the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Sep 2025
Atmospheric CO2 dynamics in a coastal megacity: spatiotemporal patterns, sea-land breeze impacts, and anthropogenic-biogenic emission partitioning
Abstract. Accurate quantification of urban carbon dioxide (CO2) emissions is essential for evaluating the efficacy of urban climate mitigation policies. However, the complex interplay of anthropogenic emissions, biogenic fluxes, and meteorological processes in coastal megacities poses significant challenges to characterizing urban CO2 dynamics. To address this, we present an observation-based framework that integrates high-precision CO2 monitoring, meteorological analyses, and ΔCO/ΔCO2 ratios (Rco) to resolve spatiotemporal CO2 variations, quantify sea-land breeze (SLB) effects, and partition anthropogenic and biogenic contributions. Applied in Guangzhou, a coastal megacity, our approach captures a pronounced urban–rural gradient. The coastal site shows the largest seasonal amplitude (25.63 ppm), resulting from wintertime transport of urban emissions and summertime inflow of marine air. Diurnally, suburban CO2 variations are dominated by biogenic activity (summer amplitude: 39.90 ppm), while urban signals reflect anthropogenic influence. SLB generally reduces coastal CO2 by 5.87 ppm but leads to a summer accumulation (+2.08 ppm) under stable, low-wind conditions with shallow boundary layers. Regression-derived Rco values (urban: 7.45 ± 1.38 ppb ppmā»1) reflect improved combustion efficiency linked to clean-air policies. Importantly, our combined observational, modeling, and Rco framework reveals that biogenic fluxes offset 60.17 % of anthropogenic CO2 emissions during summer afternoons. The framework is validated against emission inventories, Normalized Difference Vegetation Index data, and independent studies, demonstrating its robustness. This study enhances process-oriented understanding of coastal carbon cycling and underscores the integration of meteorological and biospheric dynamics in urban CO2 assessments.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2294 KB)
-
Supplement
(2112 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2294 KB) - Metadata XML
-
Supplement
(2112 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-3215', Anonymous Referee #2, 21 Oct 2025
- AC1: 'Reply on RC1', jinwen zhang, 19 Jan 2026
-
RC2: 'Comment on egusphere-2025-3215', Anonymous Referee #3, 20 Dec 2025
This study analyzes COā and CO concentration variations over 1 year and 9 months at three sites in Guangzhouāa coastal megacityāexamining their relationship with land-sea breezes. Using backward trajectory footprint modeling, it quantifies fossil fuel and biogenic contributions. Given the scarcity of direct COā observations in this region, these findings offer valuable insights into carbon sources/sinks in coastal southern Chinese megacities. The methodology is overall sound, and the work merits publication inĀ Atmospheric Chemistry and Physics. However, the introduction of research background, uncertainty analysis in source partition, the robustness of the results, and the depth of discussion could be further improved:
Ā
Specific comments:
Introduction Section: The introduction could benefit from restructuring to enhance its logical flow. The rationale for reporting urban-scale COā concentrationsāparticularly the need to clarify carbon sources and sinksāshould be more explicitly emphasized. While the study relies heavily on the COā/CO ratio and footprint modeling, there is limited discussion on the inherent uncertainties of these methods or how they compare with alternative approaches, such as 3-D atmospheric inversions. Introducing these methodological considerations would strengthen the foundation for the work.
Ā
Section 2.5: TheĀ a prioriĀ emission inventory utilized in the study should be better specified, which inventory? What spatial resolution for what year?
Ā
Line 219ļ¼for footprint simulation, using 500 particles over a 72-hour simulation period appears rather small. In this setting, the particle count per time step and per grid point is extremely low, potentially introducing substantial uncertainty in footprint estimation. It would be valuable to test and report the sensitivity of results to a larger number of particles (1000-10000) to ensure robustness.
Ā
Line 225: The statement that "footprint uncertainties are neglected under the assumption of unbiased atmospheric transport" seems to lack justification. Could the authors provide references supporting this choice or acknowledge its limitations?
Ā
Fig 3A:Ā Consider revising the x-axis label to "JanuaryāDecember" for clarity.
Ā
Lines 255ā265:Ā The urbanāsuburban differences highlighted here are insightful. Adding a figure to illustrate this spatial comparison would be valuable. Discussion of seasonal variations in this gradient would further enrich the analysis.
Ā
Line 337: āwith coastal > urban > suburban impactsā is not clear, please rephrase.
Ā
Line 425-429. This sentence is a little bit long and difficult to understand. Please rephrase and clarify.
Ā
Section 3.5: The reliability of the source analysis needs further strengthening. The current results are highly dependent on the accuracy of the prior emission inventory. However, significant discrepancies exist in fossil fuel emission estimates across different inventories, and emissions vary considerably between years. Furthermore, the biogenic contribution is derived by subtracting the fossil fuel estimates. Therefore, it is strongly recommended that the authors assess the impact of using different inventories on the conclusions.
Citation: https://doi.org/10.5194/egusphere-2025-3215-RC2 - AC2: 'Reply on RC2', jinwen zhang, 19 Jan 2026
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-3215', Anonymous Referee #2, 21 Oct 2025
This paper focuses on the atmospheric COā dynamics in Guangzhou, a coastal megacity. It does this by developing an observation-based framework that integrates ground-based CO and COā observations, as well as ĪCO/ĪCOā ratios. The paper analyses the spatiotemporal patterns of COā, assesses the influence of seaāland breezes and partitions anthropogenic and biogenic emissions. While the study presents interesting findings regarding Guangzhouās carbon emissions, some clarification of the results is necessary. The manuscript could be improved by providing more comprehensive interpretations of the methods and results, particularly with regard to the uncertainty of the results presented. Please see my comments below.
General comments:
(1) CO was measured simultaneously, but is only analysed in Section 3.4. Since CO is a good proxy for COā over urban regions, it would be interesting to check the CO results, for example in Figures 4 and 6. Please show and interpret the diurnal cycle of CO in contrast to COā.
(2) Please explain how the error bars in Figure 9 are calculated. It seems to me that the uncertainty is so large that the difference between summer and winter is not significant. Also, what is the uncertainty of the estimation in line 434 (60.17%)?
(3) The manuscript identifies seaāland breeze (SLB) days based on 24-hour wind direction transitions and a wind speed threshold of <10 m sā»Ā¹. While this threshold excludes most strong winds, the authors should clarify whether the potential influence of tropical cyclones or their peripheral circulations was considered. Even when wind speeds remain below 10 m sā»Ā¹, such events can disrupt local wind directions, potentially disturbing the regular daytimeānight-time SLB pattern. Without addressing these effects, SLB identification and subsequent COā dynamics interpretation may be biased.
Other comments:
Although the introduction highlights three key knowledge gaps in the existing research, it does not emphasise their relevance to coastal cities enough, nor does it distinguish these gaps from those in studies of inland cities. Further descriptions are required. While the introduction mentions Guangzhouās GDP, population, green coverage and seaāland breeze frequency, it does not link these to the study objectives. Green coverage, which is important for biogenic fluxes, is neither compared with that in other coastal cities nor discussed in terms of its impact on flux magnitude. The frequency of the seaāland breeze is cited without detailing its seasonal patterns or how it differs from that in other cities. Furthermore, Guangzhou's carbon mitigation policies, which could influence anthropogenic and biogenic COā emissions, are not mentioned.
Line 25: clean air policies? Please provide a little bit more details of these policies in the abstract.
Line 114: EDGAR full name.
Figure 8: Please show the density plot; the points may overlap strongly with each other.
Line 448: is the EDGAR mean yearly or monthly? How is the temporal resolution of EDGAR done? How are daytime and night-time emissions differentiated?
The STILT model releases 500 particles with a 72-hour backward trajectory and a spatial resolution of 0.08Ā° Ć 0.08Ā°, but no sensitivity tests are reported. It is unclear whether increasing the number of particles or improving the spatial resolution would significantly affect the footprint simulations. These model parameters directly impact the accuracy of COā emission estimates, so relevant validation analyses are essential.
Figure 2: The distinction between the different wind directions is unclear. The authors should consider optimising the figure, for example by using more distinct colours, line styles or annotations, to improve clarity and readability.
Figure 7: It is recommended that the legend be placed outside the figure.
Line 195: ĪCOā appears improperly formatted.
On line 288, 'Despite CH's stronger biogenic coupling (NDVI correlation: ā0.72; Fig. 3f), NS's COā levels remained 9.80 ppm lower than CH in summer and 5.80 ppm higher in winter, underscoring transport-dominated over biogenic controls at the coastal site.' This whole sentence should be rephrased to enhance logical rigour.
On line 324, it states that, at CH, 'smaller daytime weekdayāweekend differences suggest that biogenic fluxes outweigh anthropogenic variations'. This statement is somewhat too absolute. It would be better to acknowledge the uncertainties more appropriately and consider the potential influence of atmospheric transport and boundary layer dynamics.
Citation: https://doi.org/10.5194/egusphere-2025-3215-RC1 - AC1: 'Reply on RC1', jinwen zhang, 19 Jan 2026
-
RC2: 'Comment on egusphere-2025-3215', Anonymous Referee #3, 20 Dec 2025
This study analyzes COā and CO concentration variations over 1 year and 9 months at three sites in Guangzhouāa coastal megacityāexamining their relationship with land-sea breezes. Using backward trajectory footprint modeling, it quantifies fossil fuel and biogenic contributions. Given the scarcity of direct COā observations in this region, these findings offer valuable insights into carbon sources/sinks in coastal southern Chinese megacities. The methodology is overall sound, and the work merits publication inĀ Atmospheric Chemistry and Physics. However, the introduction of research background, uncertainty analysis in source partition, the robustness of the results, and the depth of discussion could be further improved:
Ā
Specific comments:
Introduction Section: The introduction could benefit from restructuring to enhance its logical flow. The rationale for reporting urban-scale COā concentrationsāparticularly the need to clarify carbon sources and sinksāshould be more explicitly emphasized. While the study relies heavily on the COā/CO ratio and footprint modeling, there is limited discussion on the inherent uncertainties of these methods or how they compare with alternative approaches, such as 3-D atmospheric inversions. Introducing these methodological considerations would strengthen the foundation for the work.
Ā
Section 2.5: TheĀ a prioriĀ emission inventory utilized in the study should be better specified, which inventory? What spatial resolution for what year?
Ā
Line 219ļ¼for footprint simulation, using 500 particles over a 72-hour simulation period appears rather small. In this setting, the particle count per time step and per grid point is extremely low, potentially introducing substantial uncertainty in footprint estimation. It would be valuable to test and report the sensitivity of results to a larger number of particles (1000-10000) to ensure robustness.
Ā
Line 225: The statement that "footprint uncertainties are neglected under the assumption of unbiased atmospheric transport" seems to lack justification. Could the authors provide references supporting this choice or acknowledge its limitations?
Ā
Fig 3A:Ā Consider revising the x-axis label to "JanuaryāDecember" for clarity.
Ā
Lines 255ā265:Ā The urbanāsuburban differences highlighted here are insightful. Adding a figure to illustrate this spatial comparison would be valuable. Discussion of seasonal variations in this gradient would further enrich the analysis.
Ā
Line 337: āwith coastal > urban > suburban impactsā is not clear, please rephrase.
Ā
Line 425-429. This sentence is a little bit long and difficult to understand. Please rephrase and clarify.
Ā
Section 3.5: The reliability of the source analysis needs further strengthening. The current results are highly dependent on the accuracy of the prior emission inventory. However, significant discrepancies exist in fossil fuel emission estimates across different inventories, and emissions vary considerably between years. Furthermore, the biogenic contribution is derived by subtracting the fossil fuel estimates. Therefore, it is strongly recommended that the authors assess the impact of using different inventories on the conclusions.
Citation: https://doi.org/10.5194/egusphere-2025-3215-RC2 - AC2: 'Reply on RC2', jinwen zhang, 19 Jan 2026
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,957 | 247 | 33 | 2,237 | 103 | 36 | 37 |
- HTML: 1,957
- PDF: 247
- XML: 33
- Total: 2,237
- Supplement: 103
- BibTeX: 36
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Jinwen Zhang
Yongjian Liang
Chenglei Pei
Bo Huang
Yingyan Huang
Xiufeng Lian
Shaojie Song
Chunlei Cheng
Zhen Zhou
Junjie Li
Mei Li
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2294 KB) - Metadata XML
-
Supplement
(2112 KB) - BibTeX
- EndNote
- Final revised paper
This paper focuses on the atmospheric COā dynamics in Guangzhou, a coastal megacity. It does this by developing an observation-based framework that integrates ground-based CO and COā observations, as well as ĪCO/ĪCOā ratios. The paper analyses the spatiotemporal patterns of COā, assesses the influence of seaāland breezes and partitions anthropogenic and biogenic emissions. While the study presents interesting findings regarding Guangzhouās carbon emissions, some clarification of the results is necessary. The manuscript could be improved by providing more comprehensive interpretations of the methods and results, particularly with regard to the uncertainty of the results presented. Please see my comments below.
General comments:
(1) CO was measured simultaneously, but is only analysed in Section 3.4. Since CO is a good proxy for COā over urban regions, it would be interesting to check the CO results, for example in Figures 4 and 6. Please show and interpret the diurnal cycle of CO in contrast to COā.
(2) Please explain how the error bars in Figure 9 are calculated. It seems to me that the uncertainty is so large that the difference between summer and winter is not significant. Also, what is the uncertainty of the estimation in line 434 (60.17%)?
(3) The manuscript identifies seaāland breeze (SLB) days based on 24-hour wind direction transitions and a wind speed threshold of <10 m sā»Ā¹. While this threshold excludes most strong winds, the authors should clarify whether the potential influence of tropical cyclones or their peripheral circulations was considered. Even when wind speeds remain below 10 m sā»Ā¹, such events can disrupt local wind directions, potentially disturbing the regular daytimeānight-time SLB pattern. Without addressing these effects, SLB identification and subsequent COā dynamics interpretation may be biased.
Other comments:
Although the introduction highlights three key knowledge gaps in the existing research, it does not emphasise their relevance to coastal cities enough, nor does it distinguish these gaps from those in studies of inland cities. Further descriptions are required. While the introduction mentions Guangzhouās GDP, population, green coverage and seaāland breeze frequency, it does not link these to the study objectives. Green coverage, which is important for biogenic fluxes, is neither compared with that in other coastal cities nor discussed in terms of its impact on flux magnitude. The frequency of the seaāland breeze is cited without detailing its seasonal patterns or how it differs from that in other cities. Furthermore, Guangzhou's carbon mitigation policies, which could influence anthropogenic and biogenic COā emissions, are not mentioned.
Line 25: clean air policies? Please provide a little bit more details of these policies in the abstract.
Line 114: EDGAR full name.
Figure 8: Please show the density plot; the points may overlap strongly with each other.
Line 448: is the EDGAR mean yearly or monthly? How is the temporal resolution of EDGAR done? How are daytime and night-time emissions differentiated?
The STILT model releases 500 particles with a 72-hour backward trajectory and a spatial resolution of 0.08Ā° Ć 0.08Ā°, but no sensitivity tests are reported. It is unclear whether increasing the number of particles or improving the spatial resolution would significantly affect the footprint simulations. These model parameters directly impact the accuracy of COā emission estimates, so relevant validation analyses are essential.
Figure 2: The distinction between the different wind directions is unclear. The authors should consider optimising the figure, for example by using more distinct colours, line styles or annotations, to improve clarity and readability.
Figure 7: It is recommended that the legend be placed outside the figure.
Line 195: ĪCOā appears improperly formatted.
On line 288, 'Despite CH's stronger biogenic coupling (NDVI correlation: ā0.72; Fig. 3f), NS's COā levels remained 9.80 ppm lower than CH in summer and 5.80 ppm higher in winter, underscoring transport-dominated over biogenic controls at the coastal site.' This whole sentence should be rephrased to enhance logical rigour.
On line 324, it states that, at CH, 'smaller daytime weekdayāweekend differences suggest that biogenic fluxes outweigh anthropogenic variations'. This statement is somewhat too absolute. It would be better to acknowledge the uncertainties more appropriately and consider the potential influence of atmospheric transport and boundary layer dynamics.