the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Sep 2024
Estimating the variability of NOx emissions from Wuhan with TROPOMI NO2 data during 2018 to 2023
Abstract. Accurate NOx emission estimates are required to better understand air pollution, investigate the effectiveness of emission restrictions, and develop effective emission control strategies. This study investigates and demonstrates the ability of the superposition column model in combination with TROPOMI tropospheric NO2 column data to estimate city-scale NOx emissions and lifetimes and their variabilities. Using the recently improved TROPOMI tropospheric NO2 column product (v2.4–2.6), we derive daily NOx emissions and lifetimes over the city of Wuhan for 335 clear sky days between May 2018 and December 2023. We find a slight weekend reduction in NOx emission with a weekend-to-weekday ratio of 0.95 and a small seasonal variation of NOx emissions over Wuhan with a summer-to-winter emission ratio of 0.87. We calculate a steady decline of NOx emissions from 2019 to 2023, and the emission in 2023 is ~15 % below the 2019 level, indicating the success of the emission control strategy. The estimated NOx lifetimes range from 0.8 h (summer) to 5.3 h (winter), with an average of 2.6 h. Meanwhile, our method shows ~30 % lower NOx lifetimes for fast wind (> 7 m s−1) speed. The superposition model method results in ~10 % lower estimation of NOx emissions when the wind direction is from distinct upwind NO2 hotspots compared to other wind directions, indicating the need to improve the approach for cities that are not relatively isolated pollution hotspots. The results of this work nevertheless confirm the strength of the superposition column model in estimating urban NOx emissions with reasonable accuracy.
- Preprint
(1240 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2641', Anonymous Referee #1, 03 Oct 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-2641/egusphere-2024-2641-RC1-supplement.pdf
-
AC3: 'Reply on RC1', Qianqian Zhang, 07 Dec 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-2641/egusphere-2024-2641-AC3-supplement.pdf
-
AC3: 'Reply on RC1', Qianqian Zhang, 07 Dec 2024
-
RC2: 'Comment on egusphere-2024-2641', Anonymous Referee #2, 22 Oct 2024
The manuscript presents an interesting investigation using TROPOMI NO2 column data in combined with superposition column model to estimate the emission and lifetime of NOx. Specifically, the study focuses on the derive the NOx emissions and lifetime over Wuhan for 335 clear sky days between May 2018 and December 2023, with the variability of emissions being evaluated to investigate the effectiveness of the emission control strategy. There are some interesting findings resulted from the study. However, the reviewer has some concerns about the novelty of the methods and the significance of the results. See detailed comments below.
Major comments:
- This paper looks like an extension of the authors’ ACP paper published in 2023. Similar methods are applied to the TROPOMI data (with version change though) over the same region, and the main difference is that this study extends the study period from 2019-2020 to 2019-2023. Because of the overlap with the authors’ previous study, the reviewer is concerned about the novelty of this manuscript, especially since the technical approach has been proposed in their 2023 paper. The authors should clarify in the introduction how this manuscript differs from the previous study, and what would be the novelty of this study.
- It’s unclear how the NOx lifetime is calculated in GEOS-Chem. The model approach gives an effective lifetime of the entire plume, but the actual chemical lifetime can vary from source to downwind. The effective lifetime can be further confounded by mixing of plumes from multiple directions. I’d suggest the authors clarify the meaning of lifetime in the manuscript, and the limitations of using the model approach to estimate NOx lifetime.
- The authors showed strong dependence of the emissions and lifetimes on wind field, which does not necessarily mean the NOx emissions vary with wind, but rather due to the limitation of the model and the way the model defines background NO2. This is not a scientific finding, so I think it’s better to be included in the uncertainty discussion.
- Figure 4: Please add error bars to this figure to reflect day-to-day variability. Considering the large variability of emissions and the uncertainties of the model and satellite observations, is the weekly cycle statistically significant?
- Section 3.2.3: Considering the large uncertainties of satellite retrievals on daily basis and the potential influences of winds, I think performing the EMG approach or superposition model over the long-term average data may actually be a better choice for studying the inter-annual variability. I don’t see any values added from performing the approach on daily basis. I suggest the authors clarify why it’s necessary to calculate daily emissions here.
Minor Comments:
Line 48: Please change "ultraviolet/visible" to "ultraviolet (UV)/visible," and use the acronym "UV" for subsequent mentions throughout the manuscript. (Line 85)
Figure 1b: Better to show the rotated plume with wind direction as x axis, and cross-wind direction as y axis.
Line 65: EMG model has been used to estimate episodic fire NOx emissions, which does not need long-term average [1]. The key is to find distinguishable plumes from TROPOMI data.
Line 136: For the title of Figure 1a, please change "origional" to "original." Additionally, could you indicate the location of Wuhan city on the map and include the corresponding radius value?
References:
[1] Jin X, Zhu Q, Cohen RC. Direct estimates of biomass burning NOx emissions and lifetimes using daily observations from TROPOMI. Atmos Chem Phys 2021; 21: 15569–15587.
Citation: https://doi.org/10.5194/egusphere-2024-2641-RC2 -
AC1: 'Reply on RC2', Qianqian Zhang, 07 Dec 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-2641/egusphere-2024-2641-AC1-supplement.pdf
-
RC3: 'Comment on egusphere-2024-2641', Anonymous Referee #3, 23 Oct 2024
Review of the Manuscript „Estimating the variability of NOx emissions from Wuhan with
TROPOMI NO2 data during 2018 to 2023“
This paper by Zhang et al. presents NOx emission and lifetime estimates for Wuhan based on TROPOMI tropospheric NO2 VCDs from mid 2018 to 2023 and a superposition column model introduced in previous studies. Seasonal patterns, the weekly cycle, interannual variability, and dependence on wind conditions are investigated and compared to similar studies. Estimated monthly NOx emissions are compared to the EDGAR and ABACAS emission inventory.
General comments:
The manuscript is a follow-up study of the Zhang et al. (2023) study. Due to a longer dataset, seasonal patterns, the weekly cycle, and the interannual variability can be investigated. The methodology is based on Lorente et al. (2019) and Zhang et al. (2023), however, some adjustments have been made.
Some parts of the methodology are hard to follow, reading only this manuscript without knowing Zhang et al. (2023) and would benefit from a bit more details (see also comments in the specific comment part below). How are data rotated and how is the mean wind calculated (over which area)? Definition of emissions and lifetime? Why an area of 90km and why was it changed compared to Zhang et al. (2023)?I am concerned that some decisions influence the emission estimates, especially the investigated seasonal patterns (see also comments in the specific comment part below):
How does the ABACAS emission inventory filter influence the results? Is the computation of NOx emissions, which seems to be restricted by a priori emissions, dampening the estimated emissions? How much is the method depending on the emission inventory data?
Is the bias correction factor of 1.2 valid for all seasons?
How would seasonal patterns change using a daily NOx/NO2 ratio instead of a fixed value of 1.26? The loss rate of NOx (k in Eq 1) is based on the fixed NOx/NO2 ratio and does not consider the temperature dependency of the NO2 and OH reaction.
I suggest publication if the raised issues are addressed.
Specific comments:
L39: You provide Goldberg et al. (2019) and Zhang et al. (2021) as references for key information about nitrogen oxides. I think it is more appropriate to cite the references which are cited in these references, e.g., Jacob, D. J. Introduction to Atmospheric Chemistry; Princeton University Press, 1999.
L51, L53, L56: You already provide several references for the different emission estimation methods and applications; since these are, however, only some examples, I would change these parts to (e.g., reference1, reference2, …)
L58: Beirle et al. (2011) have not rotated the NO2 maps, they divided the OMI data into wind sectors based on the present wind direction of the individual measurement and estimated emissions for the eight defined wind sectors. Rotation was first introduced by Pommier et al. (2013) and Valin et al. (2013), which you mention also in L62.
L60-61: Since Beirle et al. (2011) have not only estimated emissions for cities, I would suggest deleting the word city in “city NOx emissions” and replacing city with source in “over the city and its decay downwind of the city.”
L65-66: “relatively large study area”, I understand that this is meant probably in comparison to the method you use with a diameter of 90 km, still the EMG method is already possible for individual cities and powerplants. Maybe you can clarify this to avoid confusion.
You wrote that the EMG model is limited to calculating emissions only for data averaged over a longer time period. However, this is more related to the quality of the satellite product used, e.g. Goldberg et al. (2019) showed that using the EMG method with TROPOMI NO2 observations also single overpasses can deliver valuable results, which you are also mentioning in the next paragraph. You can change the sentence to something like “…and with OMI data it is limited to calculating mean NOx emissions from observations over longer time periods, like some years.”.L69: “Lorente et al. (2019) narrowed down the study area to the domain of one city”. Emission estimates for individual cities were already possible with OMI data and the EMG method.
L73: Is Valin et al. (2013) here the right reference, I think the superposition column model is not discussed in Valin et al. (2013), shouldn’t it be Lorente et al. (2019) or maybe both?
L87: “unprecedented nadir spatial resolution” Since the resolution of TEMPO with 2km x 4.5km is better than TROPOMI’s resolution, I suggest deleting “unprecedented”
L87: Since you also use data before 6 August 2019, briefly mention the spatial resolution before the change.
L90-92: You wrote that “The version 2.3.1 includes a different treatment of the surface albedo compared to earlier versions, which led to a 10~15>% increase of tropospheric NO2 columns over polluted scenes.”
This is misleading, the surface albedo in the NO2 window (OMI LER) and for the cloud product (GOME-2 LER) is replaced with the TROPOMI DLER in v2.4, which you also describe in the following sentences. The main changes in v2.3.1 compared to v1.x are the switch to the FRESCO-wide cloud product and a correction of the surface albedo for cloud-free scenes (only for specific scenes with cloud fractions < 0 and > 1). All changes together result in tropospheric NO2 VCDs that are 10-40% larger than in v1.x.
L105: You use a scale factor of 1.2 to correct the low bias of the TROPOMI data based on the ground-based validation with the Xianghe station. Can you comment on possible differences (bias variations) between different seasons, might this influence your seasonal investigations?
L112: Why have you decided on the 950hPa level?
L122: Do you use daily OH concentrations, please clarify.
L124: You decided to use a fixed value of 1.26 for the NOx/NO2 ratio. You mention that it varies less than 10% in season. Can you show the seasonal variation of the NOx/NO2 ratio over the year? I think this is especially relevant as you are investigating seasonal patterns.
L128: Information missing about the ABACAS emission inventory. Is it annual or monthly data, based on which year?
L130 & Fig. 2: EDGAR provides not only annual but also monthly time series for 2018 NOx, did you have a look at these? Relevant for L212.
Section 2.4: You write you rotate with the mean wind direction. How is the mean wind calculated, over which area? What is rotated? Is there a difference between the wind errors in Fig 1(a) and (b)? How do you determine the diameter of your circle?
Equation 1: Use points for multiplication signs instead of crosses.
L154-157: k represents the loss rate of NOx. You use a fixed value for the rate constant k’ between NO2 and OH. How large is the seasonality due to the temperature dependency of this reaction? How large is the seasonal variation of the NOx/NO2 ratio over the year (see comment above)? Relevant for section 3.2.2 about seasonal patterns:
Equation 3: for x >xi (?) missing
L163: What do you mean with “upend point”, at the upwind end point of the grid/city/circle?
L166: You write you use the monthly instead of the daily noon time mean OH concentration, can this create an issue in the weekly cycle investigation as the monthly mean is dominated by weekdays?
L178: You write clear skies here and also at other parts of the manuscript, but you mean for cloud radiance fractions < 0.5, which means there is not always clear sky.
L179: You remove overpasses with inhomogeneous wind fields and days with estimated NOx emissions beyond 0.5－1.5 times the ABACAS bottom-up emissions. First, I don’t understand the reason for the ABACAS filter, and second, can you mention how many days or overpasses are filtered for each of the filters? You only mention how much it is in total. Do you maybe filter high emission days, especially in winter, which results in dampened seasonal patterns?
L197: Why was the area changed between Zhang et al. (2023) and this study?
L201: Change November 2020 to January 2020.
Figure 2 and text: y-labels say “noontime NOx emissions”, which is true for TROPOMI, but I think not for EDGAR and ABACAS. Please correct this and also clarify in the text that the emission inventory emissions are not around noon time.
L240: Possible explanations for deviations with Lange et al. (2022): Larger area in Lange et al., this study limited to city center, maybe different behaviors and sources in the different urban/suburban areas? See comment L166, monthly OH concentrations dampening weekly cycle? Fixed NOx/NO2 conversion factor in this study compared to daily conversion factors in Lange et al.
Section 3.2.2 Seasonal pattern: You see a much more dampened seasonal emission pattern than Lange et al. Possible explanations for deviations:
See comment above: You use a fixed value for the rate constant k’ between NO2 and OH. How large is the seasonality due to the temperature dependency of this reaction? How large is the seasonal variation of the NOx/NO2 ratio over the year? Different areas, sampling issues due to a shorter period in Lange et al., different TROPOMI NO2 product versions have seasonal bias differences (van Geffen et al. 2022), bias correction used in this study.L263/264: Any ideas for the differences in summer lifetime between Zhang et al. (2023) and this study?
L272: You write that the computation of NOx emissions is restricted by a priori emissions. Is this restriction dampening your estimated emissions and a possible explanation for differences with Lange et al.? See also L179 (comment above) and L388-390 in your manuscript.
L298: Any ideas for large deviations between Lonsdale and Sun (2023) and this study, which kind of method is used in Lonsdale and Sun (2023)?
Section 3.3 Wind field dependence: You start using the term chemical lifetime in this section, I think until now you only used the term lifetime. How do you calculate lifetimes? Usually, lifetimes estimated by these kinds of models are defined as effective lifetimes (see, e.g., Beirle et al.) as they include effects of deposition, chemical conversion, and wind advection, which you are illustrating here also for your model. I think you should clarify the differences between chemical and effective lifetime in your text.
Figure 7 and Lines following L 329: I have issues understanding this figure and the text. Explanations for what is visible in the Figure are missing in the caption and also in the text below. Is this one day, averaged for several days? How are the red, yellow, green, pink, and purple curves determined? The text in L332 says that the fitted emissions are basically in line with those from the bottom-up emissions, but in the figure differences are quite large, especially close to the city center.
L357: Lange et al. are not using OH concentrations for their estimates and are not providing an uncertainty for OH.
Technical corrections:
L47: Divide the sentence into two sentences. “….ultraviolet/visible spectrum. Various satellite instruments…”
L50: Misleading, change to something like “Limited by the coarse spatial resolution of the early instruments, researchers …”
L55: when the target is an individual city.
L61: Change model to method.
L63: To avoid confusion, I would change it to “The EMG model has been first applied to OMI NO2 data…”
L70: Change to: “for daily NOx emission estimates”
L72: Change to: “avoids the bias caused by using the averaged NO2 columns in the nonlinear system”
L81: Change conclusion to concluding
L86: I suggest replacing “TROPOMI observes NO2 at 405-465nm of the UV-visible spectral band…” maybe with something like “TROPOMI NO2 columns are retrieved in the spectral range from 405-465nm…”
L95/96: Change DLER to “the DLER”.
L96/97: Check the sentence starting with “The v2.4.0 version.” The version is twice, and in general, it is not good to read.
L125: Change to: An initial guess…
L161: Missing word: and added/combined/… with the contribution from the background
L173: I suggest changing “dilute” with “reduce.”
L202: an ad hoc bias correction factor
L210/211: I would first name EDGAR, then ABACAS, following the logic of Fig. 2.
L230: Change to “NOx emission estimates.”
L237: Replace “minimum” with “reductions” and add “for Wuhan”
Fig 3 caption: “The number of valid measurement days for each day of the week is listed in the plot”
L260: Do you mean dominated instead of determined?
L279: Split into two sentences: “…2024a). We find a similar…”
L280: Change “dramatic changes” to “strong reductions”
L283: You write “We have also found” and give a reference to Zhang et al. (2023) at the end of the sentence. Do you mean “Zhang et al. (2023) have also found”?
Caption Figure 5: Two times NOx in the first sentence. Change dash to dashed. Split the second sentence into two sentences.
Citation: https://doi.org/10.5194/egusphere-2024-2641-RC3 -
AC2: 'Reply on RC3', Qianqian Zhang, 07 Dec 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-2641/egusphere-2024-2641-AC2-supplement.pdf
-
AC2: 'Reply on RC3', Qianqian Zhang, 07 Dec 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 184 | 44 | 154 | 382 | 3 | 3 |
- HTML: 184
- PDF: 44
- XML: 154
- Total: 382
- BibTeX: 3
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1