Castelli, F., & Kurucz, R. L. 2003, in IAU Symposium, Vol. 210, Modelling of Stellar Atmospheres, ed. N. Piskunov, W. W. Weiss, & D. F. Gray, A20. https://arxiv.org/abs/astro-ph/0405087
  30. [32]Chandler, C. O., Bernardinelli, P. H., Jurić, M., et al. 2026, ApJL, 1001, L35, doi: 10.3847/2041-8213/ae4b3aDOI
  31. [33]Cochran, A. L., Barker, E. S., & Gray, C. L. 2012, Icarus, doi: 10.1016/j.icarus.2011.12.010DOI
  32. [34]Cordiner, M. A., Milam, S. N., Biver, N., et al. 2020, Nature Astronomy, 4, 861, doi: 10.1038/s41550-020-1087-2DOI
  33. [37]Denneau, L., Siverd, R., Tonry, J., et al. 2025, 3I/ATLAS = C/2025 N1 (ATLAS),, Minor Planet Electronic Circular 2025-N12 doi: 10.48377/MPEC/2025-N12
  34. [38]Drahus, M., Guzik, P., Waniak, W., et al. 2018, Nature Astronomy, 2, 407, doi: 10.1038/s41550-018-0440-1DOI
  35. [39]Finson, M. L., & Probstein, R. F. 1968, ApJ, 154, 327, doi: 10.1086/149761DOI
  36. [40]Fitzsimmons, A., Meech, K. J., Matrà, L., & Pfalzner, S. 2024, in Comets III, ed. K. J. Meech, M. R. Combi, D. Bockelée-Morvan, S. N. Raymond, & M. Zolensky (University of Arizona Press), 731–766. https://arxiv.org/abs/2303.17980
  37. [41]Fitzsimmons, A., Snodgrass, C., Rozitis, B., et al. 2018, Nature Astronomy, 2, 133, doi: 10.1038/s41550-017-0361-4DOI
  38. [42]Giorgini, J. D., Yeomans, D. K., Chamberlin, A. B., et al. 1996, Bulletin of the American Astronomical Society, 28, 1158. https://ssd.jpl.nasa.gov/horizons/manual.html
  39. [44]Gray, Z., Bagnulo, S., Borisov, G., et al. 2025, The Astrophysical Journal Letters, 992, L29, doi: 10.3847/2041-8213/ae0c08
  40. [45]Graykowski, A., Bolin, B., Abron, L.-M., et al. 2026, arXiv e-prints. https://arxiv.org/abs/2609.22021
  41. [47]Guzik, P., Drahus, M., Rusek, K., et al. 2020, Nature Astronomy, 4, 53, doi: 10.1038/s41550-019-0931-8DOI
  42. [48]Hainaut, O. R., Boehnhardt, H., & Protopapa, S. 2012, Astronomy and Astrophysics, 546, A115, doi: 10.1051/0004-6361/201219566
  43. [49]Hainaut, O. R., & Delsanti, A. C. 2002, A&A, 389, 641, doi: 10.1051/0004-6361:20020431DOI
  44. [50]Hamuy, M., Suntzeff, N. B., Heathcote, S. R., et al. 1994, PASP, 106, 566, doi: 10.1086/133417DOI
  45. [51]Hamuy, M., Walker, A. R., Suntzeff, N. B., et al. 1992, PASP, 104, 533, doi: 10.1086/133028DOI
  46. [52]Hanner, M. S., Tokunaga, A. T., Veeder, G. J., & A’Hearn, M. F. 1984, AJ, 89, 162, doi: 10.1086/113495DOI
  47. [53]Hartman, J. D., Bakos, G. Á., Jordán, A., et al. 2026, The Astronomical Journal, 171, 270, doi: 10.3847/1538-3881/ae4a99DOI
  48. [54]Hinkle, J. T., Yang, B., Meech, K. J., et al. 2026, The Open Journal of Astrophysics, 9, doi: 10.33232/001c.163250DOI
  49. [55]Hoogendam, W. B., Shappee, B. J., Wray, J. J., et al. 2025, arXiv e-prints, doi: 10.48550/arXiv.2510.11779
  50. [56]Hoogendam, W. B., Kuesters, D., Shappee, B. J., et al. 2026a, The Open Journal of Astrophysics, 9, doi: 10.33232/001c.165099DOI
  51. [57]Hoogendam, W. B., Jones, D. O., Yang, B., et al. 2026b, ApJ, 1003, 245, doi: 10.3847/1538-4357/ae6243DOI
  52. [58]Hui, M.-T., Jewitt, D., Mutchler, M. J., Agarwal, J., & Kim, Y. 2026, ApJL, 999, L37, doi: 10.3847/2041-8213/ae471cDOI
  53. [59]Husser, T.-O., Wende-von Berg, S., Dreizler, S., et al. 2013, A&A, 553, A6, doi: 10.1051/0004-6361/201219058DOI
  54. [60]Hutsemekers, D., Manfroid, J., Opitom, C., et al. 2026, arXiv e-prints. https://arxiv.org/abs/2605.07652
  55. [61]Ivezić, Ž., Lupton, R. H., Jurić, M., et al. 2002a, The Astronomical Journal, 124, 2943, doi: 10.1086/344077DOI
  56. [62]Jewitt, D., Hui, M.-T., Kim, Y., et al. 2020, ApJL, 888, L23, doi: 10.3847/2041-8213/ab621bDOI
  57. [63]Jewitt, D., Hui, M.-T., Mutchler, M., Kim, Y., & Agarwal, J. 2025, ApJL, 990, L2, doi: 10.3847/2041-8213/adf8d8DOI
  58. [64]Jewitt, D., Kim, Y., Luu, J., et al. 2019, ApJL, 876, L19, doi: 10.3847/2041-8213/ab1be8DOI
  59. [65]Jewitt, D., & Luu, J. 2019, ApJL, 886, L29, doi: 10.3847/2041-8213/ab530bDOI
  60. [66]Jewitt, D., & Luu, J. 2025, ApJL, 994, L3, doi: 10.3847/2041-8213/ae1832DOI
  61. [67]Jewitt, D., & Meech, K. J. 1986, ApJ, 310, 937, doi: 10.1086/164745DOI
  62. [68]Jewitt, D., & Meech, K. J. 1988, AJ, 96, 1723, doi: 10.1086/114924DOI
  63. [69]Jewitt, D. C., & Meech, K. J. 1987, ApJ, 317, 992, doi: 10.1086/165347DOI
  64. [70]Kareta, T., Champagne, C., McClure, L., et al. 2025, The Astrophysical Journal Letters, 990, L65, doi: 10.3847/2041-8213/adbfdf
  65. [71]Kawakita, H., Tsujimoto, K., Shinnaka, Y., et al. 2026, The Astrophysical Journal Letters, 1000, L60, doi: 10.3847/2041-8213/ae4c5bDOI
  66. [72]Lazzarin, M., Mura, A. C., La Forgia, F., et al. 2026, The Astrophysical Journal Letters, 998, L30, doi: 10.3847/2041-8213/ae3dd7DOI
  67. [73]Lee, E. J., DeRocco, W., Hadden, S., & Gaudi, B. S. 2026, The Astrophysical Journal Letters, 1005, L18, doi: 10.3847/2041-8213/ae796fDOI
  68. [74]Liu, B., Zhang, X., Liu, X., & Hui, M.-T. 2026, ApJL, 1008, L26, doi: 10.3847/2041-8213/ae9a89DOI
  69. [75]Maggiolo, R., Dhooghe, F., Gronoff, G. P., de Keyser, J., & Cessateur, G. 2026, The Astrophysical Journal Letters, 996, L34, doi: 10.3847/2041-8213/ae2fffDOI
  70. [76]Martinez-Palomera, J., Tuson, A., Hedges, C., et al. 2025, The Astrophysical Journal Letters, 994, L51, doi: 10.3847/2041-8213/ae1f91
  71. [77]Mastrapa, R. M. E. 2020, Optical Constants of Water Ice: 150 K Crystalline Ice, NASA Planetary Data System, doi: 10.26033/c30a-es84DOI
  72. [78]Mastrapa, R. M. E., Bernstein, M. P., Sandford, S. A., et al. 2008, Icarus, 197, 307, doi: 10.1016/j.icarus.2008.04.008DOI
  73. [79]Medler, K., Hoogendam, W. B., Ashall, C., et al. 2026, AJ, 172, 25, doi: 10.3847/1538-3881/ae6b71DOI
  74. [80]Meech, K. J., Weryk, R., Micheli, M., et al. 2017, Nature, 552, 378, doi: 10.1038/nature25020DOI
  75. [81]Micheli, M., Farnocchia, D., Meech, K. J., et al. 2018, Nature, 559, 223, doi: 10.1038/s41586-018-0254-4DOI
  76. [82]Mommert, M., Kelley, M. S. P., de Val-Borro, M., et al. 2019, Journal of Open Source Software, 4, 1426, doi: 10.21105/joss.01426DOI
  77. [83]Morate, D., Marcio Carvano, J., Alvarez-Candal, A., et al. 2021, Astronomy & Astrophysics, 655, A47, doi: 10.1051/0004-6361/202038477DOI
  78. [84]Moreno, F., Pozuelos, F., Novákovic, B., et al. 2017, ApJL, 837, L3, doi: 10.3847/2041-8213/aa6036DOI
  79. [85]Moreno, F., Serra-Ricart, M., Licandro, J., et al. 2026, MNRAS, 550, stag1164, doi: 10.1093/mnras/stag1164DOI
  80. [86]Moro-Martín, A. 2018, The Astrophysical Journal, 866, 131, doi: 10.3847/1538-4357/aadf34
  81. [87]Mura, A. C., Kawakita, H., La Forgia, F., et al. 2026, Astronomy & Astrophysics, 712, A9, doi: 10.1051/0004-6361/202660649DOI
  82. [88]Ofek, E. O. 2012, The Astrophysical Journal, 749, 10, doi: 10.1088/0004-637X/749/1/10DOI
  83. [90]Paek, G. S. H., Im, M., Jeong, M., et al. 2026, ApJ, 1000, 53, doi: 10.3847/1538-4357/ae47f0DOI
  84. [91]Perley, D. A. 2019, Publications of the Astronomical Society of the Pacific, 131, 084503, doi: 10.1088/1538-3873/ab215dDOI
  85. [92]Picazzio, E., Figueredo, E., de Almeida, A. A., Mendes de Oliveira, C., & Churyumov, K. I. 2010, Proceedings of the International Astronomical Union, 5, 285, doi: 10.1017/S1743921310001948DOI
  86. [93]Popescu, M., Licandro, J., Carvano, J. M., et al. 2018, Astronomy & Astrophysics, 617, A12, doi: 10.1051/0004-6361/201833023DOI
  87. [94]Popescu, M., Licandro, J., Morate, D., et al. 2016, Astronomy & Astrophysics, 591, A115, doi: 10.1051/0004-6361/201628163DOI
  88. [95]Protopapa, S., Kelley, M. S. P., Yang, B., et al. 2018, ApJL, 862, L16, doi: 10.3847/2041-8213/aad33bDOI
  89. [96]Puzia, T. H., Rahatgaonkar, R., Carvajal, J. P., Navak, P. K., & Luco, B. 2025, The Astrophysical Journal Letters, 990, L27, doi: 10.3847/2041-8213/adfa0bDOI
  90. [97]Rahatgaonkar, R., Carvajal, J. P., Puzia, T. H., et al. 2025, ApJL, 995, L34, doi: 10.3847/2041-8213/ae1cbcDOI
  91. [98]Rawat, A., & Acharyya, K. 2026, The Astrophysical Journal, 1010, 106, doi: 10.3847/1538-4357/aea83cDOI
  92. [99]Raymond, S. N., Armitage, P. J., Veras, D., Quintana, E. V., & Barclay, T. 2018, Monthly Notices of the Royal Astronomical Society, 476, 3031, doi: 10.1093/mnras/sty468DOI
  93. [100]Ren, X., Yan, W., Zhao, R., et al. 2026, ApJL, 1003, L10, doi: 10.3847/2041-8213/ae61b3DOI
  94. [101]Rice, M., DeRocco, W., & Raymond, S. N. 2026, Publications of the Astronomical Society of the Pacific, 138, 082001, doi: 10.1088/1538-3873/ae9343DOI
  95. [102]Sahu, S., Gänsicke, B. T., Williams, J. T., et al. 2025, Monthly Notices of the Royal Astronomical Society, 543, 223, doi: 10.1093/mnras/staf1424DOI
  96. [103]Samarasinha, N. H., & Larson, S. M. 2014, Icarus, 239, 168, doi: 10.1016/j.icarus.2014.05.028DOI
  97. [104]Samarasinha, N. H., Martin, M. P., & Larson, S. M. 2013, Cometary Coma Image Enhancement Facility, https://cie.psi.edu/
  98. [105]Santana-Ros, T., Ivanova, O., Mykhailova, S., et al. 2025, A&A, 702, L3, doi: 10.1051/0004-6361/202556717
  99. [106]Santana-Ros, T., Sergeyev, A., Ivanova, O., et al. 2026, The Astrophysical Journal, 1007, 182, doi: 10.3847/1538-4357/ae8dbb
  100. [107]Schleicher, D. G. 2010a, Composite Dust Phase Function for Comets, Lowell Observatory https://comet.lowell.edu/comet/dustphase/
  101. [108]Schleicher, D. G. 2010b, The Astronomical Journal, 140, 973, doi: 10.1088/0004-6256/140/4/973DOI
  102. [109]Schleicher, D. G., Millis, R. L., & Birch, P. V. 1998, Icarus, 132, 397, doi: 10.1006/icar.1997.5902DOI
  103. [110]Seligman, D. Z., Micheli, M., Farnocchia, D., et al. 2025, ApJL, 989, L36, doi: 10.3847/2041-8213/adf49aDOI
  104. [112]Solontoi, M., Ivezic, Ž., Jurić, M., et al. 2012, Icarus, 218, 571, doi: 10.1016/j.icarus.2011.10.008DOI
  105. [113]Sykes, M. V., Cutri, R. M., Fowler, J. W., et al. 2000, Icarus, 146, 161
  106. [114]Tonry, J. L., Stubbs, C. W., Lykke, K. R., et al. 2012, ApJ, 750, 99, doi: 10.1088/0004-637X/750/2/99DOI
  107. [115]Tonry, J. L., Denneau, L., Alarcón, M. R., et al. 2025, The Astrophysical Journal Letters, 995, L15, doi: 10.3847/2041-8213/ae1f12DOI
  108. [116]Voitko, A., Markkanen, J., Ivanova, O., et al. 2026, Monthly Notices of the Royal Astronomical Society, 549, stag967, doi: 10.1093/mnras/stag967DOI
  109. [117]Wesolowski, M., Ryske, J., Carson, P., Gritsevich, M., & (PASP Editorial Office). 2026, Publications of the Astronomical Society of the Pacific, 138, 024402, doi: 10.1088/1538-3873/ae3eb5DOI
  110. [118]Willmer, C. N. A. 2018, ApJS, 236, 47, doi: 10.3847/1538-4365/aabfdfDOI
  111. [119]Yang, B., Kelley, M. S. P., Meech, K. J., et al. 2020, A&A, 634, L6, doi: 10.1051/0004-6361/201937129DOI
  112. [120]Yang, B., Meech, K. J., Connelley, M., Zhao, R., & Keane, J. V. 2025, ApJL, 992, L9, doi: 10.3847/2041-8213/ae08a7DOI
  113. [121]Yang, B., Li, A., Cordiner, M. A., et al. 2021, Nature Astronomy, 5, 586, doi: 10.1038/s41550-021-01336-wDOI
  114. [124]Yunerman, E. S., Price, E. M., & Öberg, K. I. 2026, The Astrophysical Journal, 1004, 63, doi: 10.3847/1538-4357/ae60feDOI

Paper details

Contents