the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Jul 2026
Characterizing Lightning NOx Production in the Ebro Valley with MTG-LI and LMA Observations
Abstract. Lightning is one of the main sources of NOₓ in the Earth’s atmosphere, yet there is considerable variability in NOₓ production during thunderstorms. In this study, we combined optical lightning data from the Meteosat Third Generation (MTG) Lightning Imaging (LI) instrument, which provides information on the occurrence rate and physical characteristics of lightning flashes, measurements from an extended Lightning Mapping Array (LMA) consisting of 24 sensors, lightning measurements from the Lightning Location System (LLS) of the Meteorological Service of Catalonia (MSC), and TROPO-spheric Monitoring Instrument (TROPOMI) cloud and NO₂ research products to investigate lightning NOₓ emissions in the Ebro Valley (Spain). WRF-Chem simulations were used to calculate the necessary air mass factors and to subtract the NOₓ not produced by lightning from the TROPOMI measurements, allowing a more accurate estimation of lightning-induced NOₓ. We found positive relationships between lightning-produced NOₓ per flash and several lightning properties, including flash channel length, radiance, footprint, duration and the length of the optical continuous signal detected from space. These results highlight the importance of combining detailed lightning observations with atmospheric modeling to quantify NOₓ emissions from thunderstorms.
- Preprint
(2541 KB) - Metadata XML
-
Supplement
(19953 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2026-3308', Anonymous Referee #1, 23 Jul 2026
- AC1: 'Reply on RC1', Francisco Javier Perez-Invernon, 29 Sep 2026
-
RC2: 'Comment on egusphere-2026-3308', Anonymous Referee #2, 23 Jul 2026
Overall Quality of the Pre-print (General Comments)
The authors examine the moles of NOx produced per flash using flash information from a satellite-based network (MTG-LI) and a ground-based LMA and meteorological/trace gas information from TROPOMI. By pairing the analysis with model simulations (WRF) they reduce uncertainties in the tropospheric background and in the profile of lightning-NOx that plague satellite-based estimates of LNOx PE. The authors provide detailed information on the method they used to estimate the PE. The discussion is good although can be focused more.
Using the 98th PCTL vertical density score as a constraint on the region-of-interest seems like a great idea. The method of determining the NOx column due to lightning appears sound.
Main Comments:
The authors should add a paragraph discussing why they believe their PE values are lower than most other estimates.
The authors need to shorten the discussion of Figure 6. Possible methods for doing that are shown below.
Figure 6 and discussion:
- This array contains too much information to be grasped easily. It would help to only show the lower-left plus the diagonal.
- You could also remove LMA-av.length or LMA-volume variables as their correlation is 0.99.
- Unfortunately, I feel the remaining array would still too large. I would then remove any variables that are not discussed or only briefly mentioned in the text. Hopefully, this will get you to a 10x10 array or less.
- You might also consider moving some of the discussion to the legend as it is tough to move back and forth between the text and the plot. At the very least, add some numbers to the main body.
- Also, is it necessary to show both the LMA and MTG variables here.
- You might also break this into 2 plots and the discussion of Figure 6 drags on and on.
- Finally, I would use bullets or numbers to emphasize the points given in the text. Otherwise, you will lose some readers here.
Minor Comments:
L37: Did Rahman et al. (2007) include an estimate of the uncertainty or is that given by the small difference between 2 and 2.4?
L46-48: Please re-write this awkward sentence; its meaning is unclear.
L77-78: Please re-write this awkward sentence. Perhaps “In this study, to reduce uncertainties in estimates of LNOx PE, we investigate relationships between … Ebro Valley.
L125: Are data collected between July 8th and October 31st valid for scientific purposes? Aren’t data available beginning July 4th?
L137-139: Be explicit as to how you determine the number of time-contiguous groups per flash and the ratio of “long-duration” to total flashes, i.e., what standard MTG-LI L2 products were used to create the new products, and how the standard products were used to create the new products.
L225: Did you use the WRF processing system to go from 34 meteorological levels to 37 model levels? Also, why did you choose a different number?
L256: It appears that you used lightning-nudging to ensure that deep (“high”) convection occurred at the location of observed flashes and to place lightning-NOx emissions? Be sure this is clear in the text.
L259-262: I don’t understand how sensitivity simulations with a different PE can be used to determine the background-NOx?
L303-305: Is the summertime ratio also elevated because summertime storms produce more lightning than spring/autumnal storms or did you control for that?
Figure 4b: I don’t understand what one obtains when you subtract VCD_strat from the total SCD and then divide by AMFstrat. I’m more used to seeing VCD_strat multiplied by AMFstrat and then subtracted from the SCD, which is of course the operational tropVCD.
L375-394: You pack a lot of interesting information into this paragraph. Yes, it probably makes sense to use groups rather than flashes when determining which grid boxes are lightning-influenced.
Table 2: I’d suggest adding a summary row containing the mean and standard deviation for each of the parameters.
L429-430: You state that the PE is significantly influenced by the PE choice used in WRF-Chem. How much did it change by when you adjusted the PE using the ratio of LMA to MTG flashes? Can you quantify the significance?
L453: I don’t understand this --- Are thunderstorms more active near sunrise?
L455-456: Wouldn’t a higher DE during the early morning mean that MTG-LI would capture more weak flashes during the morning that would be missed when the DE is lower?
L479: Isn’t the flash energy also proportional to the pressure?
L505-506: I’d be careful here. I believe GLM and presumably MTG-LI detects CG flashes more efficiently than IC flashes. What do you mean by activity?
L507-509: Interesting hypothesis: Emphasize how large the difference in correlation is between LNOx PE and duration/multiplicity. Is the difference significant?
Technical Comments:
L16: accounting up --- accounting for up
L24: are essential --- is essential
L29: or the EULINOX --- and the EULINOX
L34: unit discharge energy --- unit discharge of energy
L84: metric --- metrics
L84: cloud covers --- cloud cover
L108: research products used in this study provide --- research products provide
L126: Earth’s full (what?)
L129: define sub-satellite point
L152: ELMA or eLMA?
L167: from breaking them up into multiple--- from being broken into multiple
L168: no longer being detected --- no longer detectable
L183: calculate the total number of CG --- calculate the ratio of CG
L184: composed by --- composed of
L192: flashesy. --- flashes.
L223-225: No need to also write out “degrees” when you are using the degrees symbol.
L280: clearly seen in when comparing Figure 2(c) --- clearly see by comparison of Figure 2(c)
L346: is not exactly the same than the --- is not exactly the same as the
L446: av. Lenfth --- av. Length
L446-449: Please delete information within quotes to improve readability.
L451: flashes then to have --- flashes tend to have
L487-489: lower altitudes accordingly to the LMA --- lower altitudes
L519: which is agreement with – are compare to (I’m not sure they agree as 0.56 is less than the lower range (0.8).
L553: that variations --- that variation
L600: will provide --- provides
Citation: https://doi.org/10.5194/egusphere-2026-3308-RC2 - AC2: 'Reply on RC2', Francisco Javier Perez-Invernon, 29 Sep 2026
Status: closed
-
RC1: 'Comment on egusphere-2026-3308', Anonymous Referee #1, 23 Jul 2026
The paper describes a study to characterise NOx production from lightning in a specific region and time frame using various ground- and space-based instrumentation which is supported by various modelling and calculation techniques. The paper is incredibly detailed and clearly outlines all aspects of the study, although some further illustrations/schematics would be helpful as certain parts (particularly the method) can become confusing without something to refer too.
The inclusion of uncertainties is important, particularly as the research is essentially trying to cross-reference various measurements, and it relies on some basic assumptions which may not necessarily be true (although this is mitigated to some extent). Lightning phenomena varies widely across the world, and this one specific study in this location may not be representative of other types of storm systems / lightning cells elsewhere – It may be worth discussing how this could relate to other similar systems or regions? It is, however, a very good step forwards and a very interesting and comprehensive study.
A few minor specific comments:
Ln 14-15: The way the references are included here are a little confusing; one from 1947 (a book I cannot fully access) and another from 2007, with the sentence suggesting that NOx from lightning has, over a period of 60 years, stayed relatively stable at about 10%. But this is not true as, currently, human-made NOx sources are at around 70-80% but would not have been this high in the 1940s (the absolute amount of NOx from lightning may not have changed, but the percentage will have). Maybe some recognition that the 10% is just what the LNOx split is now but that, years ago, it likely had a much bigger split.
Ln 96: There is a large amount of text describing the methodology in great detail, but it may be helpful to include a schematic illustrating the ground-based networks and satellites used as a ‘quick reference’ – particularly as so many are mentioned and it is quite hard to understand exactly what data is associated with each specific instrument. This extends to including maps with the specific latitude and longitude ranges used both for the case study and from the modelling perspective.
Ln 257: I can understand where the 311 moles of NO per flash has come from, but does the study not distinguish / can it not distinguish between CG and IC flashes? There is also an over-simplification in the reference quoted here (Price, 1997) – 1112 for CG is just ten times 111.2 for IC; it is quite an old reference which has been observed to be incorrect (and there are some recent publications on this). I appreciate that this is just utilising a number for a model, and that if it is incorrect, it is an incorrectness which is carried through and so relative comparisons can be made.
Citation: https://doi.org/10.5194/egusphere-2026-3308-RC1 - AC1: 'Reply on RC1', Francisco Javier Perez-Invernon, 29 Sep 2026
-
RC2: 'Comment on egusphere-2026-3308', Anonymous Referee #2, 23 Jul 2026
Overall Quality of the Pre-print (General Comments)
The authors examine the moles of NOx produced per flash using flash information from a satellite-based network (MTG-LI) and a ground-based LMA and meteorological/trace gas information from TROPOMI. By pairing the analysis with model simulations (WRF) they reduce uncertainties in the tropospheric background and in the profile of lightning-NOx that plague satellite-based estimates of LNOx PE. The authors provide detailed information on the method they used to estimate the PE. The discussion is good although can be focused more.
Using the 98th PCTL vertical density score as a constraint on the region-of-interest seems like a great idea. The method of determining the NOx column due to lightning appears sound.
Main Comments:
The authors should add a paragraph discussing why they believe their PE values are lower than most other estimates.
The authors need to shorten the discussion of Figure 6. Possible methods for doing that are shown below.
Figure 6 and discussion:
- This array contains too much information to be grasped easily. It would help to only show the lower-left plus the diagonal.
- You could also remove LMA-av.length or LMA-volume variables as their correlation is 0.99.
- Unfortunately, I feel the remaining array would still too large. I would then remove any variables that are not discussed or only briefly mentioned in the text. Hopefully, this will get you to a 10x10 array or less.
- You might also consider moving some of the discussion to the legend as it is tough to move back and forth between the text and the plot. At the very least, add some numbers to the main body.
- Also, is it necessary to show both the LMA and MTG variables here.
- You might also break this into 2 plots and the discussion of Figure 6 drags on and on.
- Finally, I would use bullets or numbers to emphasize the points given in the text. Otherwise, you will lose some readers here.
Minor Comments:
L37: Did Rahman et al. (2007) include an estimate of the uncertainty or is that given by the small difference between 2 and 2.4?
L46-48: Please re-write this awkward sentence; its meaning is unclear.
L77-78: Please re-write this awkward sentence. Perhaps “In this study, to reduce uncertainties in estimates of LNOx PE, we investigate relationships between … Ebro Valley.
L125: Are data collected between July 8th and October 31st valid for scientific purposes? Aren’t data available beginning July 4th?
L137-139: Be explicit as to how you determine the number of time-contiguous groups per flash and the ratio of “long-duration” to total flashes, i.e., what standard MTG-LI L2 products were used to create the new products, and how the standard products were used to create the new products.
L225: Did you use the WRF processing system to go from 34 meteorological levels to 37 model levels? Also, why did you choose a different number?
L256: It appears that you used lightning-nudging to ensure that deep (“high”) convection occurred at the location of observed flashes and to place lightning-NOx emissions? Be sure this is clear in the text.
L259-262: I don’t understand how sensitivity simulations with a different PE can be used to determine the background-NOx?
L303-305: Is the summertime ratio also elevated because summertime storms produce more lightning than spring/autumnal storms or did you control for that?
Figure 4b: I don’t understand what one obtains when you subtract VCD_strat from the total SCD and then divide by AMFstrat. I’m more used to seeing VCD_strat multiplied by AMFstrat and then subtracted from the SCD, which is of course the operational tropVCD.
L375-394: You pack a lot of interesting information into this paragraph. Yes, it probably makes sense to use groups rather than flashes when determining which grid boxes are lightning-influenced.
Table 2: I’d suggest adding a summary row containing the mean and standard deviation for each of the parameters.
L429-430: You state that the PE is significantly influenced by the PE choice used in WRF-Chem. How much did it change by when you adjusted the PE using the ratio of LMA to MTG flashes? Can you quantify the significance?
L453: I don’t understand this --- Are thunderstorms more active near sunrise?
L455-456: Wouldn’t a higher DE during the early morning mean that MTG-LI would capture more weak flashes during the morning that would be missed when the DE is lower?
L479: Isn’t the flash energy also proportional to the pressure?
L505-506: I’d be careful here. I believe GLM and presumably MTG-LI detects CG flashes more efficiently than IC flashes. What do you mean by activity?
L507-509: Interesting hypothesis: Emphasize how large the difference in correlation is between LNOx PE and duration/multiplicity. Is the difference significant?
Technical Comments:
L16: accounting up --- accounting for up
L24: are essential --- is essential
L29: or the EULINOX --- and the EULINOX
L34: unit discharge energy --- unit discharge of energy
L84: metric --- metrics
L84: cloud covers --- cloud cover
L108: research products used in this study provide --- research products provide
L126: Earth’s full (what?)
L129: define sub-satellite point
L152: ELMA or eLMA?
L167: from breaking them up into multiple--- from being broken into multiple
L168: no longer being detected --- no longer detectable
L183: calculate the total number of CG --- calculate the ratio of CG
L184: composed by --- composed of
L192: flashesy. --- flashes.
L223-225: No need to also write out “degrees” when you are using the degrees symbol.
L280: clearly seen in when comparing Figure 2(c) --- clearly see by comparison of Figure 2(c)
L346: is not exactly the same than the --- is not exactly the same as the
L446: av. Lenfth --- av. Length
L446-449: Please delete information within quotes to improve readability.
L451: flashes then to have --- flashes tend to have
L487-489: lower altitudes accordingly to the LMA --- lower altitudes
L519: which is agreement with – are compare to (I’m not sure they agree as 0.56 is less than the lower range (0.8).
L553: that variations --- that variation
L600: will provide --- provides
Citation: https://doi.org/10.5194/egusphere-2026-3308-RC2 - AC2: 'Reply on RC2', Francisco Javier Perez-Invernon, 29 Sep 2026
Data sets
High-resolution atmospheric simulations using a modified WRF-Chem v4.3: Lightning effects on air quality in the Ebro Valley - Lightning nudging towards MTG-LI Francisco J. Pérez-Invernón https://doi.org/10.5281/zenodo.20065621
High-resolution atmospheric simulations using a modified WRF-Chem v4.3: Lightning effects on air quality in the Ebro Valley - Lightning nudging towards MTG-LI scaled to XCALMA Francisco J. Pérez-Invernón https://doi.org/10.5281/zenodo.19588416
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 192 | 89 | 36 | 317 | 69 | 33 | 32 |
- HTML: 192
- PDF: 89
- XML: 36
- Total: 317
- Supplement: 69
- BibTeX: 33
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The paper describes a study to characterise NOx production from lightning in a specific region and time frame using various ground- and space-based instrumentation which is supported by various modelling and calculation techniques. The paper is incredibly detailed and clearly outlines all aspects of the study, although some further illustrations/schematics would be helpful as certain parts (particularly the method) can become confusing without something to refer too.
The inclusion of uncertainties is important, particularly as the research is essentially trying to cross-reference various measurements, and it relies on some basic assumptions which may not necessarily be true (although this is mitigated to some extent). Lightning phenomena varies widely across the world, and this one specific study in this location may not be representative of other types of storm systems / lightning cells elsewhere – It may be worth discussing how this could relate to other similar systems or regions? It is, however, a very good step forwards and a very interesting and comprehensive study.
A few minor specific comments:
Ln 14-15: The way the references are included here are a little confusing; one from 1947 (a book I cannot fully access) and another from 2007, with the sentence suggesting that NOx from lightning has, over a period of 60 years, stayed relatively stable at about 10%. But this is not true as, currently, human-made NOx sources are at around 70-80% but would not have been this high in the 1940s (the absolute amount of NOx from lightning may not have changed, but the percentage will have). Maybe some recognition that the 10% is just what the LNOx split is now but that, years ago, it likely had a much bigger split.
Ln 96: There is a large amount of text describing the methodology in great detail, but it may be helpful to include a schematic illustrating the ground-based networks and satellites used as a ‘quick reference’ – particularly as so many are mentioned and it is quite hard to understand exactly what data is associated with each specific instrument. This extends to including maps with the specific latitude and longitude ranges used both for the case study and from the modelling perspective.
Ln 257: I can understand where the 311 moles of NO per flash has come from, but does the study not distinguish / can it not distinguish between CG and IC flashes? There is also an over-simplification in the reference quoted here (Price, 1997) – 1112 for CG is just ten times 111.2 for IC; it is quite an old reference which has been observed to be incorrect (and there are some recent publications on this). I appreciate that this is just utilising a number for a model, and that if it is incorrect, it is an incorrectness which is carried through and so relative comparisons can be made.