General Comments

The manuscript titled “TROPOMI NO₂ for urban and polluted areas globally from 2019 to 2024” presents a comprehensive analysis on NO₂ VCD changes in cities worldwide. It details the contrasts in NO₂ trends across cities, and potential drivers the embedded anthropogenic emissions, including environmental regulation, local economic growth and regional conflicts. Although the study Illustrates the latest evolution of global air pollution, and offers a valuable reference for future research, the manuscript, in its current form, contains several critical issues that warrant major revisions. Therefore, I recommend reconsideration for its publication after the authors adequately address the concerns outlined below.

The current manuscript lacks a discussion of the uncertainty of NO₂ VCDs and its potential impacts on the conclusions. This information is crucial for distinguishing trends from interannual fluctuations, and for separating meaningful emission changes from the noise inherent in satellite retrievals. However, uncertainty considerations are absent from the main text and figures. In addition, further validation of the NO₂ background values is necessary, along with sensitivity tests (e.g. evaluating the results using different percentile thresholds in the background selection). The interannual variability of the background should also be evaluated (e.g. in Fig. 12), as this could influence the interpretation of relative changes in VCD enhancements. Moreover, the spatial consistency of the background should be examined, particularly in regions where adjacent cities are expected to share similar background levels.

The manuscript includes several qualitative descriptions that are not supported by sufficient validation or statistical testing. For instance, it states that there is an accelerated decreasing trend in NO₂ VCDs in both China and European countries. However, given that the dataset used in this study begins in 2019, the time range may be too short to detect or validate such trend acceleration. Similarly, the manuscript mentions an accelerated NO₂ increase over Moscow in early 2022. Yet, Fig. S9 appears to show only a brief, anomalous spike in NO₂ VCDs, followed by a return to typical levels. These interpretations, as currently presented, are questionable and require rigorous statistical validations.

The manuscript appears to insufficiently account for the effects of seasonality on NO₂ VCDs. Given the strong seasonal variation in NOx lifetime, particularly the longer lifetime during winter, NO₂ VCDs in colder months can disproportionately influence interannual trends if seasonality is not properly addressed. However, the manuscript lacks adequate discussion or correction for these seasonal effects. Moreover, there appears to be a mischaracterization of seasons between the Northern and Southern Hemispheres. For instance, the manuscript uses data from the same calendar months to represent winter conditions in both Asia and Oceania. This approach is problematic, as most cities in Oceania are located in the Southern Hemisphere, where the seasonal cycle is inverted. As a result, the analysis may misrepresent seasonal trends in these regions, and further clarification or adjustment is necessary.

Specific Comments

Page 2, Line 56-67: I would suggest to include a brief overview about NO₂ VCD changes in India, Oceania and Africa here, since these regions also play important roles in this study.

Page 2, Line 59: “x” --> “×”. Check throughout the manuscript.

Page 4, Line 97: It should be explained why GHS-SMOD boundaries are used rather than administrative city boundaries, and clarify whether this choice affects the results.

Page 4, Line 111: According the latest ATBD (2.8.0, 2024-11-18– released) for TROPOMI NO₂, the nadir ground pixel dimensions were 7.0 × 3.5 km² before 6 August 2019. The data description here is inaccurate.

Page 6, line 142: Sensitivity tests should be conducted to assess the impact of using different percentile values in background selection. In addition, validation is needed. For example, by examining whether background values are consistent across adjacent cities.

Page 6, Line 157: The claimed acceleration in the decreasing trend requires statistical validation; otherwise, such descriptions might be just removed. (Also, for the descriptions on Page 10, Line 239, Page 10, Line 221, and Page 14, Line 310)

Page 7, Line 164: Please clarify the definition of the mining regions (including A, C in Fig. 2; B in Fig. 4; D, E in Fig. S4; and G, F, H, I in Fig. 6).

Page 7, Line 166: The texts in Fig. S3 are not clear.

Page 8, Figure 3: The information of NO₂ VCD uncertainty and significance tests on the regression is missing. In addition, please ensure consistency of significant figures or decimal precision for all numerical data throughout the manuscript.

Page 9, Line 197: Please provide the specific number and proportion (“Nearly all”).

Page 9, Figure 4: I would suggest standardizing the formatting of units throughout the manuscript for consistency.

Page 10, Line 232: What is the term “largest” referring to or being compared against in this description? (other cities or other land type? Also, for the descriptions on Page 11, Line 248, Page 11, Line 253-254, Page 12, Line 263-264 and Page 12, Line 270)

Page 10, Line 236: Please provide the specific number.

Page 12, Section 4: I would suggest to integrate Section 3 and Section 4.

Page 13, Figure 7: The figure legend could be further improved to enhance readability.

Page 14, Line 311: Typo.

Page 14, Line 311: The abnormally high NO₂ VCD values require further examination to exclude artifacts, including applying data filters based on Level-2 QA flags. It should also be verified whether any spurious outliers affect the averaging process.

Page 15, Figure 8: The figure labels/text are not clear.

Page 17, Line 349-350: Such causal relationships require careful validation. I recommend revising the statement here.

Page 18, Line 376: It is not immediately clear why population-weighted VCDs are preferred here over direct NO₂ VCDs for me. Would directly showing NO₂ VCDs make major differences?

Page 21, Line 433: Since the comparison here is based on the relative changes of VCDs and emissions with respect to 2019, it is hard to conclude that emissions are underestimated. At most, it may suggest a possible underestimation in the emission trend. (Also for Page 24, Line 481)

Page 21, Line 435: Impacts of uncertainty in VCD background need to be quantified.

Page 21, Line 435: The mean difference is likely underestimated due to the inclusion of 2019.

Page 22, Line 451: There appears to be a mischaracterization of seasons between the Northern and Southern Hemispheres, since Asia and Oceania are shown together in Fig. 13.

Page 23, Figure 13: Is the sharp increase during the winter of 2022 primarily driven by anomalously high values over Russia? If so, the authors should consider presenting additional results with Russia excluded. Intuitively, I find that this sharp increase appears inconsistent with Fig. 9c, where most cities do not show a similar increase in 2022.

Page 24: Line 481: Discussion about the impacts of NOx chemistry and its seasonality should be included.

Page 24, Line 491-492 (“tall-stack sources”): Could the authors provide supporting references for this statement?

Comments from Anonymous Referee #2

General comments:

This paper presents a comprehensive assessment of urban NO₂ changes worldwide from 2019 to 2024 using TROPOMI NO₂ VCD observations. Differences in NO₂ VCD changes over populous cities and broader areas are disclosed, probably driven by anthropogenically induced factors such as urbanization, industrial activities, government interventions, and societal disruptions. The paper also attempts to quantify the influence of background NO₂ and NO₂ seasonal variability on the trend analysis. The research topic fits in the scope of ACP, and the manuscript is already in good shape. I recommend its publication after the authors address the following comments.

Specific comments:

Line 1-2: The current title is a little general and can not convey the key point of this research. I would suggest to improve the title by using the key conclusion of this study, which can better draw readers’ attention.

Line 39-40: The remote sensing method not only relies on spectrometers aboard satellites to infer vertical columns, but also can infer vertical profiles using ground-based instruments. The statement should incorporate the profile retrieval to ensure a more comprehensive description.

Line 68-75: This paragraph only describes the different methods used to characterize the urban extent. The authors should add a brief discussion about the pros and cons of these methods, and highlight the advantage(s) of the GHS-SMOD, which is used in this study.

Section 2.2: Please briefly describe the uncertainty of the NO₂ VCD product used here.

Section 2.2.1: The structure here is a little inappropriate because there is only one sub-section. I would suggest to merge Section 2.2.1 and Section 2.2 to one section.

Line 141-142: Please justify the definition of the background NO₂ concentration here, and provide the sensitivity of the results in Section 6 to the choice of the percentile.

Section 2.4: Please briefly describe the uncertainty of the EDGARv8.1 NO_x emissions.

Line 157-158: Is the statement “the decrease accelerated after the onset of the COVID-19 pandemic” one of the findings of this study, or a knowledge cited from other papers? If the former is true, please provide a quantitative discussion to support this point; if the latter is true, please provide supporting references.

Figure 3, Figure 8 and Figure S6: please provide the confidence level of each regression to clarify the statistical significance of the characterized trends.

Line 305-306: Please provide a quantitative discussion to support the statement “The observed annual decreases in these East Asian cities were primarily driven by decreases during the winter months”.

Line 310: It is difficult to see that the increasing trend in Moscow accelerated in early 2022 from Figure S9, except that NO₂ VCDs in winter time jumped to a higher level. Please provide a quantitative discussion to demonstrate the acceleration.

Section 6: please provide a summary of this section, i.e., to what extent the influence of background NO₂ and seasonal variability can be on the analysis of urban NO₂ trends presented above?

Line 446-456 and line 482: It should be careful to define May – September and November – March as either “warm” or “cold” months, given the different hemispheres in which the continents are located. The interpretation of the results for Asia and Oceania is problematic, because May – September is summer time for Asia but is winter time for Oceania, while November – March is winter time for Asia but is summer time for Oceania. Please revise the discussions here and in Section 7.

Technical comments:

Line 50: Please check through the manuscript and replace the “x” with a times symbol at corresponding places.

Line 57: “SCHIAMACY” should be “SCIAMACHY”.

Line 63: The statement “NO₂ concentrations increased through roughly 2005” is a little confusing. I would suggest to rephrase this sentence to make it clearer.

Line 94: It is better to be 1 × 1 km².

Line 127: Please add a period after “approach”.

Line 142: It is clearer to extend “UC” to “urban cluster”.

Line 148: Do you mean “Sec. 2.2.1” here?

Line 183: “select” to “selected”.

Line 306: There is no Figure 7d.

Huber and co-workers present a global scale analysis of NO₂ trends over large cities based on satellite TROPOMI column data. The study covers 6 years of observations, from 2019 to 2024, and includes the NO₂ sudden drop and subsequent rebound in many cities due to the COVID-19 disruptions, as well as air pollution changes in response to policy regulations, economic growth, and armed conflicts. Based on population-weighted annual columns, cities of Asia, Europe and North America show a NO₂ decline, whereas African cities show a moderate increase. Although not strictly comparable, the observed trends and the anthropogenic NOx emission trends from the EDGARv8.1 bottom-up inventory give an indication that emission trends are underestimated in inventories, in particular in the Southern Hemisphere. The manuscript is well written but too lengthy, and the figures seem to support the drawn conclusions.

Although I appreciate the importance of the topic, there are several important issues with the paper. First, the trend analysis lacks an estimate of the statistical significance. I have serious doubts that the calculated trends in most cities are different from zero given the uncertainty in the satellite measurement and the limited number of available observations per city. The findings are therefore not sound. It is crucial to include a robust uncertainty analysis, and I argue that part of the conclusions will considerably change.  Second, a major limitation is that the period (6 years) of the analysis is short for attributing the observed trends to anthropogenic changes only, as short-term meteorological variability (not accounted here) might partly explain the observed trends. Furthermore, the fact that the study period includes the COVID-19 lockdowns makes the derivation of trends even more uncertain. I recommend extending the analysis to a longer period, by combining OMI and TROPOMI datasets, both having the same overpass (e.g. Glissenaar et al. https://doi.org/10.5194/essd-2024-616). Third, the seasonal variation of NO₂ data is not adequately discussed. It is unclear whether a threshold of available observations per month was used and why the May-to-September trends are not compatible with the November-to-March trends in some regions, e.g. Europe. Overall, the discussion is poor, and the scientific content is limited. To my opinion, the main problem of the manuscript is the lack of innovation and inadequate analysis. Therefore, I cannot accept this manuscript for publication in ACP.

Specific comments:

-    L.112-116: Any data filtering used?

-    L. 57: Read SCIAMACHY

-    Why not using CEDS (instead of EDGAR)? The latest CEDS version covers 2023.

-    A table summarizing the percentage changes would be useful

-    Figs 2, 4, 5, 6 are similar and could be grouped in 2

-    Unclear whether the strong column changes in Quito and Santos are due to changes in anthropogenic activities, or to data issues.  

-    For which among the cities of Fig.7 are the trends significant?