REVIEW 2 major objections 5 minor 21 references
The near infrared airglow continuum conundrum. Constraints for ground-based faint object spectroscopy
T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash
Pith's one-line read Dedicated narrow-band observations at a high-altitude Canary Islands site show that the near-infrared airglow continuum is real atmospheric emission, with zenith radiance 60–170 ph s−1 m−2 μm−1 arcsec−2 (21.4–22.8 mag arcsec−2), two to…
desk verdict Careful, honest first measurement of the ORM VIS-NIR airglow continuum; the LSF-wing estimate is the one soft spot, but the qualitative result—a dark interline sky—survives. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing technique is pre-dispersive narrow-band filtering: 60 to 120 Å filters mounted before the grism block most hydroxyl line radiation, so the broad Lorentzian and diffuse scattering wings that have contaminated earlier airglow-continuum measurements never build up. To bound what remains, a Gaussian-plus-Lorentzian line spread function is fitted to Thorium–Argon arc lines, and a deliberately conservative estimate treats the residual 1% of line flux outside the fitted core as diffuse scatter spread over the filter bandpass; that upper limit falls below the 0.6 electron systematic bias uncertainty. Zenith scaling uses a geometric layer model for the finite-thickness emitting layer at roughly 87 km altitude, together with effective optical depths for airglow and zodiacal light that account for line-of-sight scattering.
What would settle it
Take deep arc or laser spectra through the same filters and grism, long enough to expose the line wings beyond about 6 Å from bright lines, and measure the actual diffuse scattering fraction; a value well above 1% would force the reported continua downward. A complementary check is to observe the same sky fields with a hydroxyl-suppressing or double-pass spectrograph and see whether the interline continuum drops.
Extended reading notes
Core claim
On the paper's own terms, the central claim is that the interline airglow continuum in the silicon-detector near infrared is genuine sky emission, measurable without contamination from grating-scattered hydroxyl light when the bandpass is pre-filtered. The zenith-scaled radiance ranges between $60$ and $170\ \mathrm{ph\,s^{-1}\,m^{-2}\,\mu m^{-1}\,arcsec^{-2}}$, or $21.4$--$22.8\ \mathrm{mag\,arcsec^{-2}}$, making it two to four times brighter than the zodiacal light toward the ecliptic poles, the darkest foreground available from ground or space. The continuum is nearly flat in color between 6720 and 8700 Å, stable within a night with a modest decay toward morning, and comparable within errors to values reported for other major observatory sites. In the 7700 Å band the authors can only set an upper limit near 22 mag arcsec$^{-2}$ because no clean interline window exists, and in the 10,500 Å band the system sensitivity was too low for a detection.
Load-bearing premise
The conclusion that the measured continuum is not scattered hydroxyl light rests on a deliberately crude estimate in which only the residual 1% of arc-line flux beyond the fitted line core is treated as diffuse grating scatter and spread over each filter bandpass; if the true scattering wings at separations of several angstroms are stronger than 1%, part of the reported continuum would be instrumental.
Editorial extensions
If this is right
- If the continuum is as dark as $60$--$170\ \mathrm{ph\,s^{-1}\,m^{-2}\,\mu m^{-1}\,arcsec^{-2}}$, exposure-time estimates for faint-object NIR spectroscopy should budget for roughly two to four times the zodiacal-light foreground rather than the zodiacal floor alone.
- Ground-based spectrographs can meaningfully approach zodiacal-light-limited performance in the silicon NIR by adding hydroxyl-suppression masks or fiber Bragg gratings and by using double-pass or white-pupil designs that suppress grating scattering wings.
- Temporal variability of up to a factor of two on timescales of tens of minutes to hours must be built into survey strategies, calibrations, and sky subtraction for faint-object observations.
- The measured values are consistent with recent flat-continuum results from another major site, supporting the view that part of the redder 'continuum' seen in older low-resolution spectra may have been instrumental.
- The 7700 and 10,500 Å bands need deeper or cleaner measurements before the full VIS–NIR window is characterized.
Reading between the lines
- An implication the authors leave implicit is that if even a fraction of the 'continuum' is unresolved hydroxyl emission rather than a true continuum, its temporal variability and line-strength correlations should track mesospheric chemistry; the observed correlation between continuum and OH and O$_{2}$ line flux in the 8700 Å band is a hint that faint unseparated lines contribute.
- A direct test of the method's robustness is to repeat the same bandpasses with different slit widths and grating orders: a true airglow signal should be unchanged, while scattered-light contamination would scale with the optical setup.
- The same narrow-band-before-dispersion design could be applied to a network of sites with a long-term monitoring campaign, separating solar-cycle, seasonal, and meteorological drivers of the continuum.
- The 10,500 Å upper limit is sensitivity-limited rather than sky-limited, so a deeper detector or a thicker air column could convert the reddest point into a detection and constrain the red edge of the silicon window.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This paper reports an observational measurement of the near-infrared airglow continuum at Observatorio del Roque de los Muchachos using ALFOSC on the Nordic Optical Telescope. Narrow-band filters (60–120 Å wide) were placed before the grism to minimize the flux of OH lines reaching the disperser, and spectra were taken at large zenith distance in bands centered at 6720, 7700, 8700, and 10500 Å. After a careful bias subtraction (0.6 e-/px systematic), sky-line masking, flux calibration, and zenith scaling using a van Rhijn layer model and Noll et al. extinction factors, the authors derive apparent and zenith-equivalent continuum radiances. They report 60–170 ph s−1 m−2 µm−1 arcsec−2 (21.4–22.8 mag arcsec−2) depending on band and time, find the continuum to be two to four times brighter than zodiacal light toward the ecliptic poles, and argue that the signal is not dominated by grating-scattered OH light. The 7700 Å band yields only an upper limit and the 10500 Å band is detector-noise limited.
Significance. If correct, this would be one of the few modern measurements of the VIS–NIR airglow continuum at a major site, and it would support the idea that the interline sky can approach zodiacal-light-limited foreground for ground-based faint-object spectroscopy. The paper is unusually careful about detector systematics (the bias model is tested on 120 frames), and it provides a useful compilation of historical continuum measurements converted to common units. The design choice of placing narrow-band filters before the disperser is a sensible way to reduce grating scatter. However, the central quantitative claim rests on an unvalidated assumption about the line-spread-function wing, so the absolute radiances should be treated with caution until that assumption is strengthened.
major comments (2)
- [Sec. 2.4] The upper limit on grating-scattered OH light is not established by the arc-line analysis as presented. The authors state that the Grism #9 arc exposures 'appeared mainly Gaussian' and that this 'might indicate that the acquired Grism #9 arc exposures were actually not deep enough to expose the Lorentzian wings.' A Gaussian-dominated fit constrains only the core; the residual 1% of flux outside the fitted core is a property of the fitted model, not a measured bound on a broad diffuse wing. If the true LSF carries several percent of the line flux in a wing extending over tens of Å, the scattered contribution scales up from the quoted <10 ph s−1 m−2 µm−1 arcsec−2 to a level comparable to the detected 1–3 e− px−1 signal. Because Sec. 5.1's conclusion that the continuum is 'not introduced by grating scattered light' rests on this bound, the quantitative 60–170 ph s−1 m−2 µm−1 arcsec−2 claim for the 7700 and 8700 Å bands is not yet secured. The authors should either obtain deeper arc exposures that actually constrain the wing, or present the 7700/8700 radiances as upper limits.
- [Secs. 4.3-4.4] The zenith-equivalent values in Table 3 and the headline range in Sec. 6 depend on adopting the OH layer altitude (87 km) for the continuum and on extrapolating the Noll et al. fext relations to X>4, with the paper itself noting that overestimating fext,ag leads to overestimating Iag(0). No sensitivity analysis or quantitative error term is given for these choices. Because the observations are at z ≈ 75–80°, small changes in the effective airmass or fext propagate directly into the reported radiances. A short sensitivity table varying the layer height, the fext extrapolation, and the ZL model would make the '60–170 ph s−1 m−2 µm−1 arcsec−2' range and the 'two to four times brighter than ZL' statement robust.
minor comments (5)
- [Sec. 2.7] The sentence '1 Jan 2024 saw seven C-class and four M-class flares' appears to be a typo; the surrounding text describes the September 2024 run and earlier discussion refers to 1 Jan 2022. Please correct the date.
- [Sec. 6 and Table 3] The stated range '60–170 ph s−1 m−2 µm−1 arcsec−2' does not exactly match Table 3; for example, observation ID 6 gives 191 ± 49 in the emitted+scattered column. Please clarify whether ID 6 is excluded as an outlier or whether the range refers to a different column, and make the abstract and conclusions consistent with the table.
- [Sec. 2.4] The sentence 'this would lead fitting the wings to the readout noise floor, thus leading to further overestimation of the scattering wing contribution' is confusing about the direction of the bias; please reword to explain explicitly whether the adopted 1% residual is an overestimate or could be an underestimate.
- [Sec. 3] The lack of flat-fielding is justified for pixel-to-pixel variations, but the spatial illumination over the 2.4′ slit (including NB-filter vignetting on the slit wheel) is not verified to be uniform. Since the sky signal is collapsed over the full slit while standard stars are observed through a 10″ slit, a demonstration of illumination uniformity or a spatial illumination correction would strengthen the absolute calibration.
- [Throughout] There are numerous typographical issues (e.g., 'observations measurements' in Sec. 1, 'di fficult' in several places, 'increse' in Sec. 5.4, 'fits archive' in Sec. 7). A careful language edit is needed.
Circularity Check
No significant circularity: the radiance measurements are direct observational quantities with post-hoc literature comparison; the Sec. 2.4 grating-scatter estimate is an explicitly labeled upper-limit assumption, not a circular reduction.
full rationale
This paper is an observational measurement, not a derivation chain. The apparent interline continuum radiances in Sec. 4.1 are direct flux-calibrated measurements from ALFOSC spectra, and the zenith-equivalent values follow Eq. 13 using external literature inputs: airglow layer altitude and thickness from rocket experiments (Baker & Stair 1988), extinction reduction factors parameterized by Noll et al. (2012), and zodiacal light models from Kelsall et al. (1998), Wright (1998), and ESO SkyCalc. The quoted 60-170 ph s-1 m-2 um-1 arcsec-2 range is not fitted to any prior airglow continuum result; comparison to earlier measurements (Sec. 5.5, Table B.1) is performed after the fact and is explicitly post hoc. The only self-citations (Viuho et al. 2022; Andersen & Andersen 1992, Sec. 5.6) refer to prospective double-pass spectrograph designs and do not support the measured continuum. The most fragile step is the Sec. 2.4 estimate that grating-scattered OH light contributes below the systematic uncertainty; however, the authors explicitly call it a 'gross overestimate' and an upper limit, and the conclusion that the continuum is not instrumental is an inference from that assumption, not a definitional equivalence or a fitted parameter renamed as a prediction. The paper also candidly states that unresolved faint lines cannot be excluded and that it cannot distinguish true continuum from pseudo-continuum, which is an honest limitation rather than circular reasoning. No load-bearing step reduces to its own input, and no central claim is forced by self-citation.
Assumptions & free parameters
free parameters (2)
- Diffuse LSF residual fraction =
1% (assumed upper limit)
- Sky-line exclusion radius =
95% of LSF
assumptions (5)
- domain assumption Airglow continuum originates from the same atmospheric layer as OH emission, at mean altitude 87 km with FWHM 9 km.
- domain assumption Noll et al. (2012) effective extinction reduction factors fext,ag remain valid when extrapolated to airmass X > 4.
- domain assumption Line lists of Hanuschik (2003) and Rousselot et al. (2000) are complete enough that unlisted regions contain no significant sky lines.
- ad hoc to paper The diffuse grating scatter can be bounded by distributing 1% of in-band OH line flux over the NB filter FWHM.
- domain assumption Zodiacal light in VIS-NIR is described by the Kelsall et al. (1998) DIRBE model extrapolated blueward of 1.25 um.
Cite this review
Pith. "Pith review of The near infrared airglow continuum conundrum. Constraints for ground-based faint object spectroscopy." pith.science (2026). https://pith.science/paper/5XYG7RQW
@misc{pith2026250602102,
author = {Pith},
title = {Pith review of: The near infrared airglow continuum conundrum. Constraints for ground-based faint object spectroscopy},
year = {2026},
howpublished = {\url{https://pith.science/paper/5XYG7RQW}},
note = {Machine review of arXiv:2506.02102}
}
abstract
The airglow continuum in the near infrared is a challenge to quantify due to its faintness, and the grating scattered light from atmospheric hydroxyl (OH) emission lines. Despite its faintness, the airglow continuum sets the fundamental limits for ground-based spectroscopy of faint targets, and makes the difference between ground and space-based observation in the interline regions between atmospheric emission lines. We aim to quantify the level of airglow continuum radiance in the VIS -- NIR wavelength range observable with silicon photodetectors for the site Observatorio del Roque de los Muchachos in a way that our measurement will not be biased by the grating scattered light. We aim to do this by measuring the airglow continuum radiance with a minimal and controlled contamination from the broad instrumental scattering wings caused by the bright atmospheric OH lines. We measure the airglow continuum radiance with longslit $\lambda/\Delta\lambda\sim4000$ spectrograph in $\sim$100\r{A} wide narrow band passes centered at 6720, 7700, 8700 and 10500\r{A} (in line with the R, I, and Z broadbands) with the 2.5-meter Nordic Optical Telescope under photometric dark sky conditions. The bandpasses are chosen to be as clean as possible from atmospheric absorption and the OH line emission keeping the radiation reaching the grating surface at minimum. We observe the zenith equivalent airglow continuum to be 22.5mag/arcsec2 at 6720\r{A} band, and 22mag/arcsec2 at 8700\r{A}. We derive upper limits of 22mag/arcsec2 at 7700\r{A} due to difficulty to find a clean part of spectrum for measurement, and 20.8mag/arcsec2 at 10500\r{A} due to low system sensitivity. Within measurement errors and the natural variability expected for the airglow emission our results for the Observatorio del Roque de los Muchachos are comparable to values reported for other major observatory sites. (abridged)
Figures
Figures from the paper (2 more)
Reference graph
Works this paper leans on
-
[1]
Andersen, M. I. & Andersen, J. 1992, in ESO Workshop on High Resolution Spectroscopy with the VLT, V ol. 40 (European Southern Observatory), 235– 238 Arendt, R. G., Odegard, N., Weiland, J. L., et al. 1998, ApJ, 508, 74 Baker, D. J. & Stair, A. T. 1988, Physica Scripta, 37, 611–622 Bates, D. R. 1993, Proceedings of the Royal Society of London. Series A: M...
work page 1992
-
[2]
× 1 0 3 ID 3, 1 JAN 2022, UT 01:11, z = 79.76° 6700 6710 6720 6730 500 1000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 271 +/- 99 6680 6690 6700 6710 6720 6730 6740 6750 Wavelength (Å) 0.5 1.0 1.5 2.0I (ph s 1m 2 m 1arcsec
work page 2007
-
[4]
× 1 0 3 ID 5, 2 JAN 2022, UT 01:14, z = 79.58° 6700 6710 6720 6730 500 1000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 255 +/- 75 6680 6690 6700 6710 6720 6730 6740 6750 Wavelength (Å) 0.5 1.0 1.5 2.0I (ph s 1m 2 m 1arcsec
work page 2007
-
[5]
× 1 0 3 ID 12, 1 SEP 2024, UT 22:56, z = 74.96° 6700 6710 6720 6730 500 1000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 525 +/- 32 Fig. A.1. Apparent airglow spectra observed in 6720 Å band. The 2022 spectra are were recorded with 0.5 ′′slit whereas, the 2024 spectra are with 1.3′′slit. Spectral lines found in Hanuschik (2003) are indicated wit...
work page 2007
-
[6]
× 1 0 3 ID 13, 2 SEP 2024, UT 00:45, z = 77.28 ° 6700 6710 6720 6730 500 1000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 440 +/- 64 6680 6690 6700 6710 6720 6730 6740 6750 Wavelength (Å) 0.5 1.0 1.5 2.0I (ph s 1m 2 m 1arcsec
work page 2007
-
[7]
× 1 0 3 ID 14, 2 SEP 2024, UT 02:49, z = 75.89° 6700 6710 6720 6730 250 500 750 1000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 316 +/- 64 7600 7620 7640 7660 7680 7700 7720 7740 7760 7780 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0I (ph s 1m 2 m 1arcsec
work page 2007
-
[8]
× 1 0 4 ID 1, 31 DEC 2021, UT 01:22, z = 79.62° 7720 7730 7740 7750 0 2000 4000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 560 +/- 120 Fig. A.2. 6720 Å band spectra continued, and apparent airglow radiance in 7700 Å band. No clean spectral region is found without known lines, or atmospheric absorption features. An upper limit on the continuum r...
work page 2007
-
[9]
× 1 0 4 ID 10, 21 MAR 2022, UT 23:32, z = 76.99° 7720 7730 7740 7750 0 2000 4000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 560 +/- 150 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0I (ph s 1m 2 m 1arcsec
work page 2007
Show all 21 references
-
[10]
× 1 0 4 ID 2, 31 DEC 2021, UT 00:14, z = 80.29° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 536 +/- 216 ? ~8732Å 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0I (ph s 1m 2 m 1arcsec
2007
-
[11]
× 1 0 4 ID 6, 02 JAN 2022, UT 00:57 z = 79.92° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 914 +/- 233 ? ~8732Å Fig. A.3. 7700 Å band airglow spectra continued. Apparent airglow spectra in 8700Å band. 8600 8650 8700 8750 8800 Wavelength...
2007
-
[12]
× 1 0 4 ID 7, 03 JAN 2022, UT 02:02 z = 78.18° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 434 +/- 242 ? ~8732Å 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0I (ph s 1m 2 m 1arcsec
2007
-
[13]
× 1 0 4 ID 11, 21 MAR 2022, UT 22:24, z = 77.32° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 510 +/- 199 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0I (ph s 1m 2 m 1arcsec
2007
-
[14]
× 1 0 4 ID 15, 01 SEP 2024, UT 22.25, z = 74.10° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 710 +/- 196 Fig. A.4. Apparent airglow spectra in 8700Å band continued. Article number, page 15 of 18 A&A proofs: manuscript no. aa53726-25 860...
2007
-
[15]
× 1 0 4 ID 16, 02 SEP 2024, UT 00:22, z = 77.13° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 887 +/- 222 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0I (ph s 1m 2 m 1arcsec
2007
-
[16]
× 1 0 4 ID 17, 02 SEP 2024, UT 01:46, z = 77.06° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 751 +/- 248 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0I (ph s 1m 2 m 1arcsec
2007
-
[17]
× 1 0 4 ID 18, 02 SEP 2024, UT 02:07, z = 71.10° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 563 +/- 194 Fig. A.5. Apparent airglow spectra in 8700Å band continued. 8600 8650 8700 8750 8800 Wavelength (Å) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5...
2007
-
[18]
× 1 0 4 ID 19, 02 SEP 2024, UT 02:29, z = 76.28° 8700 8720 87400 1000 2000 3000 Loo & Groenenboom (2007) Hanuschik (2003) Noise floor 618 +/- 255 Fig. A.6. Apparent airglow spectra in 8700Å band continued. Article number, page 16 of 18 J. K. M. Viuho et al.: The near infrared ...
2007
-
[19]
(2000) Noise floor 3200 +/- 2600 10400 10500 10600 10700 Wavelength (Å) 0.0 0.5 1.0 1.5I (ph s 1m 2 m 1arcsec
× 1 0 5 ID 4, 1 Jan 2022, UT 00:36, z = 80.14° 10500 10550 10600 0 20000 40000 Loo & Groenenboom (2007) Rousselot et al. (2000) Noise floor 3200 +/- 2600 10400 10500 10600 10700 Wavelength (Å) 0.0 0.5 1.0 1.5I (ph s 1m 2 m 1arcsec
2007
-
[20]
(2000) Noise floor 1900 +/- 2400 10400 10500 10600 10700 Wavelength (Å) 0.0 0.5 1.0 1.5I (ph s 1m 2 m 1arcsec
× 1 0 5 ID 8, 20 MAR 2022, UT 23:10, z = 79.53° 10500 10550 10600 0 20000 40000 Loo & Groenenboom (2007) Rousselot et al. (2000) Noise floor 1900 +/- 2400 10400 10500 10600 10700 Wavelength (Å) 0.0 0.5 1.0 1.5I (ph s 1m 2 m 1arcsec
2007
-
[21]
— — " — —
× 1 0 5 ID 9, 20 MAR 2022, UT 22:27, z = 77.10° 10500 10550 10600 0 20000 40000 Loo & Groenenboom (2007) Rousselot et al. (2000) Noise floor 1500 +/- 2400 Fig. A.7. Apparent airglow spectra in 10 500 Å band. Due to low system efficiency, the continuum in 10 500 Å is indistingu...
2007
-
[84]
: Academic) Lopez-Moreno, J
(San Diego, Calif. ;: Academic) Lopez-Moreno, J. J., Rodrigo, R., Moreno, F., Lopez-Puetas, M., & Molina, A. 1987, Planetary and Space Science, 35, 1029 Maihara, T., Iwamuro, F., Hall, D. N. B., et al. 1993a, in Infrared Detectors and Instrumentation, V ol. 1946 (International...
1987
Reviewed August 7, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.