the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Nov 2024
Six years of greenhouse gas fluxes at Saclay, France, estimated with the Radon Tracer Method
Abstract. Here, we use carbon dioxide (CO2), methane (CH4), carbon monoxide (CO), nitrous oxide (N2O) and radon (222Rn) data from the Saclay ICOS tall tower in France to estimate CO2, CH4 and CO fluxes within the station footprint from January 2017 to December 2022 and N2O fluxes from February 2019 to December 2022 using the Radon Tracer Method (RTM).
We first performed a sensitivity study of this method applied to CH4 and combined with different radon exhalation maps including the improved European process-based radon flux maps developed within 19ENV01 traceRadon and back-trajectories in order to optimize it. Then, radon exhalation maps from the 19ENV01 traceRadon project, STILT trajectories from the ICOS Carbon Portal, best estimate of radon activity concentration and greenhouse gas data have been used to estimate the surface emissions. To our knowledge, this is the first study using the latest radon exhalation maps and standardized radon measurements to estimate CO2, CH4, CO and N2O surface emissions. We found that the average RTM estimates are 609 ± 402 mg m−2 h−1, 0.81 ± 0.66 mg m−2 h−1, 1.04±1.80 mg m−2 h−1 and 0.063 ± 0.079 mg m−2 h−1 for CO2, CH4, CO and N2O respectively. These fluxes are in good agreement with the literature.
CH4, N2O and CO are in fair agreement with the inventories, though with higher values. CO2 fluxes are about five times higher than modeled anthropogenic and biogenic fluxes combined. The differences mainly occur during summer, and the CO/CO2 ratio points toward a misrepresentation of the biogenic fluxes at this time by the WRF-VPRM version used here.
- Preprint
(4777 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3107', Anonymous Referee #1, 28 Jan 2025
This paper used the Radon Tracer Method (RTM) to estimate greenhouse gas (CO2, CH4, N2O) and CO fluxes at Saclay, France during the period of January 2017 – December 2022. The authors examined the sensitivity of the method to the use of different Radon exhalation maps. Radon exhalation maps from the 19ENV01 traceRadon project, STILT back trajectories from the ICOS Carbon Portal, estimates of radon activities and greenhouse gas data were then used to estimate surface emissions. They found that the estimated CO2, CH4, CO and N2O surface emissions were in good agreement with the literature and that CH4, N2O and CO fluxes were also in fair agreement with inventories. The observation-based RTM method provides an independent approach (alternative to inverse modeling) to verify greenhouse gas fluxes, as demonstrated in this study. This reviewer’s major concern is that the presentation of this paper needs improvement and in some places the texts are hard to understand (see examples below). Publication on ACP is recommended after serious editing and addressing the comments below.
Abstract, Line 12: “CH4, N2O and CO are also in fair agreement with the inventories, though with higher values” – do you actually mean “CH4, N2O and CO fluxes”? “To our knowledge, this is the first study using the latest radon exhalation maps and standardized radon measurements to estimate CO2, CH4, CO and N2O surface emissions” - Is this for any site or for Saclay only? “These fluxes are in good agreement with the literature” – Could you cite the values from the literature for each species?
Page 3, Line 5: GHG and 222Rn “concentrations”?
Page 3, Line 8: Kikaj et al. (2024) – when was this submitted? Not available to the reviewer.
Page 3, Line 10: “the radon flux was considered homogeneous over time and space” – is this said for Paris or Europe? Probably this was an assumption made in the study of Yver et al. (2009)? “as it is now known that the radon fluxes varies on space and time” – it is long known (way before 2009) that the radon fluxes vary on space and time.
Page 4, Line 4: “the nocturnal PBL was above 100m….” – I think you meant the nocturnal PBL height was above the 100 m sampling height of SAC tower.
Page 4, Line 27: “respectively, “ – add “,” before respectively (also check elsewhere in the text).
Figure 1 caption: what is the CCGCRV code?
Section 2.2: Please add references for the Radon Tracer Method at the beginning of this section since this method has previously been used.
Page 5, Line 11: Under which conditions will this (<<) be valid?
Page 8, Line 30: Please add “N” for latitude and “W” for longitude.
Page 9, Line 19: obtained BY
Page 10, Line 7: are showN.
Page 10, Lines 9-11: It’s well known that radon emissions under freezing temperatures in winter are much reduced. Is the higher soil humidity, which prevents the radon from exhaling, due to low temperature in winter?
Page 10, Line 15: remove the redundant “Bq”.
Figure 3: “the fixed flux from the literature” --- which literature?
Figure 4 caption: using either….or both the maps and the footprints.
Figure 5 caption: it’s not clear whether “fluxes” are for Rn or CH4. Please clarify to avoid confusion.
Figure 6: “CH4 2 Flux”?
Figure 8 caption: “On the left panels, …shown, in the middle panel, we show…” - Editing is needed.
Figure 9: what is “por”?
Figure 18, Line 32: FEWER events
Page 20, Line 6: “for the others, it was either the radon increase that was too low or the number of available hours” – Please clarify.
Page 21, Line 6: “an underestimation for the higher ones” – Not clear. RTM overestimates?
Page 21, Line 15: “though soil chambers” – do you mean “through soil chambers”?
Page 21, Line 22: “CO RTM and TNOf fluxes do not show a clear seasonal cycle or a trend over the period” – could you make a seasonality plot?
Page 24, Line 1: “No trend is observed” – this is also mentioned elsewhere. Did you try to do regression analysis?
Page 24, Line 2-3: “we are looking here at nocturnal fluxes without photosynthesis only respiration” – how about “…without photosynthesis (i.e., with respiration only)”?
Page 24, Line 5, Line 14: “in average” should be “on average”; change “like” to “as”.
Page 26, Line 3: do you mean “CH4, N2O and CO fluxes are in fair agreement with the inventories”?
Code and data availability: the ICOS Carbon Portal address is not provided. Both the FLEXPART trajectories and the RTM code are not provided (shared on demand only) but should be archived in a public depository (e.g., https://zenodo.org/).
Citation: https://doi.org/10.5194/egusphere-2024-3107-RC1 -
RC2: 'Comment on egusphere-2024-3107', Alan Griffiths, 25 Mar 2025
General comments
This manuscript reports on the use of the nocturnal-accumulation version of the radon tracer method (RTM), a tracer-ratio method for determining greenhouse gas emissions. In this study, radon-222 is used as a reference tracer, with known emissions, and the emission rate of several greenhouse gases is determined from the observed concentration ratio. Measurements are made over a six-year period from an inlet 100m above ground level.
The RTM has several unresolved issues. This is a fact that the authors recognise, citing Levin et al. (2021), but make the point that the benefits of the RTM make it worth exploring its application to the Saclay data set. The RTM is a relatively simple way to evaluate top-down greenhouse gas fluxes, without an inversion model. The method is therefore supported by a line of evidence which is independent from some of the uncertainties of transport models. I agree that this is a method worth exploring with these data and consider that the topic addressed by this manuscript is ultimately publication-worthy.
At this point in its development, though, there are three major issues which ought to be resolved before publication should be considered.
Specific comments
[1] First, regarding methodology, I am concerned that the STILT footprints (shown in Fig 8) have been calculated or used inappropriately, although this might also be a misunderstanding on my part arising from an incomplete description of the methodology. The footprints are used to determine the influence region of nocturnal radon measurements and therefore to calculate a representative land-surface emission rate. The RTM concerns itself with the relative increase in radon concentration since the establishment of a stable nocturnal boundary layer in the late afternoon, at a time t0. The nocturnal accumulation period lasts until the next morning, so therefore a measurement at time t should have an associated radon flux which is calculated from a footprint integrated over the period (t0, t). In contrast, as far as I can discern, the STILT trajectories have been calculated from 10-day long retroplumes (i.e. backwards trajectories, generalised to account for dispersion), based on the information supplied by the Carbo Europe website, which is given as the source of these footprints. The use of a 10-d long retroplumes explains the very large influence region visible in Fig 8, much larger than even a 10 m/s flow would travel over ~ 10h (360 km, or roughly as far as the eastern border of France). In practice, because the RTM selects nights with relatively strong radon accumulation, RTM tends to bias towards calm nights and the main influence region should be smaller again. The footprint published by Levin et al. (2021), while not directly comparable as it is from a lower height above ground, covers a much smaller region, well constrained within a 150km x 150km box.
Staying on the topic of the footprint calculation, we also see that (1) Saclay is close to a local minimum in radon emissions, according to rightmost column of Fig. 8 yet (2) a comparison between radon emissions averaged over the STILT-calculated footprint vs the Saclay pixel, seen in Fig. 7, shows that the local radon emissions are (apparently) almost always higher than those averaged over the night footprint. I would expect the night footprint emissions to be distributed around the Saclay pixel, because there are higher radon emissions to the southwest but lower radon emissions from pixels immediately east of Saclay. My suspicion is that this seemingly contradictory result is from the use of footprints which extend too far back in time combined with the inclusion of ocean pixels in the calculation. One of the main assumptions of the RTM is that emissions of the tracer of interest are distributed over a similar geographic area as the emissions of radon, meaning that ocean fluxes (where radon emissions are vanishingly small) are out of scope for this method. As mentioned above, my understanding of the RTM is that the footprints should be recalculated with a much smaller integration time (meaning that oceanic pixels barely contribute to the calculation) but in addition, on the rare occasion when a backwards plume travels over the ocean, the radon flux should be calculated from a conditional average and only include land-surface points.
[2] A second major concern is that the authors miss an opportunity to report on the observed trend in greenhouse gas emissions. One main finding of the present work, echoing what others have reported, is that the uncertainty in radon emissions is presently too high for the radon tracer method to be useful for absolute flux estimates. Consequently, it is important to report trends in emissions. Presently, the monthly mean fluxes are reported in Figs 11-13, but the data are noisy and the figure is insufficiently clear to draw conclusions. I recommend that the authors consider performing additional analysis to show a trend (naively, even extending the averaging period might be enough to better constrain the trend). If, with additional work, the RTM is not able to constrain a trend well enough to validate a priori trends in greenhouse gas fluxes then a conclusion stating this, while not a desirable outcome, would nevertheless be useful.
[3] My third concern, which is related to the previous one, is that the overall purpose of the manuscript is not altogether clear. At first glance, the mean GHG fluxes are the main result but the significance of these fluxes is undermined by the uncertainty in radon emissions, even for the most up-to-date radon emissions data (Karstens and Levin 2024). Other results from the manuscript include: a sensitivity analysis assessing the RTM, a brief comment on the VPRM biosphere exchange model, and a comparison between the reported fluxes previous studies. While these are all points worth discussing, I recommend that the main conclusion of the paper should be clarified. In my opinion, candidates for re-focusing the manuscript are either the trend in GHG emissions (observed vs. the inventories) or an analysis of the RTM, but ultimately this is a decision for the authors.
Minor and technical comments
Page 2 Line 22: “sophisticated atmospheric transport modelling” : it may be worth mentioning that the RTM provides a measurement which is independent of an atmospheric transport model, perhaps more important than the fact that it’s easier to implement. (GHG fluxes are important enough to be worth measuring, even if the method is difficult)
P3 L3: My reading of the latest flux maps for Europe (Karstens and Levin 2024) is that the new maps still have ~factor of 2 uncertainty, whereas this text gives the impression of such a large uncertainty being a thing of the past.
P3 L13: The fact that radon emissions are variable in space and time is not especially new, e.g. Schery et al (1984). I also think the flux map from Zhou et al. (2008) deserves to be cited as an earlier example of a country-scale radon emissions map.
P3 L16: I would use the word “updated” instead of “improved” because it’s not clear that the new maps have improved accuracy. The main benefit is that the updated map has a higher temporal resolution, as far as I understand it.
P3 L25: “ICOS Class 1”: Please explain the consequences of the tower being ‘Class 1’
P4 L3: The Pal and Haeffelin study is unlikely to be a good source of information about Nocturnal Boundary Layers (NBLs) below 100m. Their study used a high-power research lidar, ALS-450, which (according to the manufacturer) has full optical overlap at ~300m. It is unlikely to be able to detect NBL below 100m, and indeed their figure shows that the measured NBL *never* drops below 100m, supporting the idea that the instrument may be unable to detect such low layers. Comparing radon concentration at two heights on the tower (or using other meteorological data from the tower itself) might be a useful alternative for quantifying how often the 100-m level decouples from surface emissions.
P4 L4: “most of the time”, it would be better to quantify this, e.g as a percentage of nights, or mention that you quantify this (or alternatively, quantify the proportion of nights when the RTM was successfully used) later in the results
P4 L14: “The values found for SAC…” it is unclear whether this is a range which applies to both observations and models, or if one value is observations and the model. Also, I was confused as to why a model is involved if this is an observed flux? Please edit for clarity.
P4 L16: According to the website, the footprint function is a 10-day integral (i.e. particles are tracked 10-days backwards in time). For this study, the footprint for a nocturnal measurement should only be integrated backwards as far as the previous afternoon when the stable boundary layer began to establish itself.
P4 L28: Regarding the radon detector uncertainty, also quote the sensitivity (expected to be approx. 21 cpm/Bq/m3), based on Chambers et al. (2022) Table 1.
P5, Fig1: “…shows the absolute number of data…” By eye, the radial axis looks like windspeed for the GHG also, please double-check.
P5, L3: “..well-mixed layer..” If it is a textbook case, the NBL will not be well mixed. The assumption of a well-mixed NBL is convenient for developing the RTM equation, but it is not, in fact, a necessary assumption. Here, one could develop the RTM by starting from the footprint-based analysis, as used in the FLEXPART and STILT models, or state here that the mathematical development is simplified – an ‘illustrative’ development of the method.
P5, L5: Some terms in Eqn. (1) are not strictly defined (Δt and ΔC). I can guess that Δt is the time since the establishment of a stable boundary layer, but equally it might just mean ‘a small timestep’.
P5, L5: The radon decay term is an approximation, unless Δt -> 0. It is reasonable to assume this, provided that Δt << (the half-life of radon), but please note the approximation.
P6 L5: “in concentrations” should be “into concentrations”. As an aside, I believe that the opposite conversion would also work (converting radon into a mixing ratio).
P6 L13: “Hence, the radon flux…” Missing from this description is how far back in time the dispersion model is run for to calculate the S-R matrix, which should be only a few hours.
P8 L18: “measurements from csv files”: I don’t understand the significance of this remark, please clarify
P8 L21: It is certainly important to take into account the radon detector’s response time, however deconvolution is optional. The other option, since 1-minute GHG data are available, is to process the 1-minute data with a forward model of the radon detector's response. Computationally, and numerically, this is much simpler. Griffiths et al. (2016), Fig. 8, compares the two options. Either way, Griffiths et al. (2016) is an appropriate citation.
P8 L22: Check that you're correcting to an appropriate reference. In a previous section, you convert GHG measurements from mixing ratio to concentration (presumably concentration at ambient temperature and pressure) so radon measurements should also be reported at ambient temperature and pressure.
P8 L29: “The backtrajectories were calculated…”. There are two time variables in the footprint calculation which should be distinguished (1) the measurement time, and (2) the initial time in the source-receptor relationship. Here, it is not clear which of these two times the “24-h window” is referring to.
P8 L25: FLEXPART-WRF is a version of FLEXPART which takes WRF-model meteorological fields as input, however it is stated that “This FLEXPART model uses ECWMF ERA5 meteorological files as inputs”. This needs clarification.
P8 L29: “0-100m layer” This choice is not unreasonable, but it does make it impossible for the model to indicate when the modelled atmosphere at 100m AGL is decoupled from the surface.
P9 L15: Units of Bq/m2/h are used here for radon flux, but mBq/m2/s elsewhere. Please standardise to one or the other, and also note the justification for this value at this point in the text. It would also be helpful to indicate how this value compares to the European mean.
P10 L1: Here an R2 value of 0.6 is mentioned, which seems overly permissive and perhaps arbitrary? Is it possible to show how the flux estimate changes as a function of the R2 cut-off?
P10 L11: consider “soil moisture” rather than “humidity”, as the latter sounds related to water vapour
P11 Fig 3: “From the literature”: use a citation (abbreviated if necessary)
P11 Fig 3: “User Rn flux”: replace with a more meaningful label
P11 L7: “The standardization…” Consider not showing the non-standardised case. There is no question that the difference in instrument response should be taken into account before calculating a correlation. A more meaningful comparison would be between (1) deconvolution applied to the radon measurements or (2) the radon detector response function applied to the CO2/CH4 measurements. Or move this to a supplement.
P11 L12: “The transport models…” Edit this sentence for clarity
P12 L9: How are ocean pixels handled? If only land-based sources/sinks are being considered (and oceanic air is assumed to be 'background' concentrations) then the calculation should be a conditional average. That is, select land points, and take a weighted average of radon flux (weighted by the footprint).
P13 Fig. 5. There are outlier points, one negative and one high, in this plot. These might be worth remarking on in the text; is there anything unusual about the outlier point, around the 21st in winter? Or the negative flux, seen in summer? Do either of these cases point to the assumptions of the RTM not being upheld?
P15 Fig. 7: I assume that this the same map as the one referred to as "GLDAS Noah" in Fig 8, please use standard nomenclature.
P15 Fig. 7: As mentioned in the Major Comments, from looking at the STILT footprint map, it looks like there should be periods (especially in Summer) when the airmass comes from the SW and passes over radon emissions of > 0.06 Bq/m2/sec (216 Bq/m2/h)
P16 Fig. 8: Include map scale (e.g. with scale bar or by labelling lat/lon) and projection; Units are inconsistent with usage elsewhere; a linear colour scale would be more appropriate for this usage of the footprint OR include a contour which encloses 99% (or some other large fraction) of the footprint.
P17 L8: “However the air measured…” This would be straightforward to resolve by configuring STILT to create a footprint which has hourly-resolved emission time (that is, a 4 dimensional array with dimensions of (measurement time, emission time, latitude, longitude) ). Recording the footprint in this way would also allow the straightforward calculation of radon decay during post-pocessing.
P18 L17: “total uncertainty” is mentioned, please define how uncertainty is parameterised (e.g. one standard deviation, 95% CI, etc.)
P18 L28: Is it valid to present averages? That is, do you believe that outliers are caused by emission events, or would it be more appropriate to present a trimmed mean? The data, after all, are rather skewed.
P19 Fig 10: Not clear (from the figure caption) what time period each of the dots represent (probably nightly?)
P20 L1: “Boiler Room”, if possible be specific and name the facility. Is this a large-scale “district heating” facility, perhaps?
P20 L7: “…the correlation was too low…” It might be helpful to discuss this further, as it's a limitation of the RTM. Was this a period of strong winds, for instance? Or a period of calm conditions and stagnation?
P20 L15: Is the purpose to choose a winner out of EDGAR and TNO? This section seems to *almost* make this point, but could be more clear. It is worth stating, even if the conclusion is the unsatisfactory one that the RTM is too ambiguous to choose.
P21 L27: Is the 'boiler room', discussed as part of the CH4 section, a potential source of CO?
P21 L28: “cadran” -> “quadrant”
P24 L3: “…nocturnal fluxes without photosynthesis …” I agree with this statement, but it seems to contradict an earlier statement that “…, the air measured at midnight at Saclay travelled during the day and it is thus not straightforward to select which hours of the inventory should be favored.”
P24 L18: “1.5” -> “1.5 x 10^-3”
Some acronyms need explaining, e.g. CCGCRV, NRT
The impact of this publication would be increased if the datasets and code used to generate the results in the paper were made available more conveniently, for instance by depositing them with a repository like zenodo.
References (cited in this review)
Chambers, S. D., Griffiths, A. D., Williams, A. G., Sisoutham, O., Morosh, V., Röttger, S., Mertes, F., and Röttger, A.: Portable two-filter dual-flow-loop 222Rn detector: stand-alone monitor and calibration transfer device, Adv. Geosci., 57, 63–80, https://doi.org/10.5194/adgeo-57-63-2022, 2022
Griffiths, A. D., Chambers, S. D., Williams, A. G., and Werczynski, S.: Increasing the accuracy and temporal resolution of two-filter radon–222 measurements by correcting for the instrument response, Atmos. Meas. Tech., 9, 2689–2707, https://doi.org/10.5194/amt-9-2689-2016, 2016
Schery, S. D., D. H. Gaeddert, M. H. Wilkening, Factors affecting exhalation of radon from a gravelly sandy loam, J. Geophys. Res., 89, 7299–7309, 1984.
Zhuo, W., Guo, Q., Chen, B., and Cheng, G.: Estimating the amount and distribution of radon flux density from the soil surface in China, J. Environ. Radioact., 99, 1143–1148, doi: 10.1016/j.jenvrad.2008.01.011, 2008.
Citation: https://doi.org/10.5194/egusphere-2024-3107-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 243 | 76 | 25 | 344 | 12 | 13 |
- HTML: 243
- PDF: 76
- XML: 25
- Total: 344
- BibTeX: 12
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 107 | 32 |
| Germany | 2 | 27 | 8 |
| United Kingdom | 3 | 25 | 7 |
| France | 4 | 23 | 6 |
| China | 5 | 21 | 6 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 107