the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 May 2026
Isotope-based investigation of methane sources in Hamburg, Germany
Abstract. Methane (CH4) is the second most important anthropogenic greenhouse gas and reducing CH4 emissions can lead to climate benefits on the timescale of a decade. Knowledge of the most important sources in different regions is important for designing and implementing successful mitigation strategies. We present a detailed investigation into the source mix of CH4 emissions in Hamburg, Germany, using measurements of the CH4 mole fraction and isotopic composition by combining data from multiple observational campaigns and an atmospheric transport model. Measurements of CH4 isotopic composition were performed for eight months using isotope-ratio mass spectrometry (IRMS) at the Geomatikum building in the city centre, 82 m above ground level. The isotopic composition clearly demonstrates that the observed CH4 enhancements originated mainly from microbial sources. Supporting meteorological and hydrological data provide context for explaining the temporal CH4 variability. The highest observed CH4 enhancements are sharp peaks from microbial sources that occur only during low tides and when air is advected from the south, where the Elbe River and the Hamburg harbour are located. Measurements with a mobile analyser along the river confirm that large emissions occur from the banks of the Elbe River during low tides. Our integrated approach demonstrates the benefit of combining detailed measurements (isotopes and mobile) and high-resolution modelling for accurately attributing greenhouse gas sources in complex environments.
- Preprint
(5624 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1813', Anonymous Referee #1, 02 Jun 2026
-
RC2: 'Comment on egusphere-2026-1813', Anonymous Referee #2, 28 Sep 2026
The manuscript presents a timely and potentially valuable investigation of methane source dynamics in Hamburg, combining continuous atmospheric CH₄ isotope measurements, targeted mobile observations near the Elbe River, hydrological and meteorological data, and atmospheric transport modelling. The evidence for enhanced, predominantly microbial CH₄ signals associated with low-tide conditions and transport from the Elbe/harbour sector is interesting and relevant for improving urban methane inventories, particularly because temporally variable riverine emissions are often not adequately represented.
However, several central conclusions currently appear stronger than the available evidence supports. In particular, the manuscript should distinguish more carefully between identifying episodic microbial-like CH₄ enhancements at the Geomatikum receptor and uniquely attributing these enhancements to exposed Elbe sediments or extrapolating them to the citywide and annual Hamburg methane budget. The analysis would be considerably strengthened by clearer documentation of isotope-background selection, temporal synchronization among observational datasets, uncertainty in Keeling-plot source signatures, the representativeness of the single-site record, and quantitative evaluation of the tide-conditioned transport-model sensitivity experiment. Addressing the following major comments would improve the transparency, reproducibility, and policy relevance of the study.
1- Limits of bulk-isotope source discrimination. The manuscript uses -CH₄ and -CH₄ source signatures, together with wind direction, water level, mobile CH₄/C₂H₆ observations, and transport modelling, to attribute sharp CH₄ enhancements to microbially produced emissions associated with the Elbe River. This multi-evidence approach is valuable. However, bulk carbon and hydrogen isotope signatures alone do not uniquely discriminate among the multiple nearby microbial sources that may influence the Hamburg–Elbe/harbour footprint, including exposed river sediments, wetlands, wastewater, landfill-related emissions, manure, and biogas-related sources.
Please distinguish more clearly between a conclusion that the observed peaks are consistent with a microbial source and a stronger conclusion that they are uniquely attributable to exposed Elbe sediments. The manuscript should quantify, or at least discuss explicitly, the overlap between the observed -CH₄/-CH₄ signatures and representative signatures of plausible local microbial and mixed sources. In particular, the authors should identify which additional evidence—location-specific mobile observations, tide phase, wind-sector footprints, absence or presence of C₂H₆ enhancement, and transport-model sensitivity—rules out or constrains each alternative source.
2- Representativeness of the single-site isotope record and directional sampling sensitivity. The continuous -CH₄ and -CH₄ observations were collected at one elevated receptor at Geomatikum. The manuscript shows that the sharp CH₄ peaks are strongly associated with winds from approximately 120–190°, which corresponds to the Elbe/harbour sector, while broader diurnal enhancements occur under a wider range of wind directions. This directional dependence is central to the interpretation, but it also means that the long-term isotope record samples the Hamburg source mixture through a receptor-specific transport filter rather than uniformly across the city.
Please provide a quantitative assessment of directional sampling representativeness. At minimum, report the frequency of wind observations, measurement availability, boundary-layer height, wind speed, CH₄ enhancements, sharp-peak occurrence, and modelled source sensitivity by wind sector. The analysis should distinguish whether the pronounced Elbe-sector signal reflects comparatively high emissions, preferential transport to Geomatikum, more frequent favourable meteorological conditions, or a combination of these factors.
The authors should also quantify the fraction of the total observation period and of total CH₄ enhancement represented by each wind sector, rather than focusing principally on detected sharp events. This would allow readers to assess whether the inferred microbial dominance applies to the receptor-conditioned signal or to the broader Hamburg source mix. Conclusions should be framed accordingly: the data can demonstrate an important Elbe/harbour influence on the Geomatikum observations, but a single fixed isotope site cannot independently establish the citywide or annual dominance of that source without a spatially distributed observing network or a formally evaluated inversion framework.
3-Novelty and positioning relative to previous Hamburg–Elbe studies. The manuscript presents valuable evidence that sharp, microbially sourced CH₄ enhancements in Hamburg are associated with low-water conditions and emissions from exposed Elbe riverbank sediments. However, the novelty claim should be positioned more carefully relative to existing work. In particular, Forstmaier et al. (2023) had already identified the Elbe River and associated wetlands as an important missing CH₄ source in Hamburg’s emission inventory, demonstrated that including an Elbe emission layer improved the inverse-model performance, and reported biogenic atmospheric CH₄ enhancements associated with tidal conditions. Earlier aquatic studies, including Matoušů et al. (2017, 2019), had also documented substantial methane occurrence and dynamics in the Elbe estuary.
The present manuscript appears to advance beyond this earlier work through its targeted tide-resolved riverbank measurements, isotope-based characterization of the local enhancements, and explicit low-tide emission parameterization in the transport-model sensitivity analysis. The authors should state this distinction explicitly in the Introduction and Discussion. In particular, they should avoid implying that the Elbe–tide–methane association itself is newly discovered; instead, they should define their contribution as providing more direct, spatially resolved, and isotope-supported evidence that exposed sediments can generate strongly enhanced, episodic CH₄ emissions under low-tide conditions.
The authors should also discuss the apparent difference between their low-tide association and the tide-related signal reported by Forstmaier et al. (2023). Clarifying whether these findings reflect different tidal phases, locations along the estuary, transport conditions, or distinct CH₄-release mechanisms would improve the mechanistic interpretation and prevent an overly broad conclusion that low tide alone controls Elbe CH₄ emissions.
4- Quantification of the tide-dependent Elbe source. The manuscript provides compelling qualitative and isotope-supported evidence that sharp CH₄ enhancements observed in Hamburg are associated with microbial emissions from the Elbe/harbour sector under low-tide conditions. However, the analysis stops short of deriving a physically constrained emission estimate for this source. In contrast, the companion Hamburg study by Forstmaier et al. (2023) used an FTIR network and Bayesian inversion to estimate total city emissions of kg CH₄ h, of which kg CH₄ h-1 was attributed to anthropogenic sources.
In the present manuscript, the factor-of-100 increase applied to the Elbe emission layer under selected low-tide and wind conditions reproduces many sharp concentration peaks, but it remains an empirical sensitivity adjustment. The resulting inference is not translated into an areal, bank-specific, tidal-cycle, seasonal, or annual CH₄ flux, and no uncertainty range is provided. This limits the extent to which the results can directly inform emission inventories or mitigation prioritization.
Please consider whether the available concentration, isotope, water-level, meteorological, and transport-model data can constrain at least an order-of-magnitude low-tide Elbe emission flux and its uncertainty. Possible approaches include a formally optimized river-layer scaling factor within an inversion framework; a Bayesian or Monte Carlo sensitivity analysis that propagates uncertainties in atmospheric transport, background CH₄, exposed sediment area, source signatures, and tide phase; or a clearly bounded event-scale flux estimate for the best-characterized measurement locations. If a robust absolute flux estimate is not possible with the present dataset, this limitation should be stated explicitly and the conclusions should be framed as source-process identification rather than quantitative source apportionment.
The policy relevance would be considerably stronger if the authors distinguished between identifying a previously underrepresented episodic source and establishing its contribution to Hamburg’s annual CH₄ budget; the present evidence strongly supports the former but does not yet quantify the latter.
5-Seasonal representativeness of the isotope record. The continuous isotope measurements span 27 July 2021 to 23 March 2022, and therefore include one winter but do not cover a complete annual cycle or interannual variability. The manuscript itself indicates seasonal differences: sharp microbially sourced peaks occur more frequently in autumn than winter, whereas relatively C- and H-enriched source signatures occur more frequently in winter. These patterns suggest that the observed source mix is seasonally variable and that the eight-month record may not be representative of the annual Hamburg CH₄ source composition.
Please provide a month-by-month summary of data availability, measurement completeness, CH₄ enhancement frequency and magnitude, and the derived -CH₄ and -CH₄ source signatures, including associated uncertainty and sample/event counts. The analysis should explicitly test whether the observed isotope signatures differ among summer, autumn, and winter, rather than relying on a qualitative seasonal interpretation.
This is particularly important for the paper’s conclusion that microbial sources dominate the observed enhancements and for the proposed relevance to emission inventories and mitigation. Hoheisel and Schmidt (2024), using a six-year continuous isotope record in Heidelberg, found a 5.8 ‰ annual cycle in inferred -CH₄ source signatures, with substantially more 13C-depleted values in summer than winter, highlighting that seasonal variation in biogenic CH₄ emissions can materially affect urban isotope-based source attribution.
The manuscript should therefore clearly distinguish conclusions supported for the July 2021–March 2022 observation period from any claims concerning annual or longer-term Hamburg emission patterns. In particular, the targeted riverbank campaign was conducted only during April–May 2023; its low-tide emission observations should not be extrapolated to annual Elbe emissions without multi-season measurements or a quantitative uncertainty analysis.
At minimum, the authors should provide seasonal stratification of their existing data and report whether the inferred microbial contribution, peak occurrence, isotope source signatures, and tide-associated event frequency are robust after controlling for boundary-layer height, wind direction, and wind speed. If the available data do not permit such analysis, this should be acknowledged as a limitation rather than treating the reported period as representative of annual source dynamics.
6-Uncertainty and robustness of isotope-based source attribution. The conclusion that the sharp CH₄ enhancements are predominantly microbial relies on source signatures derived using a Keeling-plot approach and on comparison with literature- and measurement-derived isotope ranges. The manuscript describes the basic Keeling-plot procedure and reports aggregate -CH₄ and -CH₄ source signatures, but the uncertainty analysis is not yet sufficient to assess the robustness or reproducibility of the attribution.
Please provide, for each retained sharp peak and for the river-proximal samples, the number of observations, CH₄ enhancement range, fitted -CH₄ and/or -CH₄ intercept, intercept confidence interval, regression diagnostic, and the criterion used to define the event-specific background. The authors should also clarify how the 10 ppb enhancement threshold was selected and test the sensitivity of the inferred source signatures to plausible variations in this threshold, peak-selection criteria, temporal background window, and regression method.
In particular, the current interpretation appears to establish a microbial-type isotopic signature but does not quantitatively demonstrate that the signal can be uniquely assigned to exposed Elbe sediments rather than other microbial sources within the relevant footprint, such as wastewater treatment, wetlands, landfill-related emissions, manure, or biogas sources. Please distinguish clearly between (i) evidence for a microbial versus fossil origin and (ii) source-specific attribution to the Elbe River or exposed riverbank sediments.
The forward-model isotope calculations also rely on assumed source-category signatures, including literature-derived values for anthropogenic categories and local averages for the Elbe layer. Please tabulate all adopted endmember values, their provenance, representativeness, and uncertainty, and include a sensitivity analysis showing whether the principal conclusions remain unchanged when plausible source-signature ranges are used. Defratyka et al. (2025) demonstrate that the derived -CH₄ source signature from near-source atmospheric measurements can depend materially on data-processing choices, including regression approach, background treatment, and averaging strategy, particularly for lower-precision measurements. A transparent robustness analysis is therefore needed before the source attribution can be compared rigorously with other studies.
7-Definition and sensitivity of the background in isotope mass-balance analysis. The manuscript applies a Keeling-plot approach to infer isotope source signatures of CH₄ enhancements. Although the Methods state that the background is assumed to remain stable during individual sharp and diurnal peaks, the operational definition of the background is not sufficiently documented to permit reproducibility or evaluation of its influence on the fitted intercepts.
Please state explicitly, for each continuous and mobile dataset: (i) how the background CH₄ mole fraction and -CH₄/-CH₄ values were selected; (ii) the temporal window and statistical criterion used; (iii) whether background values were conditioned on wind sector, boundary-layer state, air-mass history, or tide phase; (iv) whether pre-peak and post-peak observations were treated separately; and (v) how the selected background and event window were propagated into the uncertainty of each inferred Keeling-plot intercept.
At minimum, the manuscript should report the background concentration and isotope composition used for every retained Keeling plot, together with the selected event interval, number of observations, intercept uncertainty, and regression diagnostic in a supplementary table.
8-Quantitative evaluation and independence of the transport-model analysis. The transport-model analysis is central to the manuscript’s attribution of the sharp CH₄ enhancements to tide-dependent emissions from the Elbe River. The paper usefully shows that the baseline simulation fails to reproduce the sharp peaks and that a tide-conditioned enhancement of the river emission layer can reproduce many of them. However, the evaluation remains primarily qualitative and is based on the same observations used to motivate the low-tide parameterization.
Please provide a systematic quantitative model-evaluation analysis for both the baseline and tide-adjusted simulations. At minimum, report mean bias, mean absolute error, RMSE, correlation coefficient, regression slope/intercept, and the number of observations or events evaluated. These metrics should be stratified by the full record, sharp peaks, diurnal enhancements, low-tide versus non-low-tide periods, and wind sectors associated with the Elbe/harbour versus other directions.
The authors should also evaluate timing explicitly: quantify the observed and simulated lag of CH₄ enhancements relative to low tide, compare peak onset, maximum, duration, and decline during sediment resubmergence, and report how often sharp peaks are correctly reproduced, missed, or falsely predicted. This analysis would distinguish whether the model captures a tide-dependent emission process or merely matches selected events through an empirically imposed low-tide scaling factor.
Finally, because the 100-fold river-emission multiplier appears to be selected using the observed sharp-peak events, the model result should not be presented as an independent confirmation of the tidal mechanism. Please perform an out-of-sample test—for example, calibrate the river-emission scaling on a subset of events and evaluate it on withheld periods or other low-tide episodes—or explicitly describe the exercise as a sensitivity analysis rather than model validation.
A supplementary event table should list, for each identified sharp peak, the observed and simulated peak time, observed and simulated amplitude, water level, wind direction and speed, boundary-layer height, model footprint, baseline-model residual, tide-adjusted-model residual, and classification as correctly reproduced, missed, or falsely predicted.
9- Temporal synchronization of CH₄/isotope, meteorological, hydrological, and transport-model data. The manuscript reports wind observations averaged over 10-minute intervals, whereas the IRMS system samples CH₄ and isotope composition at 20-minute intervals for -CH₄ and 30-minute intervals for -CH₄. Water-level data are hourly, and FLEXPART footprints are generated hourly. However, the manuscript does not state how these datasets with different temporal resolutions were aligned for the wind-direction analysis, peak characterization, tide-offset calculation, isotope source-signature analysis, and model–observation comparison.
Please provide a reproducible temporal-matching protocol. For each IRMS observation or identified peak, clarify whether wind speed and direction were calculated as a vector mean over the exact sampling interval, assigned from the nearest 10-minute record, or averaged over another window; how hourly water levels were interpolated or assigned to peak times; and how 20–30-minute measurements were paired with hourly FLEXPART footprints. The authors should also evaluate whether the central association of sharp microbial CH₄ peaks with low tide and southeasterly winds is robust to reasonable alternative temporal matching windows and time lags.
10-Consistency and temporal representativeness of boundary-layer-height data. The inference that sharp CH₄ peaks are associated with shallow boundary layers is important to the source-attribution framework, but the BLH dataset appears to combine direct Doppler LiDAR observations available only from 27 July to 4 August 2021 with ERA5 BLH estimates for the remainder of the July 2021–March 2022 isotope campaign. The targeted riverbank campaign, however, was conducted in April–May 2023, and the manuscript does not clearly identify a contemporaneous BLH dataset for these observations.
Please clarify precisely which BLH product was used for each analysis period, how the ERA5 BLH fields were “corrected” using LiDAR wind data, and whether this correction was calibrated and validated against contemporaneous LiDAR-derived BLH. Report the LiDAR–ERA5 comparison statistics during their overlap period, including bias, RMSE, correlation, and dependence on season/time of day where possible.
The authors should also test whether the conclusion that sharp peaks occur preferentially at BLH below 500 m is robust when using ERA5 alone, LiDAR observations alone during the overlap period, and reasonable alternative BLH thresholds. Finally, the manuscript should distinguish clearly between the 2021–2022 isotope evidence and the April–May 2023 riverbank observations. Because these datasets were collected in different years and seasons, the 2023 campaign should be presented as corroborating evidence for a plausible emission mechanism rather than direct meteorological validation of the 2021–2022 peak events.
11- Studies in Rotterdam (Varon et al.), Paris (Defratyka et al., 2022, ACP), and London (Helfter et al.) all used denser measurement networks with multiple isotopic samplers and/or continuous analyzers to achieve city-scale attribution. Hamburg, a city of comparable complexity, is served here by a single 8-month IRMS record. The authors should benchmark their methodology and uncertainty levels against these peer studies and explicitly justify why a less intensive approach is appropriate or sufficient.
Minor comments:
- Second paragraph: Please use the most recent sources and strengthen the scientific rigor of this paragraph. Please also double-check the reported value of 0.6 °C.
- Third paragraph: Please provide appropriate references for studies using atmospheric transport models.
- Line 85: Please ensure that the ¹³C and ²H isotopes are expressed using the correct notation.
- Please provide the uncertainty/accuracy of the wind and weather sensor used in the study.
- Page 5: Please clarify how many air samples were collected.
- Paragraph 145: Please revise this paragraph for clarity, as the formulas and symbols are currently not displayed or explained clearly.
- Equation 3: The equation does not appear to be presented correctly and should be checked and revised.
Citation: https://doi.org/10.5194/egusphere-2026-1813-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 435 | 159 | 50 | 644 | 43 | 49 |
- HTML: 435
- PDF: 159
- XML: 50
- Total: 644
- BibTeX: 43
- EndNote: 49
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments
This is a very interesting manuscript on the topical subject of methane emissions. It is well-written throughout, although I have many small requests for increased clarity in the detailed comments. The topic of the tidal emissions is very important and I wonder if the tidal range during the spikes could be explored more as during the autumn period when many of these events were identified there could be much lower water events and far more mud flat exposed.
The discussion is very Hamburg centric and there isn’t much comparative evidence from rivers beyond the Elbe. I think that the discussion could be strengthened with information about other locations where this tidal emissions phenomenon has been observed, to give a global perspective on this emissions inventory omission.
One thing that did annoy me slightly, and it just needs a little more clarity, is the frequent mention that the emissions only occur when the sediments are exposed, while at the same time saying frequently that they only occur when the river level is below 600cm. What is the 600cm? Is this the depth at the centre of the river? This is why it is important to include the tidal range relative to an average river water depth, which I presume is significantly greater than 600cm. For example, is it 700 ± 100cm when there is a small tidal range and 700 ±200cm when there is a large tidal range?
Detailed Comments
Introduction
Line 27 – in situ implies that the instrument is stationary. You need a bit more clarity between instruments that are mobile such as the Picarro G2210i and IRMS that are moved between different fixed sites.
Methods
Line 63 - ‘At a height of 83 m’ would be better.
Line 73 – Using ‘we’ may be appropriate in many contexts, but I don’t think it is for remote automated measurements.
Line 76 – Inlet is attached to the 70 m balcony. This is 13 m below the top of the building. Is the inlet here or at the top? Earlier you say that wind flow is not obstructed by buildings. Where are the meteorological sensors situated?
Line 85 – The 10 ppb mole fraction reproducibility is based on the IRMS peak. Is this taken into account in the modelling errors? Have you compared this with averaging of 1Hz mole fraction measurements of CH4 at the time of inlet sampling?
Figure 1 - Would be helpful to know what type of farms these are as only some types are going to be significant contributors of CH4. What is the difference between greenery and forest? Do you mean grassland?
Line 102 - When you are measuring in a tidal setting and also talking about high buildings, I wouldn’t use elevated for higher concentrations as it could be confused with height.
Line 106 – Did you use the same inlet line and height for the Picarro as for the IRMS previously in 2021?
Line 113 – For citations be consistent as per journal requirements, either alphabetical or chronological.
Lines 114-117 – Is it just a wind station or a weather station. Previously you say that the inlet is at 70 m and here you say that the weather station is at 80 m, but only 1.3 m from the air inlet. I presume there is less turbulence near the top of the building as in your discussion of location on the highest building in the area.
Line 128 – What is the minimum length for diurnal peaks? It is the inversion height that rises and lowers and the peak height responds to that, so the peaks in the summer months may be less than 8 hours duration, which could be less than your maximum for spikes.
Line 132 - Keeling and Pataki both did Keeling plots for CO2 measurements and Rockmann et al. were not the first use this method for CH4. This was used back into the 1990s.
Line 145 - Keeling plot intercepts are not always the same during the rise and fall of a peak due to atmospheric dynamics of diurnal inversion development and break-up. Are you using the whole of the diurnal peak to generate the plots or just the rise?
Line 147 – I would re-emphasise here that you are using the CH4 mole fraction estimate from the IRMS peaks with a precision of ±10 ppb, as this will contribute to your reliability cut-off point.
Line 161 – Needs more clarity here as you talk about total domain emissions of 18.3 Tg/yr, more than the whole of the EU, but earlier you mention a domain of 200 km x 160 km. I presume that the emission is for the whole of the FLEXPART model area, which is much bigger than the domain area. Just needs minor restructuring of the paragraph.
Table 2 – What year does your background isotopic signature represent? It is now closer to -48‰ for mid and high northern latitudes, but you are modelling 2021 data, so this should be clarified.
Results
Lines 184-185 – A citation for this known general phenomenon would be helpful for this type of modelling.
Figures 3 and B1 = You have a pyrogenic category in the legend, but your graphs don’t cover a big enough range of isotopic space to show it. Why do the simulations show a straight line with very little deflection toward the centre of the waste domain?
Figure 3 caption – Took me a while to realise what the coloured bars are showing so maybe ned to add ‘from light (Aug 2021) to darker (Mar 2022)’
Line 210 – ‘The sharp peaks occur more frequently in autumn than in winter.’ Have you considered greater tidal range in autumn as a contributing factor - more exposed mud flats?
Line 221 – What do you mean by ‘as the emissions of waste are relatively stable.’ The isotopic signature of waste for the region could be stable even if the emissions change significantly.
Line 244 – 6 m/s is not low and from the range presented the average could be higher than the 3.2 ± 1.9 m/s in the previous paragraph.
Figure 5 – It is surprising that you have such low boundary layer height for such high wind speeds. If the sample population is the same in (d) and (f) then there must be some boundary layers of <200m for wind speeds >10 m/s. If it isn’t the same peak population then please make it clear.
Figure 6 caption – ‘when the river level was below 600 cm.’ See general comment at start. What does the 600cm relate to, because for me 6 m water depth is not going to expose any river bed.
Figure 7 caption – As you have changed instruments for the 2024 survey it is worth mentioning that the raw data is 1 Hz frequency Aeris data.
Section 3.5.2 – this is where I think that some discussion of tidal ranges would be useful, as mentioned in the general comments, as it relates to the amount of sediment exposed. Figures D1 and D2 seem to show this information for short periods of time and appear to show tidal ranges of around 4 m.
Line 315 – Location 9 is close to the agricultural area. Could run-off from the fields be influencing or enhancing CH4 emissions at this location.
Conclusions
Lines 346-347 – Do you really think that you have a distinct isotopic signature for the rivers from the agricultural sources? It is not obvious from the isotopic cross plots. If you are convinced of it I would include a short discussion section emphasising the evidence.
Supplementary Information
The Appendix D header is before Figure C3.
Figures D1 and D2. It is not obvious in the lower parts which parameter is in red and which is in blue. Why are the wind speed scales so different between the two figures?