the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Sep 2025
Constraining urban biogenic CO2 fluxes: Composition, seasonality and drivers from radiocarbon and inventory analysis
Abstract. Urban areas play a pivotal role in achieving net-zero emissions to limit global warming to 1.5 °C, given their high carbon footprint and mitigation potential. Accurate quantification of urban CO2 sources is essential for effective carbon budgeting and targeted climate action. While fossil fuel CO2 (CO2ff) emissions are extensively studied, biogenic CO2 (CO2bio) dynamics remain poorly constrained. Here, we separate fossil and biogenic contributions to CO2 enhancements above background using Δ14C and CO2 measurements in Shenzhen, a humid subtropical Chinese megacity potentially subject to substantial biomass burning influence. We calculate human/livestock metabolic emissions (CO2HLM) at 9.32 Mt/6.22 kt per year from population/livestock data and respiratory/excretory rates, and estimate biomass burning emissions (CO2BB) at 5.05 Mt/yr using an inventory encompassing both open and domestic combustion. The residual CO2bio component is attributed to the terrestrial biosphere (CO2bio'). Integrating Δ14C with multi-source data reveals annual CO2bio contributions relative to fossil fluxes: CO2HLM (17.8 ± 3.1 %), CO2BB (9.2 ± 1.5 %), and CO2bio' (73.0 ± 3.5 %). Key findings demonstrate the terrestrial biosphere component acts as a year-round net carbon sink with significant seasonality (11.5 ppm amplitude), driven primarily by atmospheric temperature (1–2 months lag; r = –0.80, p = 0.01) rather than precipitation. This study establishes human metabolic emissions as the dominant biogenic CO2 source (17.8 % vs. 9.2 % from BB) in megacities, yet shows that concurrent biospheric sequestration can offset 63 % of fossil emissions during growing seasons, advancing understanding of urban carbon budgets.
- Preprint
(1641 KB) - Metadata XML
-
Supplement
(1133 KB) - BibTeX
- EndNote
Status: open (until 11 Dec 2025)
- RC1: 'Comment on egusphere-2025-3882', Anonymous Referee #1, 29 Oct 2025 reply
-
RC2: 'Comment on egusphere-2025-3882', Anonymous Referee #2, 13 Nov 2025
reply
This manuscript describes a set of radiocarbon (14C) in CO2 and CO2 measurements made from five sites in Shenzhen, China, during afternoons every few days over a period of one year. They use these measurements to determine the contributions of fossil fuels (CO2ff) and biogenic sources (CO2bio) to the local CO2 enhancement (CO2xs). The results are compared with modelled estimates.
The data presented in this paper has real potential improve our understanding of urban CO2 fluxes in subtropical cities. Unfortunately, the current analysis has several key erroneous assumptions that make the interpretation and conclusions invalid. From my reading of the paper, it appears that the authors have rushed to produce a manuscript without having a full understanding of the existing state of scientific knowledge of urban biogenic CO2 fluxes. My recommendation is to reject this paper, and advise the authors to spend some time and effort expanding their understanding of the existing literature to understand the well-understood challenges of interpreting urban atmospheric greenhouse gas observations. The authors could then consider how the data they have collected can and cannot be used, to write a new paper that utilises their dataset but that draws on the strengths of that data rather than attempting to draw conclusions that cannot be supported by the data. I recommend the recent IG3IS Urban Guidelines (https://doi.org/10.59327/WMO/GAW/314, currently being translated into Chinese) as a starting point to understand how urban atmospheric observations can be applied, the things that don’t work, and an extensive reference list that describes many of the papers in this field.
Key general points:
- The main conclusions of this paper are fundamentally flawed. This dataset of afternoon-only measurements cannot be used to draw conclusions about net biospheric CO2 Photosynthetic CO2 drawdown can only occur during daylight hours (since it requires sunlight), whereas respiration happens during all hours of the day. Thus measurements made only during the mid-afternoon will be biased towards increased CO2 drawdown, and the net biospheric CO2 flux across the full diurnal cycle must be smaller than the mid-afternoon values, and could even be of opposite sign. Air mass transit times across cities are typically a few hours, so that the influence function for an afternoon sample typically only captures the afternoon and possibly part of the morning. This is in contrast to regional and continental-scale studies for which the influence function may mean that the sample captures a much larger part of the diurnal cycle. Thus, it is not valid to draw conclusions about the net biogenic CO2 flux for the city from the afternoon dataset presented in this paper.
- Choice of background for the observations, the model domain and it’s spatial and temporal resolution, and metoeorology must be carfeully considered and align with one another so that valid comparisons between observations and model results can be made. The current paper does not describe these in enough detail for the reader to determine whether this has been acheived. From what is written in the paper, I suspect that there is a substantial mismatch that invalidates the conclusions.
- The FLEXPART model is described as running at 0.05° resolution (about 5 km), but the underlying CFSv2 Reanalysis Model resolution is not provided – does it align with the FLEXPART resolution?
- Do the FLEXPART footprints show that air masses come from the NW to Shenzhen – ie is Nanling an appropriate background site for air masses entering the city? If not, or if it is only appropriate in some seasons or conditions, then the choice of background needs to be reconsidered.
- The FLEXPART backtrajectories were run for 30 days, whereas the air mass transit time from the Nanling background site is likely to be only a few hours up to a day. Thus observational CO2xs and CO2bio only include any fluxes that occur as the air mass transits from the Nanling background site to the observational sites in Shenzhen (all fluxes from upwind of the background site are removed when the background is subtracted). In contrast, the FLEXPART model with 30 day backtrajectories will include fluxes from much further afield. e. the comparison of model and observation does not seem to be valid.
- Even if the model and observational background are aligned, the Nanling site is 200-300 km from Shenzhen, which means that the observations implicitly include emissions/sinks from areas between Nanling and Shenzhen – ie the observations are not just for Shenzhen city, but for a substantially larger region.
- The determination of CO2bio’ from the 14C observations is strongly dependent on the choice of background. It is clear from Figure 4d that the choice of background changes the seasonal cycle timing, it’s magnitude, and the sign of CO2bio’. Much of the discussion and conclusions in the paper relies on the CO2bio’ values determined relative to the Nanling background. Yet those CO2bio’ values appear to be inconsistent with multiple other datasets that are described in the paper, including the CASA model, eddy covariance measurements in Shenzhen, and other indices such as leaf area index. Thus all of the conclusions and inferences drawn from the CO2bio’ values are not justified.
Specific comments
35-38. Are you referring specificly to urban biogenic CO2 fluxes? There have been a few publications on this topic in the last few years, and the authors should refer to those studies.
Hardiman, B. S., Wang, J. A., Hutyra, L. R., Gately, C. K., Getson, J. M., and Friedl, M. A.: Accounting for urban biogenic fluxes in regional carbon budgets, Science of the Total Environment, 592, 366-372, 10.1016/j.scitotenv.2017.03.028, 2017.).
Sargent, M., Barrera, Y., Nehrkorn, T., Hutyra, L. R., Gately, C. K., Jones, T., McKain, K., Sweeney, C., Hegarty, J., Hardiman, B., and Wofsy, S. C.: Anthropogenic and biogenic CO2 fluxes in the Boston urban region, Proceedings of the National Academy of Sciences of the United States of America, 115, 7491-7496, 10.1073/pnas.1803715115, 2018.
Chapter 5 of the IG3IS Urban guidelines has a wealth of information on this topic.
Turnbull, J. C., Curras, T., Gurney, K. R., Hilton, T. W., Mueller, K. L., Vogel, F., Yao, B., Albarus, I., Ars, S., Baidar, S., Chatterjee, A., Chen, H., Chen, J., Christen, A., Davis, K. J., Hajny, K., Han, P., Karion, A., Kim, J., Lopez Coto, I., Papale, D., Ramonet, M., Sperlich, P., Vardag, S. N., Vermeulen, A., Vimont, I. J., Wu, D., Zhang, W., Augusti-Panareda, A., Ahlgren, K., Ahn, D., Boyle, T., Brewer, A., Brunner, D., Cai, Q., Chambers, S., Chen, Z., Dadheech, N., D’Onofrio, C., Dunse, B. L., Engelen, R., Fathi, S., Gioli, B., Hammer, S., Hase, F., Hong, J., Hutyra, L. R., Järvi, L., Jeong, S., Karstens, U., Kenion, H. C., Kljun, N., Laurent, O., Lauvaux, T., Lin, J. C., Liu, Z., Loh, Z., Maier, F., Matthews, B., Mauder, M., Miles, N., Mitchell, L., Monteiro, V. C., Mostafavi Pak, N., Röckmann, T., Roiger, A., Roten, D., Scheutz, C., Shahrokhi, N., Shepson, P. B., Stagakis, S., Tong, X., Trudinger, C. M., Velasco, E., Whetstone, J. R., Winbourne, J. B., Wu, J., Xueref-Remy, I., Yadav, V., Yu, L., Zazzeri, G., Zeng, N., and Zhou, M.: IG3IS Urban Greenhouse Gas Emission Observation and Monitoring Good Research Practice Guidelines WMO GAW Report 314, World Meteorological Organisation, Geneva Switzerland, 2025.
40-45. There exist some studies of urban CO2 fluxes in tropical and sub-tropical cities.
Velasco, E., Segovia, E., and Roth, M.: High-resolution maps of carbon dioxide and moisture fluxes over an urban neighborhood, Environmental Science: Atmospheres, 10.1039/d2ea00108j, 2023.
Chapter 9 of Turnbull, J. C., Curras, T., Gurney, K. R., Hilton, T. W., Mueller, K. L., Vogel, F., Yao, B., Albarus, I., Ars, S., Baidar, S., Chatterjee, A., Chen, H., Chen, J., Christen, A., Davis, K. J., Hajny, K., Han, P., Karion, A., Kim, J., Lopez Coto, I., Papale, D., Ramonet, M., Sperlich, P., Vardag, S. N., Vermeulen, A., Vimont, I. J., Wu, D., Zhang, W., Augusti-Panareda, A., Ahlgren, K., Ahn, D., Boyle, T., Brewer, A., Brunner, D., Cai, Q., Chambers, S., Chen, Z., Dadheech, N., D’Onofrio, C., Dunse, B. L., Engelen, R., Fathi, S., Gioli, B., Hammer, S., Hase, F., Hong, J., Hutyra, L. R., Järvi, L., Jeong, S., Karstens, U., Kenion, H. C., Kljun, N., Laurent, O., Lauvaux, T., Lin, J. C., Liu, Z., Loh, Z., Maier, F., Matthews, B., Mauder, M., Miles, N., Mitchell, L., Monteiro, V. C., Mostafavi Pak, N., Röckmann, T., Roiger, A., Roten, D., Scheutz, C., Shahrokhi, N., Shepson, P. B., Stagakis, S., Tong, X., Trudinger, C. M., Velasco, E., Whetstone, J. R., Winbourne, J. B., Wu, J., Xueref-Remy, I., Yadav, V., Yu, L., Zazzeri, G., Zeng, N., and Zhou, M.: IG3IS Urban Greenhouse Gas Emission Observation and Monitoring Good Research Practice Guidelines WMO GAW Report 314, World Meteorological Organisation, Geneva Switzerland, 2025.
53-55. 14C is not the only way to tackle this problem. See the references above for eddy covariance and modelling techniques that can and have been used.
75-78. Shenzhen is one city in a densely populated region. Are the reported emissions information for Shenzhen alone, for the region, some part of the region? In the analysis, how are the Shenzhen emissions separated from those from other cities and non-urban areas?
75-78. The two inventories given here are about 20% different. Why are they so different from each other, and how might those differences influence the analysis in this paper?
- Why are air standards that are prepared, extracted and measured in the same way as the samples not used?
- It’s a bit confusing to give uncertainty on an uncertainty.
136-137. How valid is it to use monthly mean values for the background values? Individual air masses might have quite different histories that require matching the background to the sampling dates and times. For 14C this might be ok, but for CO2 this is really tricky because of the strong diurnal cycle in CO2.
Figure 2. I comment on this later as well. Nanling does not look like a bckground site in the winter months – the dip in ∆14C looks like the influence of regional emissions. This needs to be considered and accounted for, and is likely the explanation for some of the strange seasonality calculated later.
160-165. Please reference this – Turnbull et al 2006.
169-171. I’m curious – are you basing this on reporting from power plants themselves? Has anyone measured 14CO2 emissions from Chinese nuclear power plants? Graven and Gruber (2011) and Zazzeri et al (2018) both suggest that emissions are likely to be episodic with maintenance cycles. In Figure 3, there are ∆14C values that are higher than the Nanling background – could this be due to nuclear power plant emissions?
172-182. Do you use a single correction value for all samples, or do you model it explicitly for each sample (as done in Miller et al., 2012)?
189-191. Please reference this method.
- Is 0.05x0.05° sufficient resolution for an urban study? What is the resolution of the underlying met data? What is the vertical resolution of the model? What height is used as the cutoff for “surface” influence in this calculation?
195-196. What evidence is there that CASA-GFED4s is of sufficient quality for this study? Has it been compared with observations in China? How does it handle urban areas with heterogenous terrain and impervious surface? It is well-known that regional scale biogenic flux models may not represent urban areas well (e.g. Hardiman et al., 2017).
197-199. How realistic is it to impose a single diurnal cycle on monthly data? Day-to-day variability in CO2 fluxes are large, especially for photosynthesis.
200-206. Please clarify how the ∆14C signature of heterotrophic respiration was calculated. Why is the value so much lowr than that used for biomass burning? What is the value of the correction term?
207-211. How confident are you in the quality of these biomass burning flux estimates? What is the relative contribution of open burning and anthropogenic domestic emissions (at least in the inventories you have available)? What is the value of the correction term you derive?
- “primarily from fossil fuel combustion”. Is there evidence to support this, or is it supposition?
224-225. How was this two end-member mixing analysis done? It is not described in the methods.
224-226. As shown in Figure 4, the choice of background has a massive impact on the calculated CO2bio. It is problematic to make interpretations of CO2bio based on a single background without acknowledging the uncertainties in CO2bio.
- CO2ff values of -5 ppm? The reported ∆14C uncertainties of 2.4‰ imply uncertainty in CO2ff of 1-2 ppm, so these values are too low to be explained by the uncertainties. From the figures, they are caused by the choice of background and this must be addressed in the paper.
228-231. Please reference the previous works that have discussed the impact of increased emissions and BL trapping.
231-233. Lower CO2ff concentrations do not necessarily imply lower CO2ff emissions – the concentrations observe depend on the sampling locations relative to emission sources within the city, and the atmospheric transport.
- Please show CO2xs in Figure 3 as well.
237-238. This statement is problematic because it does not recognise that by sampling only in the afternoon hours, the dataset MUST be biased to show higher photosynthetic drawdown than the full diurnal cycle would show.
239-240. When quoting data from other research, it is essential to reference those papers.
240-242. Indeed, the background selection is critical, and from Figure 4 it is clear that the interpretation of CO2bio will be very different if a different background is used. This challenge of background choice must be resolved before this paper can come to any meaningful conclusions.
Figure 3c and d. The apparent seasonal cycles in CO2ff and CO2bio are not intuitive. Are there possible physical explanations for these seasonal cycles, or are they another clue that the choice of background is problematic?
- “Excretory emission rate”? Are you referring to human respiration? Or something else?
267-272. How was the amount of livestock determined? What region was considered for the livestock, and how does that align with the region used in the observations and modelling?
- If biomass burning in Southeast Asia is important in Shenzhen, then this implies air masses coming from Southeast Asia arrive in Shenzhen. But the Nanling background is NW of Shenzhen. This seems to negate the background choice made. Surely this also implies that other source sectors, such as biogenic CO2 and fossil fuel CO2 from Southeast Asia would influence this site?
298-300. References please.
300-302. Biomass burning in Shenzhen and biofuel use in Los Angeles seem to be due to very different sources. Please clarify how this comparison is useful.
302-304. “This dominance of natural processes...” Is this supposition or can this statement be backed up by data or references?
311-312. Earlier in the paper, you say that CO2ff has seasonality of 25%, but here you claim (without references) it does not have significant seasonality.
Figure 4a,b. The seasonality in CO2bio and CO2bio’ does not look like a typical biogenic seasonal cycle, exhibiting net positive values in April, May and June, but net negative values in January, February and March. What supporting evidence is there for such an unusual seasonal cycle? Or is this an artifact of the choice of background, seasonally varying winds, or some other factor?
In Figure 4a, how was the inset pie chart calculated? Since CO2bio’ is a net negative value, it is not obvious how the pie chart is determined.
Figure 4c,d. The modeled CO2bio’ values exhibit a very different seasonal cycle than the 14C-based observations. And in Figure 4d, we see that the choice of background makes a dramatic difference to CO2bio’, changing both the seasonal cycle and often the sign (from net uptake to net release of CO2). The 3 other background choices all suggest a more typical seasonal cycle that is closer to that from the model. This suggests very strongly that indeed the choice of background is critical, and that it is likely that many of the features of CO2bio’ are artifacts of the choice of background.
333-334. Why is Shenzhen CO2bio’ being compared with a single dataset from Los Angeles, when the authors immediately acknowledge that Shenzhen and Los Angeles have very different environments? There are many other studies that have attempted to look at urban biospheric exchange that should be referred to.
Line 338. How is the CASA model sampled in the comparison with the observations shown in Figure 4c? Is it sampled for the same times/days as the observations? Does it account for the diurnal cycle in CO2bio? What region is used to calculated CO2bio from the model? Is it consistent with the region that the observations “see”?
340-341. The statement that “our results indicate that CASA-GFED4s may underestimate he seasonal amplitude of NEE fluxes in Shenzhen” is not convincing, since the observed CO2bio’ values seem to be so strongly dependent on the choice of background and it is not clear what region these CO2bio’ values represent.
342-349. The authors argue that the 4 different backgrounds they examine give similar results for CO2bio’, but this is patently incorrect, as shown in Figure 4d. The different background choices change both the amplitude of the seasonal cycle and the sign of CO2bio. This conclusion is therefore not justified by the data.
355-356. Earlier, CO2bio’ was described as having removed human respiration impacts, so one would expect a lack of correlation with population.
370-372. Do these indices support the unusual seasonal cycle shown in Figure 4a and 4b?
- This statement is fundamentally flawed, because the observational dataset is all from afternoon measurements, and does not account for other times of the day – CO2bio’ must be positive at night when photosynthesis does not occur. This observational dataset does not provide the appropriate data to determine a net annual sum of CO2bio’.
382-384. The eddy covariance data appears to be consistent with the CASA model seasonal cycle. This is yet another piece of evidence suggesting that the observed CO2bio’ presented in this study is unusual and the seasonal cycle is quite possibly an artifact of the methodology.
394-402. As noted several times before, it is problematic to draw inferences about relationships between CO2bio’ and other factors when the CO2bio’ values are not supported by other datasets, the region that they represent is unknown, and the choice of background appears to give conflicting results.
403-404. In this section of the paper, the authors refer almost exclusively to a single study from Los Angeles (Miller et al., 2020) for comparison. There are several papers that have examined CO2bio in urban areas, using a variety of methodologies, and the authors should examine that literature for comparisons with cities that are more similar to Shenzhen. See my earlier suggestions for some references to start with.
427-438. This conclusion is fundamentally flawed, as noted earlier. It is not valid to use afternoon measurements to infer net annual biogenic fluxes, since afternoon measurements are always going to be biased to higher drawdown than the daily average.
446-447. Please check your references.
Citation: https://doi.org/10.5194/egusphere-2025-3882-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 483 | 61 | 18 | 562 | 41 | 18 | 14 |
- HTML: 483
- PDF: 61
- XML: 18
- Total: 562
- Supplement: 41
- BibTeX: 18
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear editor, dear authors,
please find my review comments in the attached pdf file.