the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Building-resolving simulations of anthropogenic and biospheric CO2 in the city of Zurich with GRAMM/GRAL
Abstract. Urban areas are significant contributors to global CO2 emissions, requiring detailed monitoring to support climate neutrality goals. This study presents a high-resolution modeling framework using GRAMM/GRAL, adapted for simulating atmospheric CO2 concentrations from anthropogenic and biospheric sources and sinks in Zurich, Switzerland. The framework resolves atmospheric concentrations at the building scale, and it employs a detailed inventory of anthropogenic emissions as well as biospheric fluxes, which were calculated using the Vegetation Photosynthesis and Respiration Model (VPRM). Instead of simulating the full dynamics of meteorology and atmospheric transport, the dispersion of CO2 is precomputed for more than 1000 static weather situations, from which the best match is selected for any point in time based on the simulated and measured meteorology in and around the city. In this way, time series over multiple years can be produced with minimal computational cost. Measurements from a dense network of mid-cost CO2 sensors are used to validate the model, demonstrating its capability to capture spatial and temporal CO2 variability. Applications to other cities are discussed, emphasizing the need for high-quality input data and tailored solutions for diverse urban environments. The work contributes to advancing urban CO2 monitoring strategies and their integration with policy frameworks.
- Preprint
(13445 KB) - Metadata XML
-
Supplement
(32 KB) - BibTeX
- EndNote
Status: open (until 15 May 2025)
-
RC1: 'Comment on egusphere-2025-640', Anonymous Referee #1, 10 Apr 2025
reply
This is an interesting and impactful paper. I recommend publication after attention to the comments below.
Major Comments
- The rationale for referencing prior work seems somewhat arbitrary. There are notable recent efforts with similar observations and similar modeling that are omitted (although not combined). A more through discussion of the recent literature would help the reader put this work in appropriate context.
- It would be helpful to also include the spatial and temporal resolution of each input dataset in Table 1.
- Why are there 3 peaks in heating/hot water demand diurnal profile? I understand this is based on a simulation from CESAR-P, but what is the behavioral reason, especially for the 3-6 am peak and then the 9-12 peak? Homes then offices? Surprising to me that it’s not smoothed more in the morning
- Additional quantification of the uncertainties in each aspect of the model would be helpful.
- How should the reader thinking about the various sources of uncertainty in inversions. Presumably this approach largely aims to minimize the model representation error. Can we expect that other sources of error therefore are more important (measurement error, background error)?
- Additional discussion of whether a 10 m simulation is necessary for inverse modeling and offers significant benefit beyond lower resolution forward model would be helpful. The diurnal disagreement in Figure 15 is substantial. If this model is not replicating observed concentrations, would a lower resolution model have sufficed for inverse modeling? Some quantification of the improvement in modeled concentration of this model over others would be quite helpful or at least a discussion of how the reader might think about that question.
- For lines 427 - 438 (and Figure 16): How does the seasonal and diurnal variability in the probability density of stability classes prove that the distribution of stability classes are correct?
- I suggest adding an additional paragraph addressing the implications of this paper at the end.
Line-by-Line Edits
- Figure text is too small throughout the manuscript
- What is meant by the last line of Table 1? The line with “Example” and “how we should do.” Final line is an “example” and should be removed.
- Description Equation 1, for the unfamiliar reader, describing the units of each term would be helpful for understanding the equation
- Figure 11, is the white in the city center missing data due to the building height higher than 2 m?
- Line 36: It is unclear what “suitable measurement” means.
- Figure 5: Color key shows values of “-999” to “0.” The value -999 must be a filler for unknown values, so it should be removed or explained.
- Figure 7: Images b and c require x axis labels.
- Figure 8: How were the different vegetation patches selected as representative?
- Table 3: How was the threshold determined for match2obs selected sites.
Citation: https://doi.org/10.5194/egusphere-2025-640-RC1 -
RC2: 'Comment on egusphere-2025-640', Anonymous Referee #2, 17 Apr 2025
reply
This paper presents the GRAMM/GRAL model setup to simulate CO₂ concentrations in the city of Zurich. The manuscript is generally well-written and presents robust results, forming a solid basis for future studies. However, additional analysis and more in-depth discussion are necessary before the manuscript is ready for publication in Atmospheric Chemistry and Physics.
Major Comments
- Line 330: Selection of sites for matching
Your claim is that the model can resolve flow within street canyons. Therefore, including stations located within street canyons for model validation should be appropriate. If including these stations deteriorates the model performance, this may indicate that GRAMM/GRAL does not adequately capture the street-level flow, raising questions about the justification for using it in such environments. - Line 330: Omission of the mountain station
Excluding the mountain station is a limitation, especially since the manuscript claims GRAMM/GRAL performs particularly well at such sites. While it is true that models often struggle to represent flow at mountain tops, this issue warrants discussion. Additionally, excluding higher elevation sites may point to problems in representing the vertical concentration profile. Please address these points in the revised version. - Section 2.8: Background concentration
Please discuss how the choice of background concentration influences the results. Can you provide an uncertainty estimate, perhaps based on the spread among background stations? This is particularly important given your observation that a substantial portion of the concentration signal and its variability originates from the background. Moreover, this discussion is essential groundwork for any future inverse studies. - Line 384: Limited presentation of data
Since this paper aims to demonstrate the model’s capabilities (and is not an inverse study relying only on afternoon values), presenting only the mean diurnal cycle and two stations during the afternoon is insufficient. I recommend showing the full time series for the two selected stations and then arguing why a focus on afternoon values reduces mismatches. Otherwise, the comparison may be misleading and overly optimistic. - Figure 15: Large discrepancies in diurnal cycles
The differences between modeled and observed diurnal cycles are substantial. While you mention possible causes such as incorrect VPRM input, flawed scaling factors, or PBLH errors, the discussion remains superficial. Given Zurich’s extensive observational infrastructure, these discrepancies should be examined in more detail—for example, through comparison with other top-down estimates or vertical mixing data from tall towers. The discussion of catalog probabilities is weak and inconclusive; consider moving this part to the appendix and replacing it with a more thorough analysis of the discrepancies. - Line 475: Dynamic CO₂ simulation
You refer here to simulating CO₂ in a dynamic manner, but this concept is introduced rather abruptly in the discussion section without being presented in the results. Please clarify and provide context. This aspect deserves better integration into the manuscript. - Line 479: Potential for further investigation
As mentioned earlier, there are additional data and modeling resources that could help investigate the discrepancies observed in this study. It would strengthen the manuscript to make use of these tools or at least outline how they could be used in follow-up work.
Minor Comments
- Line 58: Please clarify what "As an alternative" refers to, or consider removing the phrase for clarity.
- Line 208ff: Remove the the brackets for Glauch et al. 2025.
- Table 1, last line: Please delete this line.
- Line 251: The sentence "Possible differences … cannot be accounted for in this way" is not entirely accurate. In principle, these differences could have been addressed by distinguishing between inner-city and outer emissions through separate source groups.
- Line 306: The statement "requires a uniform scaling of the fluxes from a given vegetation type" suggests a limitation, but this could be addressed with a more refined approach.
- Line 358: "Rather small gradients" is vague. Please provide a quantitative value or range.
- Line 461: It would be appropriate here to acknowledge that GRAMM/GRAL struggles to reproduce dynamics outside of afternoon hours.
- Literature: Please refer To Vardag and Maiwald (2024), who have already applied GRAMM/GRAL for inversion studies.
Citation: https://doi.org/10.5194/egusphere-2025-640-RC2 - Line 330: Selection of sites for matching
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
149 | 30 | 7 | 186 | 12 | 4 | 4 |
- HTML: 149
- PDF: 30
- XML: 7
- Total: 186
- Supplement: 12
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1