the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Oct 2025
New technique for aurora / dayglow separation in UV imagery: IMF By dependence of seasonal differences in auroral oval location
Abstract. We investigate how the location, size, and intensity of the auroral oval is affected by the combination of a strong dawn/dusk component of the interplanetary magnetic field (IMF By ) during northward IMF and the tilt of the Earth's magnetic dipole axis. Sunlit auroral observations are contaminated by dayglow, and its impact on average intensity estimates remains unclear. Dayglow modelling is also accompanied by significant uncertainties that increase with increasing sunlight intensity. These difficulties motivate us to develop a new technique for separating dayglow and aurora, where we assume the observed distribution of intensities consists of a convoluted distribution with separate contributions from dayglow and aurora. By performing a nonlinear fit to observed count distributions one may extract the best-fit parameters that describe the dayglow and auroral sources separately. We apply this separation method to data obtained by the Special Sensor Ultraviolet Spectrographic Imager (SSUSI) onboard the Defense Meteorological Satellite Program's (DMSP) F16–19 satellites. We demonstrate to what extent the heteroskedasticity of distributions of dayglow intensity can influence quantitative estimates of the average auroral intensities. By applying our separation method to SSUSI observations made during stable and strong IMF By conditions over many years, the auroral component is isolated, and estimates of auroral intensities during sunlit conditions are improved. The separation method produces 30–40 % higher auroral intensities than a simple calculation of the distribution mean. Statistics of the auroral component show a clear and substantial (~500 km in NH, ~430 km in SH) dawn-dusk shift in the polar cap location depending on the sign of IMF By during local summer in both Hemispheres. This shift is absent during local winter. We propose that the cause of this seasonally dependent shift in the polar cap location is likely to be related to seasonal differences in lobe reconnection rates. To our knowledge, the seasonal dependence of polar cap location has not been previously reported.
- Preprint
(19080 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-4317', Anonymous Referee #1, 31 Oct 2025
-
AC1: 'Reply on RC1', Jens Christian Hessen, 08 Jan 2026
Hessen et al.
This paper develops a new technique to separate contributions to auroral images from the auroras and dayglow. The technique is applied to images from the DMSP/SSUSI instrument. The technique assumes that the distributions of luminosity contributions from the two sources are different, and uses a probabilistic approach to separate the two. The technique cannot be applied to individual images, but to the sum of multiple images collected over several months. Hence seasonal variations in the locations of the auroras are studied. The paper concludes that dawn-dusk asymmetries in the location of the auroral oval associated with IMF By dominate in some seasons and not others. In principle the paper is of interest, but is has some major short-fallings which require major corrections to address. This issues are outlined below, on no particular order.
The paper studies dawn-dusk asymmetries in the location of the auroras during northward IMF. It is not stated in the title that the study relates to northward IMF, it is only alluded to a few times within the paper, and it is never (to my knowledge) explained why the authors ignore southward IMF. The rationale for studying northward rather than southward IMF should be clearly articulated and should be clear in the title.
- This is a good point, thank you for bringing this to our attention. We will add the following lines in the introduction of the revised manuscript to make clear to the reader that we have chosen to focus on northward IMF: "We limited the scope to northward IMF. We reserve investigation of other IMF orientations for possible future studies."
- We also agree with the reviewer that the title could be more focused. In the revision the title will be “New technique for aurora / dayglow separation in UV imagery: IMF By dependence of seasonal differences in auroral oval location during positive IMF Bz"
A major issue appears to be with the dataset used itself. The DMSP/SSUSI images have already had a dayglow removal algorithm applied to them (which introduces serious artefacts into them – see below). This paper then tries to separate a dayglow contribution, which has supposedly already been removed, from the auroral contribution. It would seem to make more sense to use the original images, without dayglow removal, and develop a new technique that does the removal better. One of the coauthors is from the DSMP/SSUSI team – is the original data available?
- This comment consists of a suggestion and a question.
Regarding the suggestion to use the original, uncorrected images: Our team’s experience with dayglow removal for both SSUSI and other imagers has shown that it is far from trivial, and we do not necessarily share the reviewer’s view that it would make more sense to attempt to develop a new dayglow removal technique with the original data. The problem of dayglow removal from SSUSI images has been addressed by one of us as recently as 2022 (Zhang et al, doi: 10.1016/j.jastp.2022.105833), as indicated in the original manuscript.
Perhaps most importantly, even if we did develop a more accurate dayglow removal method there would still be residual errors associated with counting statistics as well as processes not captured by the existing dayglow removal procedure, which is based on solar zenith angle and look angle. In other words, we would still be in a situation where the dayglow-corrected data have an associated bias and error. We therefore respectfully submit that the cost of developing a new dayglow removal procedure would far outweigh the benefit.
Regarding availability of the original data, they can be found on CDAWeb: https://cdaweb.gsfc.nasa.gov/pub/data/dmsp/. We will include this information in the data availability statement of the revised manuscript.
Artefacts in the data include a bias towards negative luminosities (even when the positive auroral contributions are included!) and spikes in the distributions at particular luminosity values. At the very least there should be a discussion in the paper about how the original dayglow removal is done, and why it leads to these artefacts. Otherwise the results from this paper are of little use. Currently, most of the paper is discussing the fact that there are these artefacts, without describing why they arise. The paper would be much better focussed if the original data was used (see previous comment).
- The reviewer is correct that the dataset exhibits an overall negative bias. The distribution of counts in bins outside the auroral oval is fairly consistent, with a mostly symmetric distribution and mean values that vary between -250R and -50R in the summer. In contrast, the mean is almost never negative in bins where the aurora is present.
There are some outliers, like extreme negative values, and the spike around -60R to -40R. The negative values are caused by data processing: removing scatters from bright emissions (H121.6nm, O130.4nm and O135.6 nm) into the LBH bands and counting errors. Isolated spikes are due to glints and MeV particle noises. - However, the contributions from these are insignificant when the method is applied. This can be seen when we truncate the window to -2000R to 5000R.
I’m not sure that I understand Figures 6 and 7. Three distributions are shown: blue is aurora and yellow is DG-residual. My understanding is that green is the sum of these two distributions, but this is clearly not the case. For instance, if green is the sum then the two bottom distributions should be bimodal. Please explain these figures better.
- The “sum” curve is fitted to the sum of two randomly distributed values (Z = X +Y). The probability distribution function (pdf) of this sum is the convolution of the pdfs of the two random variables, and not the sum of their pdfs.
We will make this point clearer by rewriting the relevant portion of the caption of Figure 5 in the revised manuscript (formerly Figure 6) so that it reads, “The ‘sum’ curve represents the convoluted distribution given by Equation 5.”
I think Figures 4 and 8 are the same, but one has more labelling. Just use the second figure.
- Thank you for pointing this out, Figure 4 will be dropped in the revised manuscript.
Citation: https://doi.org/10.5194/egusphere-2025-4317-AC1
-
AC1: 'Reply on RC1', Jens Christian Hessen, 08 Jan 2026
-
RC2: 'Comment on egusphere-2025-4317', Anonymous Referee #2, 29 Nov 2025
The article touches an important topic of observing aurora under dayglow conditions. The authors introduce an algorithm for cleaning the widely-used DMSP SSUSI dataset of auroral emissions. They suggest this new algorithm as an addition, or as an alternative, to the standard cleaning algorithm developed at the John Hopkins APL. It appears that the article is based on the master thesis of the primary author (Hessen, 2023). The proposed cleaning algorithm and its validation is interesting. My key concern is that the article is describing technical developments (details of the cleaning algorithm) and thus it does not fit the scope of Annales Geophysicae as a geophysical journal. I would recommend that the authors select a more relevant journal for this article, e.g., Atmospheric Measurement Techniques. I included some specific comments below, especially regarding the choice of IMF conditions for the validation.
Line 2: It is stated here and throughout the article that northward IMF conditions are considered, but there is no clear explanation of why southward IMF conditions are excluded. IMF By should also be a controlling factor for dawn-dusk asymmetries under southward IMF, so southward IMF cases should be also considered.
17: The statement that “the seasonal dependence of polar cap location has not been previously reported” is confusing. I would recommend the authors to remove this strong statement and summarize the actual results of their study instead, or at least to make the statement more specific.
70-74: Here I have a question about the methodology. The authors mention correctly that the dayglow removal routines have been already applied to the SSUSI dataset. Do they apply their new cleaning algorithms to the SSUSI dataset that has already been cleaned? If so, they should explain the rationale, because the original dayglow cleaning routines may have introduced certain biases. Perhaps, it would be preferable to apply the new dayglow cleaning algorithm to the lower-level, uncleaned SSUSI data?
Citation: https://doi.org/10.5194/egusphere-2025-4317-RC2 -
AC2: 'Reply on RC2', Jens Christian Hessen, 08 Jan 2026
The article touches an important topic of observing aurora under dayglow conditions. The authors introduce an algorithm for cleaning the widely-used DMSP SSUSI dataset of auroral emissions. They suggest this new algorithm as an addition, or as an alternative, to the standard cleaning algorithm developed at the John Hopkins APL. It appears that the article is based on the master thesis of the primary author (Hessen, 2023). The proposed cleaning algorithm and its validation is interesting. My key concern is that the article is describing technical developments (details of the cleaning algorithm) and thus it does not fit the scope of Annales Geophysicae as a geophysical journal. I would recommend that the authors select a more relevant journal for this article, e.g., Atmospheric Measurement Techniques. I included some specific comments below, especially regarding the choice of IMF conditions for the validation.
- We wish to thank the reviewer for their feedback and suggestions for improving the manuscript. We have strived to bring the revised manuscript in line with the reviewer’s comments.
From this and the reviewer’s last comment, we gather that the reviewer is under the impression that this study presents a new cleaning algorithm. We wish to clarify that the method we present is not intended as an additional or alternative cleaning algorithm for SSUSI data. Rather, our methodology aims to isolate separate contributions from dayglow error (not the dayglow itself!) and aurora in the convolved distribution of luminosity in each bin. This methodology allows us to obtain a better statistical estimate of the location and brightness of the auroral oval.
The actual physical result we have highlighted is that the auroral oval boundaries shift in response to IMF By. We believe this result makes the manuscript a clear candidate for Annales, instead of a journal focused solely on methodology.
Line 2: It is stated here and throughout the article that northward IMF conditions are considered, but there is no clear explanation of why southward IMF conditions are excluded. IMF By should also be a controlling factor for dawn-dusk asymmetries under southward IMF, so southward IMF cases should be also considered.
- This is a fair point. We made the choice to focus on northward IMF based on our own research interests. We do however point out in the conclusions that investigation of the role of IMF By during southward IMF conditions would likely be enlightening, and reserve this as a topic for future work.
17: The statement that “the seasonal dependence of polar cap location has not been previously reported” is confusing. I would recommend the authors to remove this strong statement and summarize the actual results of their study instead, or at least to make the statement more specific.
- We thank the reviewer for catching this. We agree that the sentence in question is unnecessary, and we will remove it from the revised manuscript.
70-74: Here I have a question about the methodology. The authors mention correctly that the dayglow removal routines have been already applied to the SSUSI dataset. Do they apply their new cleaning algorithms to the SSUSI dataset that has already been cleaned? If so, they should explain the rationale, because the original dayglow cleaning routines may have introduced certain biases. Perhaps, it would be preferable to apply the new dayglow cleaning algorithm to the lower-level, uncleaned SSUSI data?
- We thank the reviewer for helping us to understand that this aspect was unclear. We will modify the description on these lines and elsewhere to clarify that we use “dayglow-corrected SSUSI irradiance data”.
In response to the reviewer’s question about our “new cleaning algorithm”: we emphasize that our methodology does not constitute a cleaning algorithm, as we have already pointed out in our response to the reviewer’s first comment. Rather, the methodology we present is intended to enable a better statistical estimate of the auroral contribution to the SSUSI irradiance measurements. Our methodology assumes that dayglow has already been removed, and that what remains is both the auroral contribution and a statistical bias or offset (the “dayglow error”).
Citation: https://doi.org/10.5194/egusphere-2025-4317-AC2 - We wish to thank the reviewer for their feedback and suggestions for improving the manuscript. We have strived to bring the revised manuscript in line with the reviewer’s comments.
-
AC2: 'Reply on RC2', Jens Christian Hessen, 08 Jan 2026
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 285 | 255 | 29 | 569 | 22 | 21 |
- HTML: 285
- PDF: 255
- XML: 29
- Total: 569
- BibTeX: 22
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
New technique for aurora / dayglow separation in UV imagery: IMF By dependence of seasonal differences in auroral oval location
Hessen et al.
This paper develops a new technique to separate contributions to auroral images from the auroras and dayglow. The technique is applied to images from the DMSP/SSUSI instrument. The technique assumes that the distributions of luminosity contributions from the two sources are different, and uses a probabilistic approach to separate the two. The technique cannot be applied to individual images, but to the sum of multiple images collected over several months. Hence seasonal variations in the locations of the auroras are studied. The paper concludes that dawn-dusk asymmetries in the location of the auroral oval associated with IMF By dominate in some seasons and not others. In principle the paper is of interest, but is has some major short-fallings which require major corrections to address. This issues are outlined below, on no particular order.
The paper studies dawn-dusk asymmetries in the location of the auroras during northward IMF. It is not stated in the title that the study relates to northward IMF, it is only alluded to a few times within the paper, and it is never (to my knowledge) explained why the authors ignore southward IMF. The rationale for studying northward rather than southward IMF should be clearly articulated and should be clear in the title.
A major issue appears to be with the dataset used itself. The DMSP/SSUSI images have already had a dayglow removal algorithm applied to them (which introduces serious artefacts into them – see below). This paper then tries to separate a dayglow contribution, which has supposedly already been removed, from the auroral contribution. It would seem to make more sense to use the original images, without dayglow removal, and develop a new technique that does the removal better. One of the coauthors is from the DSMP/SSUSI team – is the original data available?
Artefacts in the data include a bias towards negative luminosities (even when the positive auroral contributions are included!) and spikes in the distributions at particular luminosity values. At the very least there should be a discussion in the paper about how the original dayglow removal is done, and why it leads to these artefacts. Otherwise the results from this paper are of little use. Currently, most of the paper is discussing the fact that there are these artefacts, without describing why they arise. The paper would be much better focussed if the original data was used (see previous comment).
I’m not sure that I understand Figures 6 and 7. Three distributions are shown: blue is aurora and yellow is DG-residual. My understanding is that green is the sum of these two distributions, but this is clearly not the case. For instance, if green is the sum then the two bottom distributions should be bimodal. Please explain these figures better.
I think Figures 4 and 8 are the same, but one has more labelling. Just use the second figure.