the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Sep 2026
First simultaneous retrieval of O2 A-band nightglow volume emission rate and temperature in the MLT from Tianlu-1/HDI hyperspectral observations
Abstract. The O₂ A-band nightglow contains complementary information on volume emission rate (VER) and temperature in the mesosphere and lower thermosphere (MLT), yet their simultaneous retrieval remains challenging because it requires detecting weak airglow emissions while resolving temperature-sensitive rotational structures. The Hyperspectral Atmospheric Density Imager (HDI), launched aboard Tianlu-1 in January 2025, is China’s first spaceborne limb-sounding payload based on spatial heterodyne spectroscopy and provides hyperspectral limb observations of the O₂ A-band nightglow at a spectral resolution as fine as 0.0394 nm, enabling detailed characterization of its rotational structure. Here, we present a joint algorithm for simultaneous nighttime VER and temperature retrieval and demonstrate its performance using early on-orbit HDI measurements. Results show stable simultaneous retrievals during selected nighttime windows over 86–98 km. Retrieved VER distributions are consistent with the expected primary O₂ A-band emission layer. HDI temperatures agree with collocated, window-averaged SABER observations, with monthly mean absolute differences below 3.82 K. Monthly mean posterior 1σ uncertainties are below 5.09 % for VER and 3.51 K for temperature, with monthly mean temperature degrees of freedom for signal exceeding 0.946. These results demonstrate, for the first time, the simultaneous retrieval of nighttime VER and temperature from individual hyperspectral measurements, providing intrinsically co-registered and internally consistent constraints for subsequent photochemical retrievals of atomic oxygen. This advance establishes a new spaceborne capability for characterizing the coupled thermal–photochemical state of the MLT and investigating interactions among atmospheric chemistry, radiation, and dynamics in this region.
- Preprint
(15911 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 20 Nov 2026)
- RC1: 'Comment on egusphere-2026-5511', Anonymous Referee #1, 27 Oct 2026 reply
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 372 | 137 | 57 | 566 | 68 | 56 |
- HTML: 372
- PDF: 137
- XML: 57
- Total: 566
- BibTeX: 68
- EndNote: 56
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
The paper presents simultaneous retrievals of O2 A-band nightglow VER and temperature from the HDI spatial heterodyne limb sounder on Tianlu-1, and compares the temperatures with SABER. Getting both quantities from the same hyperspectral limb frame is a useful result. My main concerns are whether the night-time screening adequately excludes solar-excited emission along the viewing path, how the retrieval uncertainties are calculated, and how closely the HDI and SABER samples are matched. The main points below need to be addressed to support the uncertainty estimates and clarify how robust the retrievals are. Beyond that, a few statements and numerical summaries need more careful wording.
While the manuscript is relevant to ACP, I would consider it more closely aligned with the scope of AMT, as its main contribution lies in demonstrating and characterizing a new measurement capability rather than providing new insights into atmospheric processes.
Specific comments
1. Is SZA > 100° strict enough? (L124-126, Fig. 2, Sect. 3.2)
Where is SZA evaluated, and at which tangent altitude? SZA > 100° can put the tangent point in shadow while higher parts of the line of sight remain sunlit. For a 98 km tangent ray, the path through 150 km can remain sunlit up to a tangent-point SZA of about 110°, depending on viewing direction. Sunlit parts of the LOS can have a big measurement contribution due to photolysis excitation being significantly stronger than just Barth-process, even higher up where the density is significantly smaller. Could solar-excited emission affect the retrievals?
Please show the SZA distribution per hemisphere and window, and check illumination along the emitting path for Figs. 2-4. A comparison using stricter cuts, e.g. 105° and 110°, or a direct estimate of the sunlit contribution would help. Report the remaining sample counts and whether the VER, temperature and SABER differences change. These cuts are sensitivity tests, not guaranteed darkness thresholds.
The cited precedent (Sun et al., 2022, Sect. 4.2) keeps observations with SZA within 100° to remove night-time data from a SCIAMACHY daytime climatology. That supports 100° as a day-side cut, not as evidence that the HDI line of sight is dark. Sun et al. suggest horizontal airglow gradients as a possible cause of retrieval problems near the terminator, though their examples concern a different band and altitude range. Since the HDI forward model assumes a spherically symmetric, layered atmosphere (L102), please discuss whether such gradients near the terminator affect the retrievals.
Do the latitude-edge enhancements in Sect. 3.2 coincide with the SZA cut? Please discuss illumination as a possible cause if relevant.
Also, does the Earth-shadow operating description at L149-150 apply throughout? Figs. 3 and 4 imply SZAs of about 146° and 101°. At 146° the satellite must have been in the Earth’s shadow while observing. Please clarify how the operating description fits these cases.
2. SABER comparison and sampling (Sect. 3.3, Figs. 3-4)
Are HDI and SABER matched in longitude and local time, or only averaged into the same weekly latitude bins? Tides and waves matter here, especially with the two-hour gap in Fig. 4. Please give the SABER version, quality and day/night filters, matching criteria and sample coverage, and explain any differences in vertical response. If close matches are too sparse, please be clear about what agreement between the averaged fields tells us about individual-profile accuracy.
3. Put a number on the signed differences (L373-386)
Alternating positive and negative differences do not show that the mean offset is small. Please report the mean signed difference and standard deviation per window and hemisphere, with counts and weighting. Say whether these compare grid-cell means or matched profiles. Differences by altitude would also help show offsets that cancel in the overall mean. Their spread should not be confused with uncertainty in the mean or attributed entirely to HDI.
4. Retrieval settings and extra fitted parameters (L129-138, Sect. 2.4)
The given uncertainties and DOFS depend on Sa and Se, which are not provided.
Please give the priors, uncertainties and correlations for the fitted parameters. How is measurement covariance estimated? Is the VER a priori only used as a priori or also used as a constraint weighted in Sa? If it is, the measurement is used twice, affecting posterior uncertainties.
For wavelength shift and ILS width, show typical fitted values and state whether they are shared across heights. How sensitive are VER and temperature to these parameters and to the assumed O2 density?
5. Emission above the retrieval range (Sect. 2.3, Sect. 3.2)
The rays also receive emission from above 98 km, where the reported retrieval ends. How is this included in the forward model? Please check sensitivity to the assumed upper-atmosphere profiles, especially where NH VER increases right up to 98 km.
6. Improvement relative to the a priori (L259-261)
Fig. 3 suggests that HDI is closer to SABER than MSIS is. Does this hold across the dataset? Comparing HDI-SABER and MSIS-SABER signed means and MAE or RMSE on the same samples would help show what the measurements add. Otherwise, please limit this claim to the example.
7. Context for the posterior uncertainties (Sect. 3.4)
Make clear that 5.09% and 3.51 K are window means across both hemispheres, not limits for individual retrievals, and note that the posterior 1σ is measurement precision only, excluding systematic errors.
How do the HDI uncertainties compare with the OSIRIS and MIGHTI results already mentioned? Please compare similar conditions, resolution and averaging, and distinguish posterior precision from systematic errors and differences between instruments.
8. Claims based on uncertainty and DOFS (L410-411, L435-436, L453-460)
Small posterior errors and high DOFS are useful evidence, but depend on the model and assumed covariances. Please soften the stronger claims and check sensitivity to reasonable changes in the priors and their uncertainties. Which calibration, pointing, background and model errors are included? The SABER comparison helps evaluate temperature, but does not by itself validate VER uncertainty.
9. Vertical sampling and resolution (L117-119, Fig. 5 caption)
The retrieval spacing is about 2 km, but Fig. 5 looks smoother. Is this plotting interpolation? Please distinguish grid spacing from actual vertical resolution and show representative averaging kernels or a resolution estimate. The kernel diagonals alone do not give the response width.
10. Number of decimal places (L250, L256-257, abstract and throughout)
Values such as 169.33 K and 2775.67 photons cm⁻³ s⁻¹ seem unnecessarily precise. Please round consistently with the uncertainties. Mean differences and DOFS can reasonably use different precision from individual temperatures.
11. Data availability (L479–480, L486–488)
The manuscript suggests that HDI can serve as a complementary source where SABER sampling is limited. For that, the data must be accessible to other users, but currently they are available only on request. Please consider depositing at least the retrieved temperature and VER profiles, with uncertainties and geolocation, in a public repository, even if the Level-1 data must remain restricted. If this is not possible, please state the restriction in the Data availability section and qualify the outlook at L479–484 accordingly.
Technical corrections
- Fig. 1: please add an intensity scale and units to the interferogram inset.
- Fig. 6: HDI - SABER might be easier to read, but either sign convention is fine if used consistently.
- Abstract: distinguish the unapodized 0.0394 nm resolution from the 0.057 nm used after apodization.
- Abstract and conclusion: call the one-week results window means, not monthly means. Avoid strict "below" or "above" wording where the displayed value equals the threshold.
- References: please cite the final version of Sun et al. (2022), Atmos. Meas. Tech., 15, 3721-3745, https://doi.org/10.5194/amt-15-3721-2022.