the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Jul 2026
Constraining anthropogenic CO2 fluxes in mixed urban source areas using eddy covariance and footprint-informed ecological parameter migration
Abstract. Urban net CO2 fluxes measured by eddy covariance (EC) cannot be treated as anthropogenic emissions without biogenic constraints, because they integrate fossil-fuel sources, biogenic processes, human respiration, and changing footprints. Existing partitioning approaches often require isotopes, tracers, multi-species fluxes, or clean vegetation/emission sectors unavailable at many urban sites. We developed a footprint-informed ecological parameter-migration framework to constrain anthropogenic CO2 fluxes (Fff) in mixed urban source areas of Guangzhou, a humid subtropical megacity. A highly vegetated, weakly disturbed suburban donor site provided ecological parameters, which were transferred to a mixed urban target site using dynamic footprints and footprint-weighted enhanced vegetation index (EVIfp). The migrated parameters produced plausible diurnal and seasonal gross primary productivity (GPP) and ecosystem respiration (Reco), supporting the donor–target constraint. At the target site, daytime biogenic uptake masked anthropogenic emissions and occasionally drove net CO2 flux toward neutral or weakly negative values. After subtracting Reco, GPP, and human respiration, Fff remained consistently positive throughout the day (1.80–5.14 μmol m−2 s−1). Fff showed morning and evening peaks consistent with NOx, CO, wind-sector source areas, and traffic indicators, while footprint-aligned inventories provided magnitude context and suggested spatial-proxy mismatch in a high-population urban functional zone with relatively low on-site combustion. Uncertainty was dominated by GPP light-response structure, with smaller effects from other perturbations. These results demonstrate the feasibility of extracting footprint-scale Fff from mixed urban source areas using conventional EC, remote-sensing, and footprint data where isotopic or multi-species flux observations are unavailable.
- Preprint
(4816 KB) - Metadata XML
-
Supplement
(3152 KB) - BibTeX
- EndNote
Status: open (until 30 Sep 2026)
- RC1: 'Comment on egusphere-2026-3381', Anonymous Referee #2, 24 Aug 2026 reply
-
RC2: 'Comment on egusphere-2026-3381', Anonymous Referee #3, 10 Sep 2026
reply
This study demonstrates how CO2 EC flux partitioning into anthropogenic and biogenic components can be done without depending on additional measurements by developing an ecological parameter-migration framework between two cities in Guangzhou, China. The two cities shared similar climate conditions but had different land cover distributions, with the donor city having mostly vegetative land cover and the target city being dominated by impervious (or built-up) land cover. The authors clearly explained most of their methodology and provided sensitivity and uncertainty calculations that supported their choices, and the results indicated that the framework they proposed worked relatively well. It would be interesting to have a bit more discussion in the paper on how feasible the authors think it would be to reproduce this at other sites or other areas of the world.
The paper is well-written and detailed in most aspects, but as highlighted below, there are a few areas that require additional information in order to improve comprehension and reproducibility, specifically regarding footprint development and emission inventory/diurnal profile implementation. Once this and a few minor comments (listed below) are addressed, I recommend this paper for publication.
Abstract
Line 34: I would replace ‘footprint-aligned’ with ‘footprint-weighted’.
Introduction
Lines 71-94: Somewhere in this paragraph, it would be good to cite a study that came out a couple of months ago (Molinier et al 2026, https://doi.org/10.1021/acsestair.6c00148) which shows another way to try to partition urban EC flux measurements into anthropogenic and biogenic components by combining footprints, emission inventories, and VPRM outputs rather than adding more measurements.
Data and methods
Lines 193 and 196: It would help to already know the measurement heights at this stage, rather than waiting until the next section.
Lines 197-198: What is the study area that is considered by the footprint model (in m^2 or km^2)?
Figure 1: Is it possible to make the cumulative footprint contours (30%-70%) more distinctive? They’re difficult to see even after zooming in.
Line 256: The authors allude to ‘footprint calculations’ here and have mentioned using ‘footprint-weighted’ parameters throughout the main text and SI without actually explaining how they implemented the model, what inputs they used, quality flagging of the footprints themselves, etc. A short paragraph or section on this topic would strengthen the paper as it would help with reproducibility.
Line 347: For clarification, does ‘11 d subsets’ refer to subsets of 11 days?
Results and discussion
Line 463: So the 2025 observations were not considered? If that is the case, the authors should specify that earlier when they mention the duration of observations.
Lines 833-834: I might have missed this, but can the authors clarify where the diurnal patterns of these species are coming from? Measurements of NOx, CO, and PM2.5 (or any other tracer, for that matter) are not mentioned in the methods section, so are these well-established profiles or are they observation-based? While there is a brief reference to observations of these species in the SI, more detail is needed on how the diurnal profiles were developed for use in this study.
Lines 944-946: It would be helpful to include the spatial and temporal resolutions of the inventories here in addition to the SI.
Supplement Information
Section S3: Out of curiosity, do the human respiration emissions calculations account for whether humans are indoors or outdoors? Is this what is meant by ‘activity level’ or ‘activity intensity’? Also, just under equation S2, the term ‘Agrid’ is not formatted with subscripts as the other variables in the text.
Section S4: The air temperature variable ‘Ta’ also is not formatted with subscripts like the other variables in the text.
Section S5: How were emissions in the inventory reported and what sectors were included? How were they converted to flux units? If the inventory is, for example, annual, how did the authors convert it to a monthly inventory? Or is the inventory already reported on a monthly basis? Lastly, did the authors use the 2023 inventory for comparison with observations from both 2023 and 2024? More details are needed regarding this aspect of the analysis.
Section S8: Since the road congestion data are from May 2026 and the EC data are from 2023-2024, why do the authors state that they ‘do not fully overlap’ when actually they do not overlap at all?
Figure S9: There is no reference to this figure that I could find in the main text, so it would be great if the authors could provide some text to explain the significance of the traces in panels a and b overlapping so much. In panel b especially, it is almost impossible to distinguish between the three curves that are apparently shown.
Citation: https://doi.org/10.5194/egusphere-2026-3381-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 218 | 94 | 34 | 346 | 60 | 29 | 27 |
- HTML: 218
- PDF: 94
- XML: 34
- Total: 346
- Supplement: 60
- BibTeX: 29
- EndNote: 27
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Zhang et al. present a novel framework for partitioning urban eddy covariance CO₂ fluxes into anthropogenic and biogenic components in mixed-source environments where conventional isotopic or multi-species approaches are infeasible. The method is applied to Guangzhou and demonstrates good consistency with independent tracers (NOx, CO), wind-sector source area analyses, and traffic congestion patterns. Overall, the manuscript is well structured and clearly written, and it benefits from extensive sensitivity analyses. I only have a few minor comments that I hope will strengthen the paper.
The selection of the CH–PY site pair is reasonably justified on the grounds of shared regional climate and similar evergreen broadleaf forest composition. However, the manuscript would benefit from a more explicit discussion of the ecological comparability between the two sites. This would help readers assess whether the methodology is transferable to other cities. In particular, urban environments often host vegetation types that differ markedly from those in surrounding natural areas, and urban soils may exhibit distinct respiration characteristics due to irrigation, management practices, and other disturbances. Could these factors affect the general applicability of the approach to other urban settings?
More generally, I would encourage the authors to offer broader guidance on how eddy covariance systems, when equipped with this fossil-fuel separation technique, could be deployed to advance our understanding of urban carbon dynamics. At present, while all the results appear reasonable, the approach provides only limited insight into what is not known or what is uncertain about the city-wide fossil-fuel CO₂ emissions, except for the inventory-based proxies used here are not applicable at fine spatial scales. A more explicit discussion of the potential and limitations of this technique for informing urban emission monitoring and policy would greatly enhance the paper's impact.
In Section 3.4.3, the authors note that the PY footprint includes a substantial proportion of water bodies (~10%), which are commonly treated as zero-emission areas in emission inventories but may contribute to the total flux. How different assumptions regarding the treatment of water fractions might influence the inventory comparison?
In Section 3.2.1, the temperature-stratified GPP scheme is found to outperform both the pure saturating and VPD-inhibition structures, and is consequently adopted as the baseline. This is an interesting finding. However, the discussion could usefully acknowledge that temperature and VPD are often correlated, making it difficult to isolate their individual effects. It is plausible that the temperature-stratified scheme may indirectly capture VPD influences; this point merits brief consideration.
Figure 6a: it would be helpful to include monthly variations in Fcr as well, as what is done in Fig. 5.
Figure 7: given that both CO and CO₂ exhibit high background concentrations, the comparison based on percentage changes may be misleading.
I am unclear on the purpose of showing the illustrative 100% uncertainty range in Figure 10. Unless this serves a specific analytical or pedagogical role, I would suggest removing it or providing a clearer justification for its inclusion.