A statistical study of the O<sub>2</sub>Atm(0-0) aurora observed by the Swedish satellite MATS

Pérez-Coll Jiménez, Judit; Ivchenko, Nickolay; Lindstein, Ceona; Krasauskas, Lukas; Hedin, Jonas; Murtagh, Donal Patrick; Megner, Linda; Linder, Björn; Gumbel, Jörg

doi:https://doi.org/10.5194/egusphere-2025-2324

Preprints

https://doi.org/10.5194/egusphere-2025-2324

Preprints

03 Jun 2025

| 03 Jun 2025

A statistical study of the O₂Atm(0-0) aurora observed by the Swedish satellite MATS

Judit Pérez-Coll Jiménez, Nickolay Ivchenko, Ceona Lindstein, Lukas Krasauskas, Jonas Hedin, Donal Patrick Murtagh, Linda Megner, Björn Linder, and Jörg Gumbel

Abstract. This study conducts a statistical analysis of the aurora observed by the Swedish satellite MATS. MATS' main instrument is a telescope that performs limb imaging at six different wavelengths intervals, among them the 762 nm wavelength emission in the O₂ atmospheric band. This emission, even though it can not be observed from the ground, plays a big role in the study of atmospheric airglow and aurora. Here, some auroral properties of this emission, such as peak altitude, geomagnetic location, and auroral intensity, are examined and compared to the global auroral activity indicator known as Kp-index. A total of 378 events are analyzed. An average geomagnetic latitude of 67.7° is found in both hemispheres, and an average peak altitude of 103 km is obtained. The peak altitude shows dependence on the magnetic local time. Auroral intensities of the order of 10²–10³ kR are observed.

Received: 19 May 2025 – Discussion started: 03 Jun 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

Judit Pérez-Coll Jiménez, Nickolay Ivchenko, Ceona Lindstein, Lukas Krasauskas, Jonas Hedin, Donal Patrick Murtagh, Linda Megner, Björn Linder, and Jörg Gumbel

Status: final response (author comments only)

RC1:
'Comment on egusphere-2025-2324', Anonymous Referee #1, 07 Jul 2025

Summary
This manuscript presents a statistical study of O2Atm(0-0) auroral emissions at 762 nm as observed by the MATS satellite during three months of operations from February to April 2023. The Authors identified auroral signatures in the limb images taken along the track of the satellite by developing a detection algorithm isolating them from the O2 airglow signatures. They visually inspected all the auroral events and determined the geomagnetic coordinates (geomagnetic latitude and magnetic local time, MLT) of the corresponding tangent point as well as the altitude of the emission peak for the 378 of them which were retained for the statistical analysis. They then analysed the distributions of the auroral events as a function of geomagnetic latitude, MLT, and geomagnetic activity as measured with the Kp index. They also examined the dependence of the emission peak altitude with MLT and Kp index, as well as the distribution of spectral intensity as a function of peak altitude. The main findings are that the auroral events occur at lower geomagnetic latitude with increasing Kp, that the peak altitude is around 103 km for most conditions, and that the spectral intensity tends to be lower when the peak is found at a higher altitude.
The manuscript is well written, the methodology is carefully detailed, and the topic of the study is very suitable for ANGEO Communicates. However, the statistical analysis is relatively superficial, and it is hence difficult to identify significant new results provided by the study. I believe that the manuscript could become suitable for publication if the following comments and suggestions are given consideration.

Major comment
As stated above, currently the main results presented in the manuscript do not really bring much novel scientific insight in O2 auroral emissions. The fact that the auroral oval expands with increasing geomagnetic activity is very well-known and documented; the 762 nm aurora peak altitude and brightness are found to be similar to those derived in past studies. However, with the unique new dataset provided by the MATS mission, and given the scarcity of the literature on O2 auroral emissions, there is a lot of room for making new findings. I may suggest a few avenues that the Authors could consider to get there, some of which would require expanding slightly the statistical treatment of the data, while others could in fact mostly rely on surveying the existing literature:
* Could the altitude thickness of the auroral emissions be evaluated (using e.g. the full width at half maximum as the metric) and be a parameter studied as a function of MLT, geomagnetic latitude, Kp index, spectral intensity, etc.?
* Could another geomagnetic index than Kp be considered – for instance an auroral electrojet such as AE or SME? These might be more insightful than Kp if looking at the properties of the aurora such as peak altitude and brightness, as they respond specifically to auroral/substorm activity.
* Rather than a mere qualitative assessment that the O2 762 nm auroral emission geomagnetic latitude decreases with increasing Kp index, could a parametrisation of this latitude as a function of Kp (or other index) be derived? It could then be discussed with respect to the latitudinal dependence of other auroral emission lines on geomagnetic activity, enabling one for instance to determine whether O2 762 nm occurs in a specific part of the auroral oval and is associated with a certain type of particle precipitation.
* Can any interhemispheric asymmetries be noted when the events do not take place close to the equinox?
* How do the results – in terms of O2 762 nm auroral emission altitude, MLT and geomagnetic latitude distribution, brightness – compare to what is known about other auroral emission lines?
* Would it be insightful to look into the cross-oval extent of the detected auroral events (again as a function of MLT, geomagnetic latitude, geomagnetic activity)? Could this be estimated by evaluating the width of the parabola associated with the signature of auroral emissions in keograms?
* Is it possible to get additional information by analysing the full images rather than keograms? In Fig. 1b–c, there seems to be some structuring of the auroral emissions visible along the horizontal direction of the images. Making use of the 2D nature of the MATS data could lead to findings that were not possible from rocket measurements, which are intrinsically 1D.
To clarify: I do not request that the Authors look into all of those questions, of course. But I think if at least one of those ideas – or another one not from this list – could be addressed, it might bring in novelty that would radically enhance the impact of the paper and make it worthy of prompt publication.

Minor comments
– l. 4–5 (abstract), “This emission (...) plays a big role in the study of atmospheric airglow and aurora”: This statement is quite vague; it could be worth giving a concrete example of what role this emission plays.
– l. 13: If possible, please provide a reference.
– l. 16: See also Kirillov & Belakhovsky (2021) for a recent work on this topic.
– l. 44–45, “and the monthly dependence of the auroral altitude”: I did not find which part of the results section addresses this point.
– l. 49: Generally, the term “lower thermosphere” is used instead of “low thermosphere” in the literature; please consider adopting it.
– l. 77: Would it be possible to explain why a third-order polynomial was chosen for the regression in the removal of the airglow contribution? Especially in the example selected for Fig. 1, it seems that the fitted background in panel f) is very flat; it this typical? What is the reasoning behind considering a third-order polynomial in the general case?
– l. 85: Please indicate (here or later in the section) the value of the retained threshold, to ensure reproducibility of the results.
– Figure 1: Please add axis labels in panels e) and f), as well as a colour bar for the data shown in panels a–e). Please define also ‘TPlat’ explicitly (for instance in the caption). You may also consider adding in panel e) a y-axis indicating the altitude of the tangent point corresponding to the pixel row numbers.
– Fig. 1 caption: One of the ‘(ii.)’s should be ‘(iii.)’.
– l. 101: Would it be possible to comment on the fact that 378 events were retained for the study, while a back-of-the-envelope calculation suggests that, during the ~81 days of MATS data used in this study, there have been approximately 4860 auroral oval crossings? Is it so that O2 762 nm auroral emission is not always present in the auroral oval? Are there limitations in the instrument’s operations related to e.g. lighting conditions in the atmosphere? Are the auroral signatures very often too complex for the events to be retained by the algorithm? It would be interesting to see the temporal distribution (i.e. as a function of the date in early 2023) of the obtained auroral events, for each hemisphere.
– l. 113: How representative are the statistics in the cases where only three events are in a given data bin? Would it remove many data points if selecting a higher threshold for calculating the mean and standard error of the mean?
– l. 117–118, “A notable feature of this plot is the lower altitude of events between 5 and 7 MLT, especially in the southern hemisphere”: It seems to me that it is in fact only the case in the southern hemisphere, as the few northern-hemisphere data points are at altitudes very closed to the average values. Please correct the statement.
– Figure 2: The chosen colour map is not adequate, as it is not accessible to people with colour vision deficiencies. Please refer to the ANGEO guidelines to revise the figure (https://www.annales-geophysicae.net/submission.html#figurestables), and consider using a suggested tool such as Coblis.
– Figure 3: Would it be possible to provide a measure of the uncertainty on the average altitude values in panel a)? Besides, panel b) is missing its y-axis label (name of the plotted parameter).
– Fig. 3 caption: “One-hour average” suggests a temporal average (e.g. from a time series), but here I think you are referring to the average of events occurring within the same 1-hour MLT bin. Please consider rephrasing to avoid ambiguity. In addition, please indicate whether the boundaries of the Kp bins are included or excluded (i.e., is the first bin from Kp = 0 to Kp = 3– or to Kp = 3? If the latter, then I presume that the second bin starts at Kp = 3+).
– l. 124: If using the phrase “clear correlation”, please calculate a relevant correlation coefficient as part of the data analysis.
– l. 127–128: The statement about the auroral intensities expressed in kR is difficult to verify by looking at the figures. Would it make sense to present the spectral intensities shown in Fig. 3b directly as limb brightness using the conversion described in l. 121–122, for instance?
– l. 131: Earlier (l. 114), it read ‘104 km’; please harmonise.
– l. 134–135, “For MLTs of 5 to 7, the average altitude of the peak tends to be below average”: As mentioned above, this seems to only be the case for the southern hemisphere; please rephrase.
– l. 135–136: More precisely than energetic electrons, in the morning sector, the role of pulsating aurora has been emphasised in producing lower auroral peak altitudes compared to the evening and midnight sectors (see e.g. Partamies et al., 2022, on the 557.7 nm and 427.8 nm aurora).
– l. 137: ‘left panel’ --> ‘right panel’
– l. 137–138: The reported general trend between peak altitude and brightness is not at all obvious from Fig. 3b. Further statistical processing and a revised figure would be necessary to make an assessment on this matter.
– l. 138–139: The reference to Cattell et al. (2006) does not seem optimal, as the statement it is meant to support is not at all the focus of the cited paper. In fact, the statement does not necessarily hold – see for instance Fig. 5 of Tesfaw et al. (2022), where it is clear that energy flux and characteristic electron energy are not always following the same trend. Please revise this sentence.
– l. 142–143: Would it be possible to provide an estimate of how often events with a complex spatial distribution of the aurora may have been missed? It would be interesting to know for instance if the O2 762 nm aurora generally consists of a single arc or if multiple structures can be seen during a single oval crossing. If only single arcs have been retained for the study due to the event selection algorithm, this may induce a bias in the results, which would be worth evaluating and discussing in more detail.
– Although it considered a different emission line of O2 (1.27 µm), it would be worth referring to the recent study by Gao et al. (2020) using 18 years of SABER data, since their methods are adjacent to yours, and it would prove insightful to discuss how your results compare to theirs. So little has been published on O2 auroral emissions that it would be worth mentioning the more recent literature addressing it.

Cited references
– Gao et al. (2020), https://doi.org/10.1029/2020JA028302
– Kirillov & Belakhovsky (2021), https://doi.org/10.1029/2020JD033177
– Partamies et al. (2022), https://doi.org/10.5194/angeo-40-605-2022
– Tesfaw et al. (2022), https://doi.org/10.1029/2021JA029880

Citation: https://doi.org/10.5194/egusphere-2025-2324-RC1
- AC1:
  'Reply on RC1', Judit Pérez-Coll Jiménez, 13 Aug 2025
  Major comment
  Thank you for your valuable suggestions on how to increase the quality of our study. After carefully considering all the suggestions, we have decided to implement two of them.
  First, we have made plots similar to Figure 2 using the AE and SME index (see Figures 1 and 2 attached). We agree with the reviewer that these geomagnetic indices are more insightful when studying the aurora. Thus, we have decided to substitute the Kp index with the SME index, as it is a better representative of the auroral current strength.
  Next, we have fitted the latitude of the auroral events as a function of the kp-index, AE-index, and SME index (see attached Figure 3). We plan to include two new panels in Figure 2 with the fit for both hemispheres and compare the results with auroral oval position models, e.g. ovation. Like above, we will keep the plots with the SME index.
  These two modifications will be incorporated into the revised manuscript. As for the other suggestions, we explain below the reasons for not fully implementing them for this particular manuscript.
  Altitude extent of aurora. Evaluating the vertical extent of the auroral emissions could be considered for future publications. However, there are a couple of concerns that make it challenging to do for this manuscript. First, to evaluate the full width at half maximum, we rely on a precise 0 value for the background. The challenge of removing the background is the presence of the airglow and distinguishing it from the aurora. The paper presents how this is done in our case, and in combination with the threshold, this gives an objective way of selecting clear events of the aurora. The overlap of aurora and airglow below the peak affects where exactly the “half maximum” is located, and any uncertainty of the background (airglow) subtraction will contribute to the uncertainty of the altitude extent. In addition, the vertical extent of our images is limited and, even though the maximum altitude of emission is generally inside the field of view, in many cases the brightness of the emissions at the top of the image may still be above the half of the peak brightness, making it impossible to evaluate the vertical extent of the auroral layer in these cases.
  
  Interhemispheric asymmetries. MATS’s orbital parameters combined with the configuration of the magnetic field introduce a significant amount of biases to studying interhemispheric asymmetries, as the satellite does not have symmetrical passes in both hemispheres. For this reason, evaluating any interhemispheric features would require a separate study (and preferably a satellite in a non-sunsynchronous orbit - covering a range of local times), which is beyond the scope of this paper.
  
  Comparison between O2 762 nm and other emissions. While it would be good to have a full comparison of different parameters among several emissions, we believe that only studies conducted by limb imaging would be significant for our work. These, however, are very scarce. In addition, we are aiming to publish a short communication (letter), which limits how much we can add to the current manuscript. We will, however, search for studies on other auroral emissions measured by limb imaging and, if we are successful, we will include a short comparison of the relevant parameters.
  
  Cross-oval extent of auroral events. In the future, a three-dimensional tomographic reconstruction of the events is planned, which would provide a full approach to this problem and several others. This, however, relies on the validated tomographic reconstruction algorithm (which is currently work in progress applied to MATS data). The algorithm needs certain adaptations for auroral tomography compared to airglow tomography, as the aurora is not earth-fixed in the way the airglow is. Finally, it only works for cases where the emission distribution is sufficiently stable in time, an assumption which is often violated in the aurora. Thus, we consider this topic to be out of the scope of this paper.
  
  Analyzing full images. Yes, that is an excellent point. There is more information in the full images than what we can extract from only the keograms. Looking at full images will indeed bring forth a lot of new information. Consequently, it is difficult to organize into a few parameters that fit into this paper. For this reason, it is out of the scope of this paper, and we will look into it in the future.
  
  Minor comments
  Thank you for taking the time to carefully look at the manuscript and make minor comments. We will thoroughly address all of the typos and rephrasing issues and make the modifications in the manuscript. Below, we reply to the questions included in the comments and some comments that we thought required clarification:
  Third-order polynomial. The reason for choosing a third-order polynomial is that whenever the satellite moves towards the dayside or nightside, there is a strong variation in the background. After several tests, we concluded that a third-order polynomial worked best when handling the background in these cases. Whenever the keogram is entirely in the nightside or the dayside, the polynomial looks flat, as in Figure 1. We chose to show this event because it was a clear and clean event, which we thought would make things easy to understand. We will clarify the reason for the third-order polynomial in the text and consider whether to change the example figure to one that does not appear flat.
  
  378 events out of 4860 auroral oval crossings. Thank you for your question. We always observe the airglow layer in our images. To detect auroral emissions, they need to be spatially distinct enough from the airglow layer, and the brightness must be above a certain threshold. In addition, we don’t consider auroral events that are too complex, i.e., auroral traces in the keogram overlapping each other. We will clarify this in the manuscript. Attached is a distribution of the detected auroral events over the analyzed period for both hemispheres. In addition, we discuss the estimation of complex events vs events without distinguishable aurora three points below.
  
  Figure 3, panel a). We have evaluated the number of data points remaining if we consider bins with a larger number of elements. With at least 3, we get 33 data points. Increasing the threshold to 4, we get 26 data points, and only 1 data point with high kp remains. Increasing it to 5, we get 24 data points, with no data points in the range 6-9 kp. Events with high kp are not common, and we wanted to keep some in the plot; that is why we chose to keep the bins with at least 3 data points and not increase the threshold. Regarding the possibility of adding the uncertainty, we will add the standard deviation/sqrt (n) as the uncertainty of the mean estimate in the text or caption.
  
  Figure 3, panel b). We agree that it makes sense to plot the brightness in kR (integrated over the filter passband) instead of the spectral intensity in phnm-1 m-2 sr-1 s-1, we will change it. We will also add the quantity that we are plotting in the y-label. Regarding the fact that the reported trend between peak altitude and brightness is not obvious, statistically, there is a relation between the energy flux and the characteristic energy of auroral electrons (going back to Knight, 1973), which was the motivation behind our analysis. We believe this relationship may also be observable in the plot, but we agree that it is not very convincing. We will review and add relevant references for the discussion around this point, and the reasons for the (weak) relation in our data.
  
  Estimate of missed complex events. We have conducted an estimate of how many events with a complex distribution have been omitted, compared to the total number of crossings and the total number of auroral events. The estimation has revealed that in about 90 % of the crossings, the aurora is not visible or too weak to be detected by the algorithm. Then, 2% of the crossings detect complex auroral distributions (indicating closely spaced multiple auroral structures) that are omitted, while about 8% detect aurora clear enough to keep in the study. It is also relevant to mention that due to operational reasons, some data gaps reduce the dataset to about 85% to 90% of the expected crossings. When editing the manuscript, we will clarify that some bias may have been introduced by omitting complex events that may correspond to multiple arcs, while mostly keeping single-arc events.
  
  Again, thank you for your comments, suggestions, and references. Regarding all the points that we haven’t mentioned above, we plan to implement your suggestions as suggested.
  
  Citation: https://doi.org/10.5194/egusphere-2025-2324-AC1

Judit Pérez-Coll Jiménez, Nickolay Ivchenko, Ceona Lindstein, Lukas Krasauskas, Jonas Hedin, Donal Patrick Murtagh, Linda Megner, Björn Linder, and Jörg Gumbel

Viewed

Total article views: 1,334 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
1,296	27	11	1,334	52	73

HTML: 1,296
PDF: 27
XML: 11
Total: 1,334
BibTeX: 52
EndNote: 73

Views and downloads (calculated since 03 Jun 2025)

Month	HTML	PDF	XML	Total
Jun 2025	111	14	6	131
Jul 2025	42	7	4	53
Aug 2025	434	3	1	438
Sep 2025	709	3	0	712

Cumulative views and downloads (calculated since 03 Jun 2025)

Month	HTML	PDF	XML	Total
Jun 2025	111	14	6	131
Jul 2025	42	7	4	53
Aug 2025	434	3	1	438
Sep 2025	709	3	0	712

Viewed (geographical distribution)

Total article views: 1,308 (including HTML, PDF, and XML) Thereof 1,308 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 17 Sep 2025

Short summary

This study uses images taken by the Swedish satellite MATS to conduct a statistical analysis of the molecular oxygen atmospheric band emissions in the aurora. This auroral emission can not be observed from the ground, making it one of the least understood auroral emissions. Our results provide a new dataset with information on the peak altitude, geomagnetic location, and auroral intensity of 378 events detected between February and April 2023.


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

A statistical study of the O2Atm(0-0) aurora observed by the Swedish satellite MATS

Viewed

Viewed (geographical distribution)

A statistical study of the O₂Atm(0-0) aurora observed by the Swedish satellite MATS