Review of Manuscript:

Title: Evaluating urban methane emissions and their attributes in a megacity, Osaka, Japan, via mobile and eddy covariance measurements

Author(s): Masahito Ueyama et al.

MS No.: egusphere-2024-3926

MS type: Research article

Does the paper address relevant scientific questions within the scope of ACP?             YES

Does the paper present novel concepts, ideas, tools, or data?                    YES

Are substantial conclusions reached?                                YES

Are the scientific methods and assumptions valid and clearly outlined?                YES

Are the results sufficient to support the interpretations and conclusions?            YES

Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?                    YES

Do the authors give proper credit to related work and clearly indicate their own new/original contribution?                                            YES

Does the title clearly reflect the contents of the paper?                        YES

Does the abstract provide a concise and complete summary?                    YES

Is the overall presentation well-structured and clear?                                   YES

Is the language fluent and precise?                                YES

Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?    YES

Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?                                            NO

Are the number and quality of references appropriate?                        YES

Is the amount and quality of supplementary material appropriate?                YES

Recommendation:

Accepted!

In my opinion, this is an excellent paper! 

Temporal and spatial variability of greenhouse gas exchange between the atmosphere and various ecosystems is one of the most critical problems of global climatology/ecology/environmental science. Despite the growing number of measurement sites worldwide, it should be noted that (for various reasons) their number in cities is insufficient. And yet, cities are such intensive sources of GHGs to the atmosphere! In addition, the existing urban sites mainly focus on measuring carbon dioxide fluxes, and long-term measurements of methane fluxes are still few (the results of only a few long-term measurement campaigns in the UK, Poland, Japan, or Italy have been published). Thus, it should be emphasized that the research results presented in the manuscript are valuable to knowledge of urban methane emissions.

The article presents the results of a detailed planned research experiment during which measurements were made using the EC method and during mobile measurements with methane concentration sensors installed on a car and a bicycle. The study is written clearly, and the Authors describe the results in detail. Noteworthy is the detailed description of the methodology and the extensive discussion of the results' quality (and their comparison with inventories). Some discrepancies in the results obtained from mobile measurements by car and bicycle are also discussed in detail by the Authors. The Authors also devoted much attention to estimating CH4 flux components, which I consider the most critical achievement in the presented study. 

I have only two comments:

P4, figure 4 should be more detailed (larger?). In the current figure, the differences in development, and especially the EC footprint, are poorly visible!

P6L168 – should be (Vickers and Mahrt, 1997)

Reviewer comments on “Evaluating urban methane emissions and their attributes in a megacity, Osaka, Japan, via mobile and eddy covariance measurements”, by Ueyama et al.

General comments

This paper reports on a study of CH₄ emissions in two Japanese cities using both eddy-covariance (EC) and mobile measurements of CH₄ and C₂H₆ at street level. The mobile measurement campaign covered a lot of ground using two modes of transport (car and bicycle). The aim of the study was to identify different sources of CH₄, quantify their emissions, upscale them to city level and compare them to local emissions inventories. This is an interesting subject because there is a mounting body of evidence demonstrating that urban emissions inventories tend to underestimate CH₄ emissions; identifying the potential missing urban sources of CH₄ and/or using in situ measurements to correct emission factors is crucial for improving current inventories and planning mitigation strategies. However, in my opinion, the experimental datasets generated in this study were not used to their full potential. Great effort was put into upscaling localised emissions derived from mobile measurements to the city scale; this required ‘calibrating’ the mobile fluxes using EC values obtained for very different spatial and temporal scales. This must introduce substantial uncertainties, which seem almost impossible to fully quantify.

More specifically, the design of the study could have been improved. A second mobile measurement campaign in a contrasting season would have been interesting since the EC time series demonstrates that there is a strong seasonality in emissions. The authors acknowledge that the chosen routes for the bicycle and car measurements might have biased the fluxes because of plausible differences in sources. This could have been tested by conducting a sampling experiment along the same routes with both modes of transport. Overall, there are many caveats, unknowns and sources of uncertainties associated with the mobile measurements, whilst the EC approach seem to have produced a robust time series exhibiting interesting temporal patterns. More could have been done with the EC data; e.g. the spatial distribution of EC fluxes was not even mentioned.

It is no surprise that EC and street level mobile fluxes cannot be reconciled since the spatial scales and micrometeorological conditions are vastly different for the 2 approaches. Fundamentally, trying for/forcing a reconciliation feels like shoehorning. Why not acknowledge that the 2 approaches yield very different estimates (for obvious reasons), shorten the discussion on this and emphasise instead the unique information that mobile measurements can provide and that EC cannot? For example, the findings about increased emissions near restaurants is interesting as is the hypothesis that these might be caused by the type of stoves used.

I would like to see a rewrite where the use of the mobile flux data is mostly limited to partitioning the street-level aggregated fluxes into natural gas, biogenic etc… whilst the EC measurements are used to calculate robust seasonal and annual budgets. Combining the two outputs (fraction of fossil fuel to biogenic emissions from mobile measurements) and seasonal and annual budgets can give a more robust estimate of the magnitude of emissions segregated by source, and improve the comparison with values from emissions inventories.

Specific comments

Equations 1-4: change the asterisk (*) to a multiplier sign (×) or omit altogether.

Lines 211-212: “The local background concentration was defined as the 5th percentile value during a 5-minute moving window”. How were these limits/criteria set?

Line 220: “spatial resolution of 12 s”.

• How was this aggregation threshold obtained?
• Consider changing the sentence to “a spatial resolution equivalent to 12 seconds of travel time (approximately 37 m).”
• I don’t understand the leak detection procedure and the data aggregation. I read the first sentence as meaning that the maximum enhancement within the 12 s block is used, but the next sentence states that multiple data points are available within each 12 s time chunk.

Lines 229 – 231: “We found that it underestimated the regional CH₄ fluxes in the empirical model, as noted below. Consequently, the use of units in the emission rate might be misleading; therefore, we used the unit of CH₄ enhancement (i.e., ppm) but a definition consistent with that used in previous studies.”

• What underestimates the regional CH₄? Weller’s empirical model? How was the regional flux calculated?
• Why is the use of units misleading?
• What definition of enhancement was used? What studies is it consistent with?

Additionally, these comments read like results and should therefore not be in the methods section.

Lines 233 – 235: so you forced the estimated CH₄ fluxes obtained from the mobile measurements to match the EC values by applying a correction factor. This seems problematic because of the difference in spatial scales, source distribution, intensity etc… Essentially, the assumption is that street-level spot measurements must be identical to EC values from a larger, spatially-integrated flux footprint. This is a dubious assumption and I doubt that it is defensible.

Line 244: why not use data from the LI-7810 measured at 1.85 above street level?

Equation 6:

• A: correction factor. Why not use the EC fluxes directly since you’re forcing the spot measurements to match the EC values? This correction factor bothers me because the assumption seems to be that most emissions of CH₄ are from roads (whether it be from leaks in the gas distribution network or else), but what about buildings etc…? If I’m barking up the wrong tree, please correct me and rewrite the relevant part of the methods section to make the approach more explicit.
• Em_mean: what area was Em averaged over?
• Multiplication by total road length within a ward: the underlying assumption is that the hotspots density and intensity is uniformly distributed along all roads. Is this a reasonable assumption?

Lines 282 - 283: “We estimated the upscaled CH₄ fluxes via Eqs. 5, 7, 8, and 9 and considered the range of upscaled fluxes as an uncertainty.” Does this mean that estimates using Eq.6 were not used after all? If that is the case, there is no point in discussing Eq.6 at all.

Line 295: another correction factor (0.64). I assume that this is adjust the mobile flux to the average diel cycle of the EC measurement, but it isn’t clear from the text how it was obtained. What time period was covered? Was the correction applied to the upscaled fluxes at ward or city level? I am concerned about the mounting number of simplifying assumptions, correction factors, etc… and the usefulness of the end estimates given that uncertainties are probably extremely large and difficult to quantify.

Lines 297 -299: “The annual emissions were only calculated for the vehicle measurements because the bicycle measurements were only conducted for the city center and residential areas and might underrepresent the rural areas.” This sentence seems to contradict the statement at line 262 “To determine the A factor, we used bicycle measurements […]”. The A correction factor forces fluxes from street-level spot  measurements to match EC values, which arise from all sources of CH4 within the EC flux footprint, road and non-road. In this context, it would seem more logical to derive the A correction factor from the car measurements since a more varied landscape was sampled.

Lines 322 -329: This section is out of place. It would make sense to move it to the end of the sentence at line 242 since the justification of why the 0.5-m height measurements was favoured.

Fig.3: The red dots are the data for which the correlation coefficient > 0.7, the blue dots <= 0.7; so what are the grey dots?

Fig.6-8: consider a different colour scheme because red, blue and green are difficult to distinguish from one another for some people.

Line 505: The agreement between EC and upscaled mobile measurements is perhaps a little misleading and circular because the upscaling was forced to match the EC data. I think that the authors are trying to highlight that mean annual EC values are close to the upscaled mobile data obtained for 2 months of the year only.

Line 585: In my opinion, this conclusion is somewhat overreaching because despite the agreement between car- and bicycle-borne measurements, there is a demonstrated discrepancy between the street-level fluxes and the EC fluxes. So, in terms of estimating urban emissions the mobile measurements are limited by the fact that they require calibration using EC measurements, and that they only capture temporal snapshots. In fact, calculating the emission budgets from the EC measurements comes with fewer uncertainties than using mobile measurements. However, the strength of the mobile measurements lies in the ability to identify hotspots and in source attribution. The authors should tone done the claims of usefulness in terms of estimating emissions at higher temporal and spatial scales.

Line 647: what “21 uncertainties”? Do you mean sources of uncertainties? If so, name a few.

Line 688: refrain from using a non-standard acronym (LI, in this instance) in the conclusion as some readers might skim the paper and only read the abstract and the conclusion initially.

The paper by Ueyama et al. describes eddy covariance and mobile observations of methane and ethane in the Osaka metropolitan area in Japan. It covers an important and relevant research topic. The dataset is quite detailed and unique in that the authors try to combine different datasets from eddy covariance to mobile measurements. Regarding the interpretation of their results, I have a couple comments and questions. I recommend publication after consideration of my comments as outlined below.

Major comments:

Section 2.5: the proposed method of Weller to infer fluxes from road-side observations is fundamentally challenging and uncertain, because it assumes that methane is emitted at street level (e.g. leaks below ground), for which the method has been designed. However there is growing evidence (see also Stichaner et al., 2024, doi: 10.1016/j.atmosenv.2024.120743 ), that methane emissions are largely emitted via standard operation procedures (e.g. pre-flush or partially burned natural gas) and then emitted at roof-top level through smoke stacks. Methane sources are therefore not solely distributed horizontally, but also exhibit a strong vertical gradient within the urban roughness layer. This could easily bias the approach of Weller by a factor 2-3 and therefore the approach can be qualitative at best. One would not expect to see a good correspondence with EC observations conducted above the urban roughness layer. The EC observations in this study also provide indirect evidence that methane is linked to natural gas consumption (e.g. diurnal cycle) rather than leaks, as other cited studies have shown. (e.g. Pawlaw and Fortuniak, 2016, Helfter et al. 2016)

Section 3.4: why would sewage treatment plants be a big methane source? They are typically run aerobically, and methane emissions should be very low for well managed plants.

Section 3.5:  Diurnal variations and weekend to weekday variations suggest sources other than leaks, which would be independent of activity (ie. gas usage) as gas distribution networks are typically maintaining a constant pressure and leak related emissions therefore are largely independent from temperature and activity changes.

Fig. 10: The authors only state that there is a seasonal cycle with two peaks, which is quite unusual (all other sites with multi-year direct EC CH4 observations show a clear seasonal cycle with only one peak). What is the cause of this – can the authors provide a reasoning?  The only explanation I can think of is drastic seasonal changes in the flux footprints, which the authors have not analysed to the extent necessary to understand diurnal and seasonal variations of EC fluxes.

Generally I miss a more thorough analysis of the flux footprints to understand the bimodal seasonal cycle that has not been observed at other sites. Is there a significant sea-breeze effect? What is the diurnal variation of the flux footprint? Could seasonal synoptic changes lead to the methane peak in summer? Are flux footprints changing seasonally? Antropogenic CH4 emissions would be expected to lead to a peak in winter (more heating demand), while biogenic emissions (e.g. wetlands) would show an opposite seasonal cycle. In this context it would be interesting to investigate whether CH4 fluxes in Sakei City are temperature dependent, and whether there is a different temperature dependence in different seasons. Perhaps this could provide additional insights in their interpretation of EC flux data. Are there wetland emissions within the flux footprint or is there evidence of a superemitter (e.g. Stichaner et al., 2024) biasing the seasonal cycle? 

Line 593: Why would sewage treatment plants be such a big source of methane? They are typically operated under aerobic conditions and methane emissions from such plants have shown to be comparatively small. This is actually in line with the street level observations suggesting that the sewage treatment plant is not a big methane emitter, which would be expected for a well operated facility.

Minor comments:

Line 58 - 565: it is Helfter and not Helfer – this typo needs to be corrected throughout the manuscript

Line 56, “The EC method has been used for measuring fluxes over terrestrial ecosystems (Baldocchi, 2014) but has been applied in urban areas to understand decade long greenhouse gas emissions. “ change ‘but’ to ‘and’