GHOST aurora &ndash; continuum emission produced by hot N<sub>2</sub>

Dayton-Oxland, Rowan; Ball, Fiona; Whiter, Daniel; Samaddar, Srimoyee; Partamies, Noora; Barthelemy, Mathieu; Herlingshaw, Katie; Karvinen, Eero

doi:10.5194/egusphere-2025-5317

Preprints

https://doi.org/10.5194/egusphere-2025-5317

Preprints

16 Dec 2025

| 16 Dec 2025

Status: this preprint is open for discussion and under review for Annales Geophysicae (ANGEO).

GHOST aurora – continuum emission produced by hot N₂

Rowan Dayton-Oxland, Fiona Ball, Daniel Whiter, Srimoyee Samaddar, Noora Partamies, Mathieu Barthelemy, Katie Herlingshaw, and Eero Karvinen

Abstract. We investigate the origin of the continuum emissions observed in the poleward boundary dayside aurora discovered in Partamies et al. 2025, known as GHOST, and propose that they arise from highly excited, hot N₂. Using spectral modelling and fits to ground-based measurements of high-resolution GHOST spectra, we demonstrate that vibrationally and rotationally excited N₂ and N₂⁺ can reproduce the observed structured continuum without requiring emission from NO. Spectral fitting indicates that GHOST events coincide with extreme ion heating and high neutral temperatures. Background conditions from additional events indicate that strong ionospheric flows are typically present, which can help to provide the necessary energy input for producing hot neutral and ionised N₂. Proton aurora observations and EISCAT incoherent scatter radar measurements of ionospheric plasma parameters indicate that two of our three events are located in the cusp. These results suggest that the combination of strong flow, heating, particle precipitation, and cusp conditions produce thermally excited N₂ populations which can account for the continuum spectrum of GHOST.

Received: 27 Oct 2025 – Discussion started: 16 Dec 2025

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Rowan Dayton-Oxland, Fiona Ball, Daniel Whiter, Srimoyee Samaddar, Noora Partamies, Mathieu Barthelemy, Katie Herlingshaw, and Eero Karvinen

Status: open (until 27 Mar 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2025-5317', Anonymous Referee #1, 17 Mar 2026 reply
Overall, this is a very strong and impressive study, and I would like to emphasize the quality of both the dataset and the modelling effort. The manuscript provides a compelling framework to interpret complex auroral spectra, and the clarification that the “continuum” corresponds to a blend of band systems rather than a true blackbody is particularly useful, and offers a strong alternative to the NO continuum.
That said, several aspects of the interpretation would benefit from clearer justification and consolidation, especially regarding the physical conditions required for GHOST emissions.
Section 3 – interpretation of Event 1 and the definition of “continuum”
I compared the discussion of Event 1 from Partamies et al. (2025) with my own reading of their Figure 4, and I think this part would benefit from a more explicit comparison in the manuscript. The spectrum appears to show both: 1. Strong red emissions (630/636 nm), suggesting a significant contribution from low-energy electrons, and 2. clear N₂⁺ features (e.g. 427.8 nm, 470.9 nm), indicating that higher-energy electrons are also involved. A straightforward interpretation would therefore be a mixture of low- and high-energy electron precipitation. However, this does not immediately make it obvious that the spectrum should also be understood as a blend of multiple molecular band systems (e.g. N₂ IRA, N₂⁺, etc.), as discussed later in the manuscript. From a quick view, the reader may naturally interpret the spectrum in terms of electron energy distributions, and not grasping yet the idea about the role of the different molecular contributions in shaping the continuum-like appearance. It would therefore help to guide the reader more explicitly on this point, for instance by briefly revisiting this event (with the ASC image and spectrum) and clarifying how one moves from the “low/high energy mixture” interpretation to the “continuum as a blend of band systems”.
Section 4

Lines 144–146: the purpose of this paragraph is not entirely clear. Is Figure 3 intended to show the Hα emission in both the GHOST and background spectra? This could be clarified.

Figure 3: the Hα rest wavelength at 656.3 nm is quoted for air. Does this mean that the full spectrometer calibration is performed using the air wavelength convention?

Section 4.1

The authors may consider adding, perhaps in an appendix, the complete list of wavelengths/species/band included in the modelling. This would be valuable as a reference for future work.

Section 4.3 – modelling results and interpretation
First, I would like to note that the fitting results are excellent, especially given the large number of free parameters involved. That said, I still have several questions regarding the interpretation.
Line 242: it would be helpful to include an ASC image from these events, in order to compare them directly with Figure 1. The manuscript refers here to a “normal aurora”, and a visual comparison would be useful.

Figure 5 (top), GHOST spectrum: I would strongly suggest removing the N₂ VK curve from the plot. It took me some time to distinguish it from the N₂ 1P contribution, and since N₂ VK does not seem to contribute significantly here (unless I misunderstood), its inclusion may unnecessarily complicate the figure. By contrast, the authors emphasize that N₂⁺ M is important for shaping the spectrum. It would therefore help the reader if some specific spectral examples were pointed out explicitly, for example around 737.0, 739.3, and 740.1 nm.

Figure 5 (bottom): the N₂ IRA contribution has indeed increased compared with Figure 4 (bottom), especially above 732 nm, although it already seemed to be present at somewhat shorter wavelengths in Figure 4 as well. I am less convinced, however, by the interpretation of the role of N₂⁺ M in distinguishing Figure 5 from Figure 4, since this component also appeared important in Figure 4 (bottom).

More generally, I am not entirely convinced that the spectrum in Figure 5 (bottom) represents a clear transition stage between GHOST and “normal” aurora. To my eye, it still looks rather similar to Figure 4 (bottom). Since the N₂ 1P bands are less pronounced and likely affected by lower SNR, is it possible that the fitting procedure compensates by assigning a stronger contribution to N₂ IRA and N₂⁺ M?

Line 260: when referring to “the continuum fit”, do the authors mean the spectrum shown in Figure 5 (top), i.e. the GHOST case? This should be stated unambiguously.

Lines 264–267: I found this part difficult to follow. My understanding is that the best fit for Figure 5 (top) was obtained with a rotational temperature of 881 K, but the manuscript then states that this is not sufficient to thermally excite N₂ IRA. Yet the model still includes some N₂ IRA signal in both the top and bottom panels of Figure 5. Does this imply that another mechanism is contributing? Please elaborate.

"The high rotational temperatures generally flatten all band systems, which would make the spectrum more continuum-like at all wavelengths". I am not sure to follow. Indeed we should expect a more continuum-like at all wavelengths, but this is not what the figure 5 (top and bottom) are showing, is it?

Since the authors already present a HITIES spectrum around Hα in Figure 3, could the model also be applied to this spectral range? Would this help constrain the N₂ bands, or perhaps provide additional information about N₂ IRA? Even if it is outside the main scope of the paper, a short comment on this would be interesting.

Line 295: could the authors specify the typical electron energies discussed here? I would have expected the IMF orientation to influence primarily the flux and access conditions of magnetosheath electrons, not necessarily their characteristic precipitation energy. Some clarification would be useful.

Is the code publicly available, or only available upon request?

Can also the code be used for a different part of the spectral range (not only Ha region) ? If so, did the authors try to degrade the spectral resolution of the synthetic spectra to fit eight MISS or T-Rex spectra, or another instrument? This model is particularly valuable and should probably be interesting for the community.
Sections 5–6 – mechanism and occurrence of GHOST
The proposed mechanism is physically plausible and well motivated, but the different contributing conditions are somewhat dispersed throughout the text. It would therefore help to summarize these conditions more explicitly, for instance in a table or within the schematic/cartoon already presented, distinguishing the main ingredients (ionospheric priming, dayside activity, sunlit conditions, convection/heating).
In addition to the convection patterns, the manuscript identifies other important conditions, such as an already energized dayside ionosphere (proton and low-energy electron precipitation) and a sunlit atmosphere enhancing both ionization and resonant scattering (notably for N₂⁺).
Taken together, these suggest that GHOST emissions require a combination of conditions. However, each of these ingredients seems, at least individually, relatively common in the cusp. This raises a key question of their actual joint occurrence: are these conditions truly rare when combined, or should one expect GHOST-like emissions to occur more frequently? A short discussion on this point would significantly strengthen the interpretation.
Line 343: the statement on ion upwelling driven by low-energy precipitation would benefit from clarification and/or references. It may help to better distinguish between convection-driven heating/upwelling and subsequent excitation by soft electrons.

Minor comments:
Line 67: NO instead of NO₂?

Figure 1: please add the date of the event.

Section 3.3: please clarify the HITIES FoV on the ASC image.

Figure 2 (09:24:01 UT): are the whitish features near zenith expected?

Reply
Citation: https://doi.org/10.5194/egusphere-2025-5317-RC1
RC2: 'Comment on egusphere-2025-5317', Anonymous Referee #2, 22 Mar 2026 reply

The paper under review investigates the origin of emissions observed in the poleward boundary dayside aurora that was first reported by Partamies et al. (2025). The authors propose that the emissions arise from highly excited, hot N2. Using spectral modelling and fits to ground-based measurements of high-resolution GHOST spectra, the authors claim that vibrationally and rotationally excited N2 and N2+ can reproduce observed emissions without requiring the presence of NO, as had been suggested in prior published work. Spectral fitting using simulated annealing indicates that GHOST events coincide with extreme ion heating and high neutral temperatures. Background conditions suggest that strong ionospheric flows are typically present and can help provide the necessary environment for producing hot N2 and N2+. The authors conclude that a combination of strong flow, heating, particle precipitation, and cusp conditions produce thermally excited N2 populations that may account for the occurrence of GHOST emissions.
I find the paper to be interesting, well-written, and thorough, in most respects. The figures are certainly interesting and help the reader to understand both the observations and the methodology used to interpret them. The bibliography is mostly complete, though there is prior related work that the authors are either unaware of or may not consider to be relevant. With some additional information and clarifications, I would consider the manuscript suitable for publication. I hope my comments are viewed as helpful and constructive.
Section 1 Introduction
Lines 37-39: The N2 VK and N2+ 1NG (first negative) band systems excited by energetic particle precipitation can produce emission in the wavelength region 502-507 nm. What evidence can the authors provide to support their claim that there is no auroral emission from these band systems for the three events considered in their study?
Lines 51-55: The language here is imprecise and difficult to interpret. Perhaps the authors mean “…with extreme ion drift velocity and electron temperature in the highest 5% and 3%, respectively, compared with non-STEVE SAID events, corresponding to about 3 km/s and 4000–10000 K, respectively…”
Section 4 Spectral identification
This section is quite important as it describes the model used to fit the observed spectra. However, there are missing details that prevent repeatability and, more importantly, are required to support assumptions made in the fitting methodology. For example, the authors state, “We model only the shape of the emission spectrum for each species based on temperature, and do not need to model specific atmospheric conditions since the fit to data leaves the intensities of each molecular vibrational band as free parameters.” First, the atmospheric conditions can only be ignored if such conditions do not affect the shape of the emission spectrum. At lines 213-214 the authors indicate that intensities are corrected to account for water vapour absorption. Presumably this is a wavelength dependent correction. If so, the shape of the emission spectrum is indeed adjusted to account for atmospheric conditions, correct? Second, the authors state they have “checked that the high resolution spectra of NO2…does not show any specific coincidence Hα HITIES spectra…” but they do not show these checks in the manuscript. These comparisons should be included in the manuscript.
Section 4.1 Synthetic spectra
Line 154: Please explain in the text why the N2 2PG (second positive group) and N2+ 1NG band systems were not included in the spectral modeling.
Section 4.2 Fit to observations
There are some missing details in this section that make repeatability of the methodology difficult, at best. Specifically, the authors use a simulated annealing technique because it can escape local minima to find the global minimum. However, this is not guaranteed. The authors chose limits of 90 iterations or a total of 3000 iterations, and then the whole process is repeated 3 times. What is the justification for these limits (90, 3000, 3)? The authors state that details of the technique are described by Price (2021), but in a different context. How different is the context? Have the authors performed any simulations to demonstrate that they can retrieve the original state of free parameters used to produce the simulated spectra with an appropriate instrument sensitivity (i.e., random variability), along slit calibration differences, atmospheric conditions (i.e., absorption by water vapour), etc.? The authors’ claim that the global minimum can necessarily be found would be more convincing if demonstrated using simulated observations. This is standard practice when using fitting routines to retrieve geophysical parameters.
Line 209: Please include in the text a justification for assuming that the N2+ and O2+ bands have the same rotational temperature. Can the authors cite any laboratory measurements to support their claim?
Lines 213-214: Are just atomic line intensities corrected to account for water vapour absorption or are all atomic and molecular lines corrected to account for water vapour absorption? Please clarify the text.
Lines 217-218: Isn't there a degeneracy between the Gaussian width and the rotational temperature? Did the authors consider determining the width of Gaussians as a preprocessing step to avoid any degeneracy with rotational temperature? Perhaps the simulated annealing process does effectively determine the Gaussian width based on atomic lines, but that would require that the atomic lines are always detectable in the spectra. Please provide additional details that demonstrate there is no degeneracy between Gaussian width and rotational temperature.
Section 4.3 Results
The quality of the fits to observed spectra reported in this section is difficult to assess since the authors do not provide measurement uncertainties in Figs. 4 & 5. These measurement uncertainties are important since they drive the minimization of the weighted RMS of the residuals. Please add representative error bars to the black plusses in the figures showing observed spectra. Another issue I have with this section is that the authors do not provide their fitted free parameters for each fitted spectrum. These parameters ultimately provide the basis for the authors’ conclusions, but the reader has no access to these values. These values should be provided in tables. While the authors briefly discuss the fitted rotational temperatures they obtained, they provide no details on the fitted vibrational populations. These values may provide useful information on the excitation mechanism. For example, how do the fitted vibrational populations compare with theoretical Franck-Condon factors? Are the fitted vibrational populations even physically realistic? Have the authors compared their results with published laboratory measurements of the vibrational populations for any of the important band systems (e.g., N2 1P, N2+ Meinel).
Lines 252-253: The statement that there is no indication of the presence of NO is dubious since the authors do not include NO2 band systems in their fit. The statement would be more convincing if they included NO2 in their fit and the fit parameters indicate no detectable presence of NO.
Lines 254-257: Why do the authors conclude that the increased contribution from N2 IRA and N2+ Meinel could be “continuum-like” emission when they acknowledge that “normal” N2 aurora is present and N2 IRA and N2+ Meinel band systems are known to be produced in “normal” N2 aurora (Cartwright et al., 1975; doi.org/10.1029/JA080i004p00651; Gattinger and Jones, 1974; doi.org/10.1139/p74-305).
Lines 258-260: What is the source of the temperature quoted by the authors? Is it from their fits or from uncited published work?
Lines 265-267: Are the authors claiming that they observed emission coming from the E-region? If so, how do they know the altitude of the emission? Please provide a figure showing how the structure of the band systems would change with rotational temperature. The statement that a high rotational temperature spectrum would be more continuum-like is confusing. A high rotational temperature would indeed make each band appear spectrally wider, but it would not affect the band intensities and therefore the overall “shape” of a band system would be preserved. Changing the rotational temperature is similar to observing a spectrum with two instruments, one with high and one with low spectral resolution.
Section 5 Background conditions
There are several statements in this section relating GHOST event broadness and thickness to favourability of prevailing conditions. How do broadness and thickness relate quantitatively to favourable conditions? What determines the broadness and thickness of the events? Could the broadness and thickness simply be due to observational bias (i.e., geometrical effects)?
Section 6 Proposed mechanism
The language used by the authors to describe their proposed mechanism for GHOST events is rather confusing to me. The authors suggest that when precipitation collides with an ionosphere primed with upwelling, heating, and excitation, a continuum made up of electronically and vibrationally excited N2, N+2 and O+2 is produced. Note that precipitation colliding with an ionosphere (primed as described by the authors or not) will produce rovibrational emission if the energy of the particles is higher than the excitation threshold. How is the scenario proposed by the authors truly different from “normal” aurora? Isn’t “normal” aurora defined to be a visual manifestation of a continuum made up of electronically and vibrationally excited N2, N2+ and O2+ (i.e., for the wavelength range considered by the authors)? The authors may or may not be aware of results reported by Ellingsen et al. (2021; doi.org/10.5194/angeo-39-849-2021) who analyzed rocket observations of sunlight aurora in the cusp region and emission from N2+ in an environment with ion and neutral upwelling. The relevance and significance of this recent work should be discussed.
Section 7 Conclusions
As noted above, the language used by the authors in their concluding remarks is confusing and needs clarification. It’s not entirely clear to me what the authors mean by their statement “The infra-red afterglow and Meinel bands have not been considered in previous Studies…” These emissions (as well as a continuum emission) were certainly considered and discussed in the aforementioned work by Gattinger and Jones (1974). Indeed, I’m surprised that this work was not cited by the authors.
An additional source of confusion is the authors’ statement “We propose that the combination of upwelling, heating, and shear flow provides the energy needed to excite the N2, N2+, and O2+ to produce the observed pseudo-continuum.” There is no mention of particle precipitation as a source of energy in this statement. Are the authors suggesting that particle precipitation only plays a role in priming the ionosphere and has no direct role in exciting N2, N2+, and O2+? If this is the argument, and absent the “normal” sources of auroral excitation - energetic particles (possibly with solar illuminations) - how are upwelling, heating, and shear flow able to exceed the excitation threshold of N2, N2+, and O2+ (at least 8 eV) resulting in continuum emission?
Minor comments:
Lines 93-94, 121: Please provide transitions and wavelengths for the emissions for the benefit of readers who are not familiar with these details.
Figures 4 & 5: Discrimination of the colors in the figures will be difficult for color blind people. The use of symbols or different line styles would make the figures much easier to interpret.

Reply

Citation: https://doi.org/10.5194/egusphere-2025-5317-RC2

Rowan Dayton-Oxland, Fiona Ball, Daniel Whiter, Srimoyee Samaddar, Noora Partamies, Mathieu Barthelemy, Katie Herlingshaw, and Eero Karvinen

Viewed

Total article views: 623 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
377	211	35	623	35	58

HTML: 377
PDF: 211
XML: 35
Total: 623
BibTeX: 35
EndNote: 58

Views and downloads (calculated since 16 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	107	55	13	175
Jan 2026	132	49	5	186
Feb 2026	42	45	3	90
Mar 2026	96	62	14	172

Cumulative views and downloads (calculated since 16 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	107	55	13	175
Jan 2026	132	49	5	186
Feb 2026	42	45	3	90
Mar 2026	96	62	14	172

Viewed (geographical distribution)

Total article views: 560 (including HTML, PDF, and XML) Thereof 560 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 28 Mar 2026

Short summary

GHOST is a pale mauve aurora, usually seen near noon in Svalbard during polar night, emitting a continuum spectrum across the visible range. Our fitting of high-resolution spectra shows that GHOST emission is best explained by extremely hot nitrogen and oxygen molecules, not by the more scarce nitrous oxide which is usually assumed. We propose that strong ionospheric flows, daylight, and incoming solar‑wind particles combine to heat the upper atmosphere, priming it to produce GHOST aurora.


Total:	0
HTML:	0
PDF:	0
XML:	0

GHOST aurora – continuum emission produced by hot N2

Viewed

Viewed (geographical distribution)

GHOST aurora – continuum emission produced by hot N₂