the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Nov 2023
From CO2 emissions to atmospheric NO2 mixing ratios: simulating chemical processes and their impacts on TROPOMI retrievals over the Middle East
Abstract. As many large metropolitan areas have pledged for a rapid decrease of their greenhouse gas emissions through ambitious climate mitigation policies, the need for rapid and robust quantification methods became more pressing. At the global scale, the scarcity of satellite carbon dioxide (CO2) observations remains the major roadblock to producing independent city-scale CO2 emissions estimates from atmospheric data, except for a handful of cities benefiting from a dense network of ground-based CO2 sensors. In this study, we quantify the potential of assimilating indirect measurements from spaceborne sensors (here nitrogen dioxide – NO2) to constrain fossil fuel CO2 emissions, relying on the co-emission of these two species during combustion. We developed a modeling framework using a NOx-aerosol chemistry transport model (WRF-Chem) and performed simulations of NO2 and CO2 over the Middle-East, an area known for its large cities, its frequent clear sky conditions and a fairly constant albedo from its desertic land. We first demonstrate the importance of production/destruction processes impacting NO2 lifetime at short and long distances from the source, suggesting that simplified approaches may be impacted by large errors. In comparison to TROPOMI satellite observations, the simulated NO2 plumes from emissions inventories (EDGAR) revealed large misattribution of NO2 emissions at fine scales, hence an uncertain disaggregation of national emissions to single point sources in the industrial and energy sectors in the EDGAR inventory. We further studied the relationship between NO2 and CO2 during summer and winter seasons by simulating the enhancement ratios (δNO2:δXCO2) in plumes produced by cities and point sources We found that the enhancement ratios are consistent with the observed ratios derived from the ESA TROPOMI (NO2) and NASA OCO-3 missions (XCO2). We conclude here that the spatial misattribution of NO2 emissions parallels the misattribution of CO2 emissions, and that improved NO2 inventories could therefore be used to improve the monitoring of CO2 emissions at sub-national scales in current global inventories when a sufficiently-large amount of NO2 satellite measurements are available.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(3403 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-2487', Anonymous Referee #1, 04 Dec 2023
Overall this paper is scientifically sound. It investigates NO2 in the Saudi Arabia and Qatar regions of the Middle East by comparing satellite to model simulations: a full chemical transport simulation and a passive tracer simulation. Some conclusions are also made about CO2 emissions after the NO2 inter-comparison It is a good advancement and worthy of publication, but there are several areas of the manuscript that require major revisions.
After reading the manuscript, I did find the title to be a bit misleading. This paper is mostly about NO2. And because NO2 is so wrong in the model (driven by EDGAR NOx), CO2 must be off as well. I suggest a new title focus mostly on NO2 and less on CO2. Further the CO2/NO2 ratio portion of this project is not as insightful as they could have been (see next comment). A new suggested title: Observed and simulated NO2 mixing ratios in Saudi Arabia and Qatar and linkages with CO2
The conclusions from results in Section 3.3 are incomplete. The problem is what is left unsaid, not that something incorrect was stated by the co-authors. Given the small sample size, I don’t think we can say we any certainty that the NOx to CO2 ratios are correct. This is an important point, and not explicitly stated by the authors. For example even a 20% error in this NOx to CO2 ratio has big implications, and I don’t think this study demonstrates NOx to CO2 ratios within that uncertainty bounds. I do agree with the authors that CO2 is incorrectly estimated because NOx is incorrectly estimated, but that does not give any certainty about the NOx-to-CO2 ratio. A longer discussion on the uncertainties in needed in Section 3.3, and also a clarification in the Abstract to note these uncertainties.
Finally, I found the figure labeling and figure captions to be insufficient (see specific suggestions in the line by line comments). At times, it was hard to follow because clear figure labels were missing. This should be an easy fix on your end, but on my end, this made it difficult and time consuming to fully evaluate this manuscript. Further, the authors made a choice to use UTC throughout but this also made the manuscript difficult to evaluate, because I was constantly adding 3 hours, since it is easiest to evaluate NO2 with respect to the local solar hour. For this reason, local time should be used throughout. 10 UTC = 13:00 local, 0 UTC = 3:00 local
Line by line comments are below:
Line 37. 1% compared to 2021?
Line 50. I’d argue it’s the lack of precision and re-visit times of satellite instruments that is the biggest issue (as you discuss in the next sentence). The ground station networks are helpful for validating satellite data, but I don’t see it feasible for 100s of urban areas to have a dense enough network to do an atmospheric inversion. In other words, the goal should be for a better satellite, not for 10s of greenhouse gas instruments in 100s of cities in order to do an inversion. I’d suggest a rephrase here.
Line 82. “Highly uncertain” is probably not the correct phrase here. Maybe say “dependent on ambient reactants and ambient conditions” as you discuss two sentences later. Please rephrase.
Line 122. Never heard of a “”child” domain. Suggest rephrase to “one-way nested domain”
Line 130. Prefer for times throughout to be in local time rather than UTC. Are time in Figure 2 local or UTC? I prefer they be in local time.
Line 150. Is “energy” sector equivalent to power generation? If not, can you clarify how the energy sector is different than other sectors? How does the “energy” sector differ from “industrial”? Please clarify. All NOx emissions come from the combustion of some type of “energy” so I personally do not like this label, but I understand this is the EDGAR labeling convention.
Line 180. At first it was hard to understand what you were saying, but I think you mean that there is no blue shade overlaying some of the red or black circles. Can you state that directly in the text here (and in the Figure 5 caption) referring directly to the colors? Are these power plants not built and EDGAR randomly put them there (can confirm with Google Maps imagery)? What are the lat/lons of these not built or under-emitting power plants? More details here would be helpful.
Line 190. It seems that the scaling factors are an average over the whole domain and not individual sectors. Do individual sectors have different hourly scaling factors? Please be clear about this.
Figure 6. Same aforementioned issue with time. Prefer local time which is 3 hours different than UTC
Line 222. “Instantaneous” is perhaps not the correct word. “Within minutes” or “within seconds” is a better terminology
Figure 8. Which cities/locations are being shown in the panels? This figure is very unclear. Also, text needs to be bigger on the figure, and much of it can be removed. Titles of individual panels are unnecessary. For lat/lons maybe show only the integer values. Use a single colorbar.
Figure 9. Please make it clear directly on the figure that the left panels are the WRF-Chem chemistry simulation, and right is WRF-Chem tracer simulation. Same comment about time.
Figure 9a. Is O3 being produced from NO2 at 3 AM local time? This does not make sense if I am following correctly. NO2 should be going to NO3 nitrate at night, and consuming O3.
Line 270, Clarify that averaging kernel is applied to the model simulation.
Figure 10. Prefer for the colors in the difference plot to be reversed. Usually difference are calculated by: model - obs. The value closer to the “truth” should be the baseline value.
Lines 280-285. Some mention of WRF-Chem being unable to simulate the wind should be mentioned, and you can allude to Figure 2. In the Zone 1, we see a “dipole” indicating that wind direction is not properly simulated by WRF-Chem. Therefore agreement in Zone 1 of the NOx emissions themselves is probably better than what the Figure 10 implies. Some more discussion on this would be helpful.
Line 310. I had hoped the authors would be more explicit with which NOx (and CO2) emissions are missing in addition to what is listed in Table 2. Can you provide a list of missing power plants? And vice versa, the lat/lons for power plants that the inventory says are there, but are either not operating or operating much cleaner than anticipated? In essence, I am requesting a follow-on table to Table 2 that has more detailed information. Future researchers and modelers of Middle East air quality would benefit from this updated information.
Line 317. desertic --> desert
Figure 13. It would be helpful to underlay some geographic features, such as the Riyadh city boundary and/or nearby cities / power plants.
Line 352. Some mention that the NOx-to-CO2 ratio cannot be conclusively determined needs to be mentioned here. An uninformed reader might take liberty to assume that because the deltaNO2/delta CO2 generally lie along the 1-to-1 line that the NOx to CO2 ratio is therefore correct, but I don’t think that can be said with any certainty. The problem is what is left unsaid, not that something incorrect was stated by the co-authors. Hopefully this makes sense. Additional discussion here and in the Abstract is necessary. This is my biggest concern with this manuscript.
Citation: https://doi.org/10.5194/egusphere-2023-2487-RC1 -
RC2: 'Comment on egusphere-2023-2487', Anonymous Referee #2, 21 Dec 2023
This paper investigates use of plumes of NOx and CO2 as observed by TROPOMI and OCO-2/3 respectively in combination with WRF-CHEM modeling to infer emissions. The paper has a number of flaws, is not well organized and does not build appropriately on the very extensive literature using satellite remote sensing observations of NO2 to understand emissions from cities and power plants. I recommend it be rejected.
Most of this paper is about NO2 and I focus my comments on those aspects.
- Retrieval accuracy: There is a significant and sophisticated literature on the role of spatial and temporal resolution in the accuracy of the NO2 column retrieved from space-based observations. I refer the authors to papers from the Cohen group at Berkeley and the Burrows group at Bremen. There are many others.
- Method of analysis: The literature on the use of plumes to infer emissions is extensive. There are too many papers to recap the literature here. It is common in the NO2 plume literature to rotate plumes along the winds to increase the signal to background. The authors choose a method of concentric circles for their analysis that does not seem like it would be as effective. It adds quite a lot of background to an analysis degrading the signal to background ratio. A comparison to the common practice in recent literature is warranted.
- Use of plumes to infer emissions: Most of the recent literature uses an exponentially modified Gaussian to interpret the plumes. Some authors have used days with slow winds to isolate emissions and fast winds to isolate chemical lifetimes.
- Novelty: It is not clear to me that any of the analysis of NO2 in plumes is new. For example Beirle et al. investigates the exact same region: https://www.science.org/doi/10.1126/sciadv.aax9800. And several authors have previously pointed out the utility of space based NO2 for correctly geolocating large emission sources.
- Chemistry and turbulent mixing in plumes: The paper describes and propagates a number of fundamental misunderstandings about the chemistry of nitrogen oxides in plumes. There are three different aspects. The rapid reaction of NO with O3, turbulent mixing with air from the edges of the plume, and oxidative chemistry. These 3 physicochemical aspects of the evolution of a plume operate on different time scales. First, the lifetime of NO and O3 is set by the rate of their reaction. Under typical circumstances (298K, 60ppb O3, less than 1ppb NO) that lifetime of NO (1/e) to reaction with O3 is about 100 seconds. In daytime, the lifetime of NO2 to photolysis is also 100 seconds. Thus, NO and NO2 reach a photostationary state on a time scale of minutes. However, in a plume where NO is in large excess compared to O3, this analysis breaks down as all of the ozone is rapidly consumed. In these circumstances, the relevant lifetime is set by the time scale for reintroducing O3 into the plume by turbulent mixing of air from outside the plume. The oxidative chemistry that reduces NOx occurs on a longer time scale, set by the production rate of OH at high NO2 to VOC ratios and by the NOx to VOC ratio when the two are at concentrations ratios where VOC compete with NO2 for reaction with OH. These ideas have been extensively observed in aircraft experiments downwind of power plumes by, for example, Ryerson et al. from the NOAA group. They have also been studied in wildfires. And of course, there is an extensive literature on urban ozone production. Finally, nighttime chemistry shouldn’t matter much (if at all) for observations at midday.
- WRF-CHEM Modeling: The issues described above all play out in the accuracy of WRF-CHEM modeling. See Valin et al. https://doi.org/10.5194/acp-11-11647-2011, 2011 for one example.
- Overall NO2 assessment: What accuracy is needed for useful analysis of the NO2? A clear statement is needed along with an evaluation of what can be achieved by the method described.
- Presentation: The logic in the paper is hard to follow. There are many figures and what the reader is expected to take from each one is difficult to discern. The maps all have different axes. They should be reformatted to be the same map area for every map in the paper. I recommend describing the approach for a single plume and then extending to a larger suite of plumes.
Citation: https://doi.org/10.5194/egusphere-2023-2487-RC2
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-2487', Anonymous Referee #1, 04 Dec 2023
Overall this paper is scientifically sound. It investigates NO2 in the Saudi Arabia and Qatar regions of the Middle East by comparing satellite to model simulations: a full chemical transport simulation and a passive tracer simulation. Some conclusions are also made about CO2 emissions after the NO2 inter-comparison It is a good advancement and worthy of publication, but there are several areas of the manuscript that require major revisions.
After reading the manuscript, I did find the title to be a bit misleading. This paper is mostly about NO2. And because NO2 is so wrong in the model (driven by EDGAR NOx), CO2 must be off as well. I suggest a new title focus mostly on NO2 and less on CO2. Further the CO2/NO2 ratio portion of this project is not as insightful as they could have been (see next comment). A new suggested title: Observed and simulated NO2 mixing ratios in Saudi Arabia and Qatar and linkages with CO2
The conclusions from results in Section 3.3 are incomplete. The problem is what is left unsaid, not that something incorrect was stated by the co-authors. Given the small sample size, I don’t think we can say we any certainty that the NOx to CO2 ratios are correct. This is an important point, and not explicitly stated by the authors. For example even a 20% error in this NOx to CO2 ratio has big implications, and I don’t think this study demonstrates NOx to CO2 ratios within that uncertainty bounds. I do agree with the authors that CO2 is incorrectly estimated because NOx is incorrectly estimated, but that does not give any certainty about the NOx-to-CO2 ratio. A longer discussion on the uncertainties in needed in Section 3.3, and also a clarification in the Abstract to note these uncertainties.
Finally, I found the figure labeling and figure captions to be insufficient (see specific suggestions in the line by line comments). At times, it was hard to follow because clear figure labels were missing. This should be an easy fix on your end, but on my end, this made it difficult and time consuming to fully evaluate this manuscript. Further, the authors made a choice to use UTC throughout but this also made the manuscript difficult to evaluate, because I was constantly adding 3 hours, since it is easiest to evaluate NO2 with respect to the local solar hour. For this reason, local time should be used throughout. 10 UTC = 13:00 local, 0 UTC = 3:00 local
Line by line comments are below:
Line 37. 1% compared to 2021?
Line 50. I’d argue it’s the lack of precision and re-visit times of satellite instruments that is the biggest issue (as you discuss in the next sentence). The ground station networks are helpful for validating satellite data, but I don’t see it feasible for 100s of urban areas to have a dense enough network to do an atmospheric inversion. In other words, the goal should be for a better satellite, not for 10s of greenhouse gas instruments in 100s of cities in order to do an inversion. I’d suggest a rephrase here.
Line 82. “Highly uncertain” is probably not the correct phrase here. Maybe say “dependent on ambient reactants and ambient conditions” as you discuss two sentences later. Please rephrase.
Line 122. Never heard of a “”child” domain. Suggest rephrase to “one-way nested domain”
Line 130. Prefer for times throughout to be in local time rather than UTC. Are time in Figure 2 local or UTC? I prefer they be in local time.
Line 150. Is “energy” sector equivalent to power generation? If not, can you clarify how the energy sector is different than other sectors? How does the “energy” sector differ from “industrial”? Please clarify. All NOx emissions come from the combustion of some type of “energy” so I personally do not like this label, but I understand this is the EDGAR labeling convention.
Line 180. At first it was hard to understand what you were saying, but I think you mean that there is no blue shade overlaying some of the red or black circles. Can you state that directly in the text here (and in the Figure 5 caption) referring directly to the colors? Are these power plants not built and EDGAR randomly put them there (can confirm with Google Maps imagery)? What are the lat/lons of these not built or under-emitting power plants? More details here would be helpful.
Line 190. It seems that the scaling factors are an average over the whole domain and not individual sectors. Do individual sectors have different hourly scaling factors? Please be clear about this.
Figure 6. Same aforementioned issue with time. Prefer local time which is 3 hours different than UTC
Line 222. “Instantaneous” is perhaps not the correct word. “Within minutes” or “within seconds” is a better terminology
Figure 8. Which cities/locations are being shown in the panels? This figure is very unclear. Also, text needs to be bigger on the figure, and much of it can be removed. Titles of individual panels are unnecessary. For lat/lons maybe show only the integer values. Use a single colorbar.
Figure 9. Please make it clear directly on the figure that the left panels are the WRF-Chem chemistry simulation, and right is WRF-Chem tracer simulation. Same comment about time.
Figure 9a. Is O3 being produced from NO2 at 3 AM local time? This does not make sense if I am following correctly. NO2 should be going to NO3 nitrate at night, and consuming O3.
Line 270, Clarify that averaging kernel is applied to the model simulation.
Figure 10. Prefer for the colors in the difference plot to be reversed. Usually difference are calculated by: model - obs. The value closer to the “truth” should be the baseline value.
Lines 280-285. Some mention of WRF-Chem being unable to simulate the wind should be mentioned, and you can allude to Figure 2. In the Zone 1, we see a “dipole” indicating that wind direction is not properly simulated by WRF-Chem. Therefore agreement in Zone 1 of the NOx emissions themselves is probably better than what the Figure 10 implies. Some more discussion on this would be helpful.
Line 310. I had hoped the authors would be more explicit with which NOx (and CO2) emissions are missing in addition to what is listed in Table 2. Can you provide a list of missing power plants? And vice versa, the lat/lons for power plants that the inventory says are there, but are either not operating or operating much cleaner than anticipated? In essence, I am requesting a follow-on table to Table 2 that has more detailed information. Future researchers and modelers of Middle East air quality would benefit from this updated information.
Line 317. desertic --> desert
Figure 13. It would be helpful to underlay some geographic features, such as the Riyadh city boundary and/or nearby cities / power plants.
Line 352. Some mention that the NOx-to-CO2 ratio cannot be conclusively determined needs to be mentioned here. An uninformed reader might take liberty to assume that because the deltaNO2/delta CO2 generally lie along the 1-to-1 line that the NOx to CO2 ratio is therefore correct, but I don’t think that can be said with any certainty. The problem is what is left unsaid, not that something incorrect was stated by the co-authors. Hopefully this makes sense. Additional discussion here and in the Abstract is necessary. This is my biggest concern with this manuscript.
Citation: https://doi.org/10.5194/egusphere-2023-2487-RC1 -
RC2: 'Comment on egusphere-2023-2487', Anonymous Referee #2, 21 Dec 2023
This paper investigates use of plumes of NOx and CO2 as observed by TROPOMI and OCO-2/3 respectively in combination with WRF-CHEM modeling to infer emissions. The paper has a number of flaws, is not well organized and does not build appropriately on the very extensive literature using satellite remote sensing observations of NO2 to understand emissions from cities and power plants. I recommend it be rejected.
Most of this paper is about NO2 and I focus my comments on those aspects.
- Retrieval accuracy: There is a significant and sophisticated literature on the role of spatial and temporal resolution in the accuracy of the NO2 column retrieved from space-based observations. I refer the authors to papers from the Cohen group at Berkeley and the Burrows group at Bremen. There are many others.
- Method of analysis: The literature on the use of plumes to infer emissions is extensive. There are too many papers to recap the literature here. It is common in the NO2 plume literature to rotate plumes along the winds to increase the signal to background. The authors choose a method of concentric circles for their analysis that does not seem like it would be as effective. It adds quite a lot of background to an analysis degrading the signal to background ratio. A comparison to the common practice in recent literature is warranted.
- Use of plumes to infer emissions: Most of the recent literature uses an exponentially modified Gaussian to interpret the plumes. Some authors have used days with slow winds to isolate emissions and fast winds to isolate chemical lifetimes.
- Novelty: It is not clear to me that any of the analysis of NO2 in plumes is new. For example Beirle et al. investigates the exact same region: https://www.science.org/doi/10.1126/sciadv.aax9800. And several authors have previously pointed out the utility of space based NO2 for correctly geolocating large emission sources.
- Chemistry and turbulent mixing in plumes: The paper describes and propagates a number of fundamental misunderstandings about the chemistry of nitrogen oxides in plumes. There are three different aspects. The rapid reaction of NO with O3, turbulent mixing with air from the edges of the plume, and oxidative chemistry. These 3 physicochemical aspects of the evolution of a plume operate on different time scales. First, the lifetime of NO and O3 is set by the rate of their reaction. Under typical circumstances (298K, 60ppb O3, less than 1ppb NO) that lifetime of NO (1/e) to reaction with O3 is about 100 seconds. In daytime, the lifetime of NO2 to photolysis is also 100 seconds. Thus, NO and NO2 reach a photostationary state on a time scale of minutes. However, in a plume where NO is in large excess compared to O3, this analysis breaks down as all of the ozone is rapidly consumed. In these circumstances, the relevant lifetime is set by the time scale for reintroducing O3 into the plume by turbulent mixing of air from outside the plume. The oxidative chemistry that reduces NOx occurs on a longer time scale, set by the production rate of OH at high NO2 to VOC ratios and by the NOx to VOC ratio when the two are at concentrations ratios where VOC compete with NO2 for reaction with OH. These ideas have been extensively observed in aircraft experiments downwind of power plumes by, for example, Ryerson et al. from the NOAA group. They have also been studied in wildfires. And of course, there is an extensive literature on urban ozone production. Finally, nighttime chemistry shouldn’t matter much (if at all) for observations at midday.
- WRF-CHEM Modeling: The issues described above all play out in the accuracy of WRF-CHEM modeling. See Valin et al. https://doi.org/10.5194/acp-11-11647-2011, 2011 for one example.
- Overall NO2 assessment: What accuracy is needed for useful analysis of the NO2? A clear statement is needed along with an evaluation of what can be achieved by the method described.
- Presentation: The logic in the paper is hard to follow. There are many figures and what the reader is expected to take from each one is difficult to discern. The maps all have different axes. They should be reformatted to be the same map area for every map in the paper. I recommend describing the approach for a single plume and then extending to a larger suite of plumes.
Citation: https://doi.org/10.5194/egusphere-2023-2487-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 307 | 146 | 24 | 477 | 13 | 19 |
- HTML: 307
- PDF: 146
- XML: 24
- Total: 477
- BibTeX: 13
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1