the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A WRF-Chem study on the variability of CO2, CH4 and CO concentrations at Xianghe, China supported by ground-based observations and TROPOMI
Abstract. The temporal variability of both surface concentrations and column abundances of CO2, CH4 and CO at the Xianghe site in China are analyzed with the Weather Research and Forecast model coupled with Chemistry (WRF-Chem). Simulations of these in situ (PICARRO) and remote sensing (TCCON-affiliated) measurements are produced by the model's passive tracer option, called WRF-GHG, from September 2018 until September 2019. Our analysis found a good model performance with correlation coefficients between observations and simulations up to 0.85 for CO2 and 0.69 for CO. Key source sectors for every gas are revealed by tracking the anthropogenic fluxes in separate tracer fields. While there are slight variations in the relative impacts of these source sectors between surface and column observations, owing to differences in the sensitivity footprint of each observation type, the primary sectors influencing the various species are evident. For CO2 the industry, energy and biosphere sectors are found to be the primary contributors to the total simulated concentration, whereas CH4 concentrations are predominantly attributed to the energy, agriculture and residential & waste sectors. For CO, industry is the largest contributing sector at Xianghe, followed by residential and transportation sources. Differences among the various observation types were particularly visible in the contributions of the biosphere to CO2 and the energy sector to CH4, as their largest sources are located further away from Xianghe. Further, the influence of meteorological factors on the variability observed in the different time series was analyzed. We found that southwest winds typically bring polluted air masses from the North China Plain to the site, while northern winds are associated with cleaner conditions. Variability in surface measurements is primarily driven by the daily cycle of accumulation and atmospheric mixing linked with the planetary boundary layer height. Furthermore, the study demonstrates the ability to detect strong regional sources at Xianghe depending on wind direction. To address inconsistencies between the simulations and observations of CH4, we looked at TROPOspheric Monitoring Instrument (TROPOMI) satellite observations. We found that the model underestimation of CH4 in summer and overestimation in winter may result from a combination of a similar bias in the lateral boundary conditions and an incorrect monthly variation of the CH4 emissions in the agriculture and/or waste sectors of the CAMS-GLOB-ANT inventory over north China. Additionally, WRF-GHG simulations indicated a possible overestimation of coal mine emissions nearby Tangshan, which could not be confirmed nor contradicted by the TROPOMI observations. In summary, our findings highlight the value of WRF-GHG to interpret both surface and column observations at Xianghe, offering source sector attribution and insights in the link with local and large-scale winds based on the simultaneously computed meteorological fields. However, given the long lifetime of the considered species and the fact that WRF-GHG is a regional model, accurate initial and lateral boundary conditions remain crucial. The dependence on precise input emission data on the other hand, can be used to evaluate the existing bottom-up inventories.
- Preprint
(12854 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2023-2103', Anonymous Referee #1, 06 Dec 2023
Review Comment on "A WRF-Chem study on the variability of CO2, CH4 and CO concentrations at Xianghe, China supported by ground-based observations and TROPOMI" by Callewaert et al.
The article presents an analysis of the temporal variability of surface concentrations and column abundances of CO2, CH4, and CO at the Xianghe site in China using WRF-Chem. The study spans from September 2018 to September 2019, utilizing in situ and remote sensing measurements. The model's passive tracer option, WRF-GHG, is employed to simulate these measurements. The study reveals key source sectors for each gas and examines the influence of meteorological factors on observed variability. Additionally, the study addresses inconsistencies in CH4 simulations with TROPOMI satellite observations, suggesting potential biases in emission inventories and lateral boundary conditions.
Major Concerns:
The literature review is narrowly focused on recent studies, with minimal references before 2017. This approach overlooks significant earlier research, especially in GHG model-data comparisons in urban settings, as seen in projects like INFLUX, LA Megacity, and European initiatives like Mega-Paris and COBRA.
The study's sensitivity tests concentrate on PBL, surface layer, and radiation schemes, but the rationale for this focus is unclear. Notably, the land surface model variety, crucial for PBL variations, is absent, which is a considerable oversight in the experimental design. See Díaz-Isaac et al., 2018 (ACP).
The method to compensate for WRF-Chem's limitation at 50 hPa for XCO2 calculations involves integrating TCCON a priori profiles. A more consistent approach would be to use the global-modeled CO2 that provides lateral boundary conditions, here CAMS values, above 50 hPa, ensuring a fairer comparison with TROPOMI results. See Butler et al., 2020 (Atmosphere)
The model's performance is somewhat overstated. For example, a mean bias of -1.43 ppm in XCO2 is significant. The high CO2 correlation coefficients are likely influenced by seasonal variation, necessitating detrending for accurate assessment. Additionally, the model's capture of the CO2 diurnal cycle shows notable discrepancies, especially during nighttime. This also reflects the lack of a literature review on the authors’ end.
The authors' conclusions about error sources, particularly the underestimation of XCO2, seem speculative and lack a detailed description of the model configuration, including aspects like data assimilation and restarts. Without the info, it’s difficult for me to judge if the authors’ interpretation makes sense or not.
A Lagrangian approach might be more suitable for tracing back signals to their sources and sinks, as the current Eulerian perspective may not effectively disentangle these signals.
The manuscript suffers from poor organization and numerous language and grammatical errors, making it challenging to follow. For instance, the placement of CO discussion in the CO2 section and the absence of a dedicated section for CO2 biases, despite their significance, are confusing.
Specific Comments:
- Line 67: Expand citations to include key studies like Feng et al., 2016 (ACP), Lauvaux et al., 2016 (JGR-A), etc. See my comments above regarding those urban projects for more references.
- Line 105: Reevaluate the relevance of the Fast et al. (2006) citation.
- Lin e 165-167: Clarify the intended message for better understanding.
- Line 186: Correct the statement about TROPOMI's orbit.
- Line 201: Provide justification for the chosen two-week spin-up period. My experience is that sometime after 20 days, I still see influences from initial conditions.
- Figure 6: Narrow down the wind direction ranges for NW and SW.
- Figure 7: high correlations after diurnal cycle removal are expected in terms of synoptic scale variations.
- Line 333: there is a surface layer under the PBL.
Considering the major concerns and fundamental issues in study design and manuscript presentation, I recommend a comprehensive revision of the manuscript. The authors should address the concerns in detail, improve the organization, and enhance the clarity of the language before the manuscript can be considered for publication.
Citation: https://doi.org/10.5194/egusphere-2023-2103-RC1 -
RC2: 'Comment on egusphere-2023-2103', Anonymous Referee #2, 19 Dec 2023
Review of 'A WRF-Chem study on the variability of CO2, CH4 and CO concentrations at Xianghe, China supported by ground-based observations and TROPOMI'
manuscript by: Callewaert et al., review by Anonymous Reviewer
December 2023
Overview
The authors present an overview of the performance of their WRF-Chem modelling framework, focusing on comparing the model performance against the in situ and column measurements of CO2, CH4 and CO performed at Xianghe, a measurement site in the vicinity of Beijing, China, over a period between September 2018 to September 2019. In order to alleviate the issues with interpreting regional aspects with localized observations, they also use spaceborne observations from TROPOMI instrument where CH4 is discussed.
The authors do not use a typical (Intro - Methods - Results - Discussion - Conclusion) paper structure. After the introduction (Sec. 1), describing the measurement site (Sec. 2) and model framework (Sec. 3), in which authors describe the input data as well as sensitivity studies performed to select the final model configuration. Further, they describe the methodology of comparison against TROPOMI data (concerning only CH4) in Sec. 4, after which they sequentially analyse various aspects of the model-observation comparisons. In Sec. 5, the overall model results and their comparison against observations are discussed, and larger issues are identified. In Sec. 6, the authors discuss primary sector-contributors to the observed signals for each analysed compound. Then, in Sec. 7, the reasons for seasonally-varying bias in CH4 is analysed with particular attention to the emission data, further supported by TROPOMI observations. In Sec. 8 the authors are discussing meteorological causes for FTIR model-data comparison, as well as PBL variability at the site and its influence on detecting local emissions, arriving at possible discrepancies in the local attribution of coal-mine methane fluxes from nearby coal-mining area. Sec. 9 concludes the paper. The manuscript also contains two appendices, Annex A containing additional details on the sensitivity study discussed in Sec. 3, and Annex B containing extra illustrations with supplemental information.
The modelling framework is set up very well, with clear indication of the authors knowledge about practicalities of running such complex system for the purpose of comparisons against the real-world scenarios. The results demonstrate that the selected configuration was well chosen and I commend the authors on their setup. Thanks to that, the study provides much interesting information on the greenhouse gas transport in the regionally- and globally relevant GHG emission region of NE China.
The manuscript, however, suffers from analyses that feel unfocused and too shallow. In multiple places the authors appropriately identify issues that require further study, but in none of the cases (e.g. bias in long-term CH4, potential underestimation of coal-mine CH4 sources) the analysis goes deep enough to determine whether the hypotheses brought up to explain them are adequate.
Here follows a list of some major concerns:
- The unorthodox structure of the text makes it at times difficult to follow. The flow of information seems to reflect way the experiments were performed, describing the 'results > analysis > extra-results-to-test-hypothesis-X > analysis' chain, but this might not be the best choice for the final publication. I strongly recommend that the authors streamline and clarify the text structure.
- In multiple places in the manuscript, the quantitative analysis and/or information is needed and is not given. For example, in L213 the authors write that "a similar pattern [of negative bias] was found when comparing CAMS reanalysis data set with the TCCON data at Xianghe and other sites in that part of the globe", but do not provide neither the size of that bias nor the reference to the study where it was published. Another example, in L404: "We find slightly elevated XCH4 values nearby the coal mines of Tangshan (∼6◦N, 118.4◦E) in both WRF-GHG and TROPOMI maps" -- it is unclear what does "slightly elevated" mean here? Numeric value here would be objective and reproducible. I've listed some of the places where more quantitative approach is needed below (see "Detailed comments").
- Seasonal bias of CH4 is heavily discussed, but the authors do not attempt to quantify how much of this effect is related to background variability, and how much is local. While this doesn't allow for direct data to model comparison, this could still inform on relevance of regional vs local influences.
- Discussion on biases for CO2 is limited to 7 (seven) lines and focuses on the jump of summer of 2019. Looking at the data quality makes me wonder about stability of the measurement system at that time (especially considering concurrent gap in the insitu system - was that related?), but regardless of that, the latter part of the data is ignored, the CO2 discrepancy is still worth investigating.
As it stands, the paper requires major revision, with second review recommended after addressing the issues listed. Considering already large amount of material that is present in the manuscript, it might be worth considering to divide local and regional aspect into separate, focused, more detailed studies.
Below please find, other, more detailed comments.
Detailed comments:
L4: passive tracer is also an option from base WRF. In order to avoid confusing GHG module with that separate module, I suggest: "are produced by model's greenhouse gas module WRF-GHG"
L5: add correlation coeff. also for CH4 for completeness
L9: 'sectors' are usually used for anthropogenic activities, 'biosphere sector' is rarely used (if at all). Consider using 'biosphere fluxes' or 'biospheric activity' / 'NEE (Net Ecosystem Exchange)' instead throughout the text.
L10: 'Residential \& waste' is meant as a clumped sector as defined in this work, but it requires reading the paper to understand its usage in the abstract. Rephrase / clarify.
L32: "also been rising for the last 200 years"
L63: It's sufficient to ignore chemical reactions change if the residence time in the limited area domain is short enough that the change in mole fractions is smaller than the precision of the measurements. It can be shown for CH4, however, assuming exponential decay, over five days it can already decay by over 2 ppb, and depending on the domain size the residence time of air can be even longer -- depends on the circulation. In most of the cases this can be ignored, but I believe this should be addressed in here for both CH4 and CO, because the domain is relatively large. It should also be discussed later in the paper, where bias of CH4 is described, as the effect of omitting the CH4 destruction could have some effect on observed differences.
L78: Please provide more exact location of the site, at the moment this points to a random location when checked on publicly available mapping portals.
L88: I've never heard about calling CO 'Xgas' before. While interesting, I would suggest removing this.
L91: The instrument is installed on a tower, or is there tubing coming from the tower? Please clarify.
L95: typing error in Sentinel-5 ('-nal')
L97-99: "In our study, we use...". Please add appropriate reference to the dataset.
Same place: Also, evaluation of L2 product is discussed but what about evaluation of L3 that is used here?
L101: -0.6 % and -0.39 % translate into roughly 11 and 7 ppb, respectively. I have some reservations against treating such bias as 'small'. It is substantial.
L102: 'demonstrate great' quality of TROPOMI XCH4 data' - see above. To be clear, it would be absolutely unfair to say that TROPOMI provides poor quality data, but it is an extremely challenging measurement and we should be careful and realistic about what can be achieved.
L126: 'choice of flux inventory is likely important for...' -- understatement, consider '...choice of flux inventory is critical for...'
L137-: I recommend removing 3.2 as separate section (together with Table 1). Extract only information necessary for the manuscript (final config - Table 2) and put it in the description of "WRF-GHG modelling system". Move everything else to the appendix, where it fits better at the moment.
L139: Consider relabeling test case E as BASE (or A).
L147: Where the anthropogenic fluxes also distributed vertically? Relevant since the discussed station is located near lots of anthropogenic activity. See: Brunner et al., 2019
L155: Please add info on TCCON measurement frequency for clarity.
L188: Please provide info on the regridding method.
L197: when comparing to TCCON, WRF output for methane was extended. Why not extend also here? At the very least this extension of 40 ppb could be added to the background (or as extra 'offset') to WRF numbers to avoid explaining the shifting of the colour scales when discussing each figure.
L213-215: The 2 ppm bias is brushed over, and the reference to comparison CAMS-TCCON is missing. See also major comment no. 2.
L221: Again comparison CAMS-TCCON mentioned without reference. Numbers should be given, with uncertainty if possible.
L230: In previous section, the authors followed the order CO2 - CH4 - CO when discussing compounds. Now this order is reversed. Would make it easier to navigate if the ordering was consistent.
L231: 'According to WRF-GHG' > 'Based on our results'
L233: 'Energy sources and biomass burning are not important for the observations at Xianghe. Both residential and transportation tracers show larger values in winter, which is in agreement with higher emissions in that period of the year due to colder air temperatures.' - all these statements would use support from numbers (comment relevant throughout the manuscript).
L240: 'The main sectors contributing to 240 the CO2 data at Xianghe' - consider: 'The main sectors contributing to the modelled CO2 variability at Xianghe'
L241: Note that CO2 for VPRM model in WRF relies on modis surface classification and will predict zero fluxes over urban areas, while in fact CO2 fluxes from urban biosphere can still be substantial. Unless VPRM was modified to handle these fluxes, this might have higher than anticipated impact if urban classification dominates in the vicinity of your station (in the model).
L249: '...the year, which is in agreement with the general emissions patterns in China' - add a reference.
L254-L256: Again, numbers needed to support the statements about emission contributions.
L276-280: 'Unfortunately, the source...'. Maybe I misunderstand. The authors suggest that factors for CH4 emissions from agriculture are constant - and yet there is a peak in emissions? That sounds like a bug in the emission preprocessing code. Please clarify.
L286-289: 'In CAMS-GLOB-ANT, the waste sector is the most important one in the Xianghe region...' - numbers needed; also, why does this need support from Fig4, where direct emissions are available?
L299-300: '...ensemble against GOSAT observations by Parker et al. (2020), a general underestimation of the seasonal amplitude in China 300 was found. This would mean an underestimation of the wetland CH4 emissions in summer.' - Needs clearer phrasing. As it stands the sentence is imprecise. Note that the underestimation of amplitude doesn't automatically translate into underestimation of wetland emissions.
L325: 'More specifically, we looked at the daily mean column concentrations above the background for every wind direction...' Daily means from time when observations were available, correct?
L347-L365: While the importance of PBL height is paramount to the correct interpretation of in situ data, this section doesn't bring a lot of insight into the discussion. It could be quite illuminating if expanded.
L352: 'This stable nocturnal layer is quite shallow and...' - how shallow? Numbers needed.
L361: 'Remark that WRF-GHG is very well capable at simulating this diurnal variation of both CO2 and CH4 in situ observations.' - Again, quantitative analysis would be more than welcome. Also: 'Remark that' > 'Note that'
L372: 'enhancement (sum of all WRF-GHG tracers above the background) is smaller' - delete fragment in parentheses (redundant)
L381: 'The highest values overall...' - how high?
L382: '...are coming from the east. To the east are...' > '...are coming from the east, where...'
L384-385: Concluding the cause before proof is presented. Overestimation of coal-mine emissions cannot be argued before next section, where TROPOMI data is used as extra argument. By this point of the text also bias in the night-time PBL height can explain the observed CH4 in situ bias.
L388: Section 'Assessing CH4 emission sources' is very interesting but sadly mostly based on subjective visual analysis of the figures. More quantitative results should be presented to support the conclusions.
L396-L403: Please briefly expand here. I assume this is a mixture of cloud + albedo effects. Is there enough confidence that the issues causing missing TROPOMI data are completely gone in the areas immediately neighbouring the "white spots" on the map in Fig12?
To clarify what I mean: it is clear that the highest discrepancy is present immediately to the east of 114 deg. E, where many of the gridded cells with those high values immediately neighbour the area of missing TROPOMI data. If the filtration procedure used for TROPOMI is not 100 \% sure, then the discrepancy might be an artefact rather than real signal.
I would appreciate if this can be addressed.
L406: Why not add 40 ppb as offset and use same scales? Would be easier to interpret. See comment above.
L421: In some sections results from 9 x 9 km2 are used for comparisons.
L430: '...a small underestimation of -1.43 ppm and -3.03 ppb with respect to TCCON XCO2 and XCH4, respectively.' Please clarify that these are averages. Add uncertainties.
L431: 'slightly overestimated' - I don't think 'slightly' fits here, these numbers are substantial (again, to be clear: I do not mean the results are bad)
L438: 'This difference is likely because of the lack of strong photo-synthetically active vegetation in the neighborhood of Xianghe.' - in the model or physical world? Or both?
L439: 'For CH4...' > 'On average, for CH4,...'
L453-454: 'peak values are found in the early morning around sunrise when local emissions are strongest...' - Please clarify. Not all the compounds have strongest emissions in the morninig (CH4 for example).
Table 3.: Consider reordering columns, having 'This study' as first for easier reader experience.
Figure 7: It appears that the fonts in this figure are smaller, but maybe it's my eyesight. Please check. Also, consider using ΔXCO2 and ΔCO2 as axis labels.
References
Brunner, D., Kuhlmann, G., Marshall, J., Clément, V., Fuhrer, O., Broquet, G., Löscher, A., and Meijer, Y.: Accounting for the vertical distribution of emissions in atmospheric CO2 simulations, Atmos. Chem. Phys., 19, 4541–4559, https://doi.org/10.5194/acp-19-4541-2019, 2019.
Citation: https://doi.org/10.5194/egusphere-2023-2103-RC2 - AC1: 'Final author comments', Sieglinde Callewaert, 14 Jun 2024
Status: closed
-
RC1: 'Comment on egusphere-2023-2103', Anonymous Referee #1, 06 Dec 2023
Review Comment on "A WRF-Chem study on the variability of CO2, CH4 and CO concentrations at Xianghe, China supported by ground-based observations and TROPOMI" by Callewaert et al.
The article presents an analysis of the temporal variability of surface concentrations and column abundances of CO2, CH4, and CO at the Xianghe site in China using WRF-Chem. The study spans from September 2018 to September 2019, utilizing in situ and remote sensing measurements. The model's passive tracer option, WRF-GHG, is employed to simulate these measurements. The study reveals key source sectors for each gas and examines the influence of meteorological factors on observed variability. Additionally, the study addresses inconsistencies in CH4 simulations with TROPOMI satellite observations, suggesting potential biases in emission inventories and lateral boundary conditions.
Major Concerns:
The literature review is narrowly focused on recent studies, with minimal references before 2017. This approach overlooks significant earlier research, especially in GHG model-data comparisons in urban settings, as seen in projects like INFLUX, LA Megacity, and European initiatives like Mega-Paris and COBRA.
The study's sensitivity tests concentrate on PBL, surface layer, and radiation schemes, but the rationale for this focus is unclear. Notably, the land surface model variety, crucial for PBL variations, is absent, which is a considerable oversight in the experimental design. See Díaz-Isaac et al., 2018 (ACP).
The method to compensate for WRF-Chem's limitation at 50 hPa for XCO2 calculations involves integrating TCCON a priori profiles. A more consistent approach would be to use the global-modeled CO2 that provides lateral boundary conditions, here CAMS values, above 50 hPa, ensuring a fairer comparison with TROPOMI results. See Butler et al., 2020 (Atmosphere)
The model's performance is somewhat overstated. For example, a mean bias of -1.43 ppm in XCO2 is significant. The high CO2 correlation coefficients are likely influenced by seasonal variation, necessitating detrending for accurate assessment. Additionally, the model's capture of the CO2 diurnal cycle shows notable discrepancies, especially during nighttime. This also reflects the lack of a literature review on the authors’ end.
The authors' conclusions about error sources, particularly the underestimation of XCO2, seem speculative and lack a detailed description of the model configuration, including aspects like data assimilation and restarts. Without the info, it’s difficult for me to judge if the authors’ interpretation makes sense or not.
A Lagrangian approach might be more suitable for tracing back signals to their sources and sinks, as the current Eulerian perspective may not effectively disentangle these signals.
The manuscript suffers from poor organization and numerous language and grammatical errors, making it challenging to follow. For instance, the placement of CO discussion in the CO2 section and the absence of a dedicated section for CO2 biases, despite their significance, are confusing.
Specific Comments:
- Line 67: Expand citations to include key studies like Feng et al., 2016 (ACP), Lauvaux et al., 2016 (JGR-A), etc. See my comments above regarding those urban projects for more references.
- Line 105: Reevaluate the relevance of the Fast et al. (2006) citation.
- Lin e 165-167: Clarify the intended message for better understanding.
- Line 186: Correct the statement about TROPOMI's orbit.
- Line 201: Provide justification for the chosen two-week spin-up period. My experience is that sometime after 20 days, I still see influences from initial conditions.
- Figure 6: Narrow down the wind direction ranges for NW and SW.
- Figure 7: high correlations after diurnal cycle removal are expected in terms of synoptic scale variations.
- Line 333: there is a surface layer under the PBL.
Considering the major concerns and fundamental issues in study design and manuscript presentation, I recommend a comprehensive revision of the manuscript. The authors should address the concerns in detail, improve the organization, and enhance the clarity of the language before the manuscript can be considered for publication.
Citation: https://doi.org/10.5194/egusphere-2023-2103-RC1 -
RC2: 'Comment on egusphere-2023-2103', Anonymous Referee #2, 19 Dec 2023
Review of 'A WRF-Chem study on the variability of CO2, CH4 and CO concentrations at Xianghe, China supported by ground-based observations and TROPOMI'
manuscript by: Callewaert et al., review by Anonymous Reviewer
December 2023
Overview
The authors present an overview of the performance of their WRF-Chem modelling framework, focusing on comparing the model performance against the in situ and column measurements of CO2, CH4 and CO performed at Xianghe, a measurement site in the vicinity of Beijing, China, over a period between September 2018 to September 2019. In order to alleviate the issues with interpreting regional aspects with localized observations, they also use spaceborne observations from TROPOMI instrument where CH4 is discussed.
The authors do not use a typical (Intro - Methods - Results - Discussion - Conclusion) paper structure. After the introduction (Sec. 1), describing the measurement site (Sec. 2) and model framework (Sec. 3), in which authors describe the input data as well as sensitivity studies performed to select the final model configuration. Further, they describe the methodology of comparison against TROPOMI data (concerning only CH4) in Sec. 4, after which they sequentially analyse various aspects of the model-observation comparisons. In Sec. 5, the overall model results and their comparison against observations are discussed, and larger issues are identified. In Sec. 6, the authors discuss primary sector-contributors to the observed signals for each analysed compound. Then, in Sec. 7, the reasons for seasonally-varying bias in CH4 is analysed with particular attention to the emission data, further supported by TROPOMI observations. In Sec. 8 the authors are discussing meteorological causes for FTIR model-data comparison, as well as PBL variability at the site and its influence on detecting local emissions, arriving at possible discrepancies in the local attribution of coal-mine methane fluxes from nearby coal-mining area. Sec. 9 concludes the paper. The manuscript also contains two appendices, Annex A containing additional details on the sensitivity study discussed in Sec. 3, and Annex B containing extra illustrations with supplemental information.
The modelling framework is set up very well, with clear indication of the authors knowledge about practicalities of running such complex system for the purpose of comparisons against the real-world scenarios. The results demonstrate that the selected configuration was well chosen and I commend the authors on their setup. Thanks to that, the study provides much interesting information on the greenhouse gas transport in the regionally- and globally relevant GHG emission region of NE China.
The manuscript, however, suffers from analyses that feel unfocused and too shallow. In multiple places the authors appropriately identify issues that require further study, but in none of the cases (e.g. bias in long-term CH4, potential underestimation of coal-mine CH4 sources) the analysis goes deep enough to determine whether the hypotheses brought up to explain them are adequate.
Here follows a list of some major concerns:
- The unorthodox structure of the text makes it at times difficult to follow. The flow of information seems to reflect way the experiments were performed, describing the 'results > analysis > extra-results-to-test-hypothesis-X > analysis' chain, but this might not be the best choice for the final publication. I strongly recommend that the authors streamline and clarify the text structure.
- In multiple places in the manuscript, the quantitative analysis and/or information is needed and is not given. For example, in L213 the authors write that "a similar pattern [of negative bias] was found when comparing CAMS reanalysis data set with the TCCON data at Xianghe and other sites in that part of the globe", but do not provide neither the size of that bias nor the reference to the study where it was published. Another example, in L404: "We find slightly elevated XCH4 values nearby the coal mines of Tangshan (∼6◦N, 118.4◦E) in both WRF-GHG and TROPOMI maps" -- it is unclear what does "slightly elevated" mean here? Numeric value here would be objective and reproducible. I've listed some of the places where more quantitative approach is needed below (see "Detailed comments").
- Seasonal bias of CH4 is heavily discussed, but the authors do not attempt to quantify how much of this effect is related to background variability, and how much is local. While this doesn't allow for direct data to model comparison, this could still inform on relevance of regional vs local influences.
- Discussion on biases for CO2 is limited to 7 (seven) lines and focuses on the jump of summer of 2019. Looking at the data quality makes me wonder about stability of the measurement system at that time (especially considering concurrent gap in the insitu system - was that related?), but regardless of that, the latter part of the data is ignored, the CO2 discrepancy is still worth investigating.
As it stands, the paper requires major revision, with second review recommended after addressing the issues listed. Considering already large amount of material that is present in the manuscript, it might be worth considering to divide local and regional aspect into separate, focused, more detailed studies.
Below please find, other, more detailed comments.
Detailed comments:
L4: passive tracer is also an option from base WRF. In order to avoid confusing GHG module with that separate module, I suggest: "are produced by model's greenhouse gas module WRF-GHG"
L5: add correlation coeff. also for CH4 for completeness
L9: 'sectors' are usually used for anthropogenic activities, 'biosphere sector' is rarely used (if at all). Consider using 'biosphere fluxes' or 'biospheric activity' / 'NEE (Net Ecosystem Exchange)' instead throughout the text.
L10: 'Residential \& waste' is meant as a clumped sector as defined in this work, but it requires reading the paper to understand its usage in the abstract. Rephrase / clarify.
L32: "also been rising for the last 200 years"
L63: It's sufficient to ignore chemical reactions change if the residence time in the limited area domain is short enough that the change in mole fractions is smaller than the precision of the measurements. It can be shown for CH4, however, assuming exponential decay, over five days it can already decay by over 2 ppb, and depending on the domain size the residence time of air can be even longer -- depends on the circulation. In most of the cases this can be ignored, but I believe this should be addressed in here for both CH4 and CO, because the domain is relatively large. It should also be discussed later in the paper, where bias of CH4 is described, as the effect of omitting the CH4 destruction could have some effect on observed differences.
L78: Please provide more exact location of the site, at the moment this points to a random location when checked on publicly available mapping portals.
L88: I've never heard about calling CO 'Xgas' before. While interesting, I would suggest removing this.
L91: The instrument is installed on a tower, or is there tubing coming from the tower? Please clarify.
L95: typing error in Sentinel-5 ('-nal')
L97-99: "In our study, we use...". Please add appropriate reference to the dataset.
Same place: Also, evaluation of L2 product is discussed but what about evaluation of L3 that is used here?
L101: -0.6 % and -0.39 % translate into roughly 11 and 7 ppb, respectively. I have some reservations against treating such bias as 'small'. It is substantial.
L102: 'demonstrate great' quality of TROPOMI XCH4 data' - see above. To be clear, it would be absolutely unfair to say that TROPOMI provides poor quality data, but it is an extremely challenging measurement and we should be careful and realistic about what can be achieved.
L126: 'choice of flux inventory is likely important for...' -- understatement, consider '...choice of flux inventory is critical for...'
L137-: I recommend removing 3.2 as separate section (together with Table 1). Extract only information necessary for the manuscript (final config - Table 2) and put it in the description of "WRF-GHG modelling system". Move everything else to the appendix, where it fits better at the moment.
L139: Consider relabeling test case E as BASE (or A).
L147: Where the anthropogenic fluxes also distributed vertically? Relevant since the discussed station is located near lots of anthropogenic activity. See: Brunner et al., 2019
L155: Please add info on TCCON measurement frequency for clarity.
L188: Please provide info on the regridding method.
L197: when comparing to TCCON, WRF output for methane was extended. Why not extend also here? At the very least this extension of 40 ppb could be added to the background (or as extra 'offset') to WRF numbers to avoid explaining the shifting of the colour scales when discussing each figure.
L213-215: The 2 ppm bias is brushed over, and the reference to comparison CAMS-TCCON is missing. See also major comment no. 2.
L221: Again comparison CAMS-TCCON mentioned without reference. Numbers should be given, with uncertainty if possible.
L230: In previous section, the authors followed the order CO2 - CH4 - CO when discussing compounds. Now this order is reversed. Would make it easier to navigate if the ordering was consistent.
L231: 'According to WRF-GHG' > 'Based on our results'
L233: 'Energy sources and biomass burning are not important for the observations at Xianghe. Both residential and transportation tracers show larger values in winter, which is in agreement with higher emissions in that period of the year due to colder air temperatures.' - all these statements would use support from numbers (comment relevant throughout the manuscript).
L240: 'The main sectors contributing to 240 the CO2 data at Xianghe' - consider: 'The main sectors contributing to the modelled CO2 variability at Xianghe'
L241: Note that CO2 for VPRM model in WRF relies on modis surface classification and will predict zero fluxes over urban areas, while in fact CO2 fluxes from urban biosphere can still be substantial. Unless VPRM was modified to handle these fluxes, this might have higher than anticipated impact if urban classification dominates in the vicinity of your station (in the model).
L249: '...the year, which is in agreement with the general emissions patterns in China' - add a reference.
L254-L256: Again, numbers needed to support the statements about emission contributions.
L276-280: 'Unfortunately, the source...'. Maybe I misunderstand. The authors suggest that factors for CH4 emissions from agriculture are constant - and yet there is a peak in emissions? That sounds like a bug in the emission preprocessing code. Please clarify.
L286-289: 'In CAMS-GLOB-ANT, the waste sector is the most important one in the Xianghe region...' - numbers needed; also, why does this need support from Fig4, where direct emissions are available?
L299-300: '...ensemble against GOSAT observations by Parker et al. (2020), a general underestimation of the seasonal amplitude in China 300 was found. This would mean an underestimation of the wetland CH4 emissions in summer.' - Needs clearer phrasing. As it stands the sentence is imprecise. Note that the underestimation of amplitude doesn't automatically translate into underestimation of wetland emissions.
L325: 'More specifically, we looked at the daily mean column concentrations above the background for every wind direction...' Daily means from time when observations were available, correct?
L347-L365: While the importance of PBL height is paramount to the correct interpretation of in situ data, this section doesn't bring a lot of insight into the discussion. It could be quite illuminating if expanded.
L352: 'This stable nocturnal layer is quite shallow and...' - how shallow? Numbers needed.
L361: 'Remark that WRF-GHG is very well capable at simulating this diurnal variation of both CO2 and CH4 in situ observations.' - Again, quantitative analysis would be more than welcome. Also: 'Remark that' > 'Note that'
L372: 'enhancement (sum of all WRF-GHG tracers above the background) is smaller' - delete fragment in parentheses (redundant)
L381: 'The highest values overall...' - how high?
L382: '...are coming from the east. To the east are...' > '...are coming from the east, where...'
L384-385: Concluding the cause before proof is presented. Overestimation of coal-mine emissions cannot be argued before next section, where TROPOMI data is used as extra argument. By this point of the text also bias in the night-time PBL height can explain the observed CH4 in situ bias.
L388: Section 'Assessing CH4 emission sources' is very interesting but sadly mostly based on subjective visual analysis of the figures. More quantitative results should be presented to support the conclusions.
L396-L403: Please briefly expand here. I assume this is a mixture of cloud + albedo effects. Is there enough confidence that the issues causing missing TROPOMI data are completely gone in the areas immediately neighbouring the "white spots" on the map in Fig12?
To clarify what I mean: it is clear that the highest discrepancy is present immediately to the east of 114 deg. E, where many of the gridded cells with those high values immediately neighbour the area of missing TROPOMI data. If the filtration procedure used for TROPOMI is not 100 \% sure, then the discrepancy might be an artefact rather than real signal.
I would appreciate if this can be addressed.
L406: Why not add 40 ppb as offset and use same scales? Would be easier to interpret. See comment above.
L421: In some sections results from 9 x 9 km2 are used for comparisons.
L430: '...a small underestimation of -1.43 ppm and -3.03 ppb with respect to TCCON XCO2 and XCH4, respectively.' Please clarify that these are averages. Add uncertainties.
L431: 'slightly overestimated' - I don't think 'slightly' fits here, these numbers are substantial (again, to be clear: I do not mean the results are bad)
L438: 'This difference is likely because of the lack of strong photo-synthetically active vegetation in the neighborhood of Xianghe.' - in the model or physical world? Or both?
L439: 'For CH4...' > 'On average, for CH4,...'
L453-454: 'peak values are found in the early morning around sunrise when local emissions are strongest...' - Please clarify. Not all the compounds have strongest emissions in the morninig (CH4 for example).
Table 3.: Consider reordering columns, having 'This study' as first for easier reader experience.
Figure 7: It appears that the fonts in this figure are smaller, but maybe it's my eyesight. Please check. Also, consider using ΔXCO2 and ΔCO2 as axis labels.
References
Brunner, D., Kuhlmann, G., Marshall, J., Clément, V., Fuhrer, O., Broquet, G., Löscher, A., and Meijer, Y.: Accounting for the vertical distribution of emissions in atmospheric CO2 simulations, Atmos. Chem. Phys., 19, 4541–4559, https://doi.org/10.5194/acp-19-4541-2019, 2019.
Citation: https://doi.org/10.5194/egusphere-2023-2103-RC2 - AC1: 'Final author comments', Sieglinde Callewaert, 14 Jun 2024
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
461 | 227 | 37 | 725 | 31 | 45 |
- HTML: 461
- PDF: 227
- XML: 37
- Total: 725
- BibTeX: 31
- EndNote: 45
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1