General comments

This paper by Laitinen et al. presents two important datasets of long-term atmospheric CO2 and CH4 observations in northern Finland. Although they follow the same general guidelines defined by the WMO for in-situ GHG observations, including the adoption of the same calibration scale, these datasets were produced at the Pallas site using two completely independent observing systems (i.e. different sampling systems, different calibration standards, different data flagging procedures). The manuscript mainly focuses on the comparison of the two datasets (one produced within the GAW/WMO framework and the other within the ICOS-RI) to show the high consistency between them and to briefly discuss the reasons for the small observed deviations.

The longer CO2 and CH4 dataset (i.e. the GAW/WMO one) was also compared with the widely used NOAA MBL data product to put it into a global context.

The topic of comparison and attribution of deviations in contiguous measurement sites is an important and, I would say, emerging issue (see e.g. https://doi.org/10.1088/1748-9326/abe74a, https://doi.org/10.5194/amt-11-1599-2018, https://doi.org/10.5194/amt-16-2399-2023), so I appreciate the appearance of this paper.

Having said that. I recommend publication and my comments can be seen as mostly minor or technical suggestions/recommendations.

Specific comments

Throughout the paper: I think the nomenclature "mixing ratio" should be changed to "mole fraction" as recommended by WMO.

An interesting point is that the GAW and ICOS data sets are produced using two different sampling inlets. I suggest that the author also add a figure or map to Figure 1 showing the different locations of the two inlets.

Line 86. “Due to its remote…representative for unpolluted air”. Is there a reference to quote?

Line 88. For the reader not familiar with ICOS, can you briefly explain the difference between the different classes of ICOS stations (Class 1 vs. Class 2)?

Section 2.2: For the ICOS instrument, you can report the instrumental uncertainty as indicated by the continuous measurement repeatability and the short - long term repeatability. Are these (or similar) indications available for the GAW instrument?

Line 120: How did you obtain the correction factors to correct for residual water vapour interference? Did you do the “droplet test”? How often? This should also be reported for the GAW instrument (line 150).

Section 2.2.3. I agree with the comments by Tonatiuh Guillermo Nuñez Ramire that flask results should be shown, especially the flask vs. ambient air comparison. Or at least the authors should refer to them in some way.

Line 186: "...last and first days of the year". Can you be more precise?

Figure 6/12: Since there is an apparent impact of the implementation of Nafion on the ICOS instrument, have you tried to analyse DCO2 (and DCH4) as a function of ambient water vapour values to better quantify possible impacts? This would also help to better attribute the differences for CH4 highlighted in lines 336 - 368 .

Lines 290 - 292: A "large" spread is also visible for the comparison of ICOS and GAW instruments. As stated by the authors the observed differences are mostly within the WMO compatibility goal but I’m wondering what happen if you subset the data as done for the trend analyses (i.e. only keeping data with standard deviation less than 0.5 ppm). This would in some way minimize the impact of the different inlet locations by catching conditions characterized by low CO2 variability.

Lines 298-305 and 362 - 366: These paragraphs have been added to place the results of the intercomparison between ICOS and the GAW instrument at Pallas in a broader context. It would be interesting to add information from the other intercomparisons carried out in other WCC-EMPA audits.

Line 336-338: How can you say "better" performance? Can it not be "different performance"? Or do you have some quantitative test to show that the water vapour correction on the GAW analyser is more accurate?

Line 388 – 391: “The largest differences were observed between the GAW and WCC systems in both CO2 and CH4, however the largest spread (CI range) in the differences was between ICOS and GAW for CO2 and between GAW and the ICOS Mobile Lab for CH4. This is expected, as the GAW system is the only one measuring from its own inlet and all the other systems are connected to the same inlet”. A comparable spread for CO2 (Table 2: 0.17 ppm) was also observed for WCC and GAW vs. ICOS Mobile Lab.

Line 394: “This is likely due to the filtering of the data based on the wind speed, assuring well mixed air”. I think you should apply this filtering to the audit as well and see if the deviations decrease (see also my previous comment about lines 290 - 292).

Technical comments

Data citation: the authors only provide data availability at the end of the manuscript. I recommend that the analysed dataset can be fully cited in the main manuscript, since doi are available for both GAW and ICOS datasets.

Line 255, but also in the rest of the manuscript: I would suggest that CI can be expressed as e.g. [-8.0, 11.8] ppm.

Line 262: “(7 bottom) -> (Fig. 7 bottom). The same in line 264. In addition, I suggest that Figure 7 can be moved to Figure 6, as it is referenced earlier in the manuscript.

Line 271: "before"

Line 291: I would suggest to be consistent and continue to use "ICOS Mobile Laboratory" or "ICOS Mobile Lab".

I suggest moving Figure 9 and Figure 14 to Supplementary Material.

Sections 4.1 and 4.2: I would split these two sections into two/tree sub-sections as the following topics are considered: 1) Comparison of Pallas measurements with NOAA PBL dataset, 2) Long-term comparison of GAW and ICOS measurements at Pallas, 3) Combined GAW and ICOS audit.

Figures 7 and 11: Please describe in the caption what the error bars represent.

Line 373: “global development” -> “global tendencies”

The article "Long-term observations of atmospheric CO₂ and CH₄ trends and comparison of two measurement systems at Pallas-Sammaltunturi station in Northern Finland" by Laitinen et al. presents the atmospheric observations of the greenhouse gases CO₂ and CH₄ from two measurement programs, GAW and ICOS, conducted at a boreal station in northern Finland. The authors intercompared the measurements and also performed a concurrent audit of WCC-EMPA and the ICOS Mobile Lab. Long-term trends were compared and discussed, including a comparison with the Northern Hemispheric mean derived by NOAA. The results show that the comparability and compatibility of the network, including their audit, are mostly fulfilled. Deviations from the Northern Hemispheric trends and growth rates were also addressed.

Overall, the paper is well-structured and makes an important contribution to the comparability and compatibility of the European ICOS and global GAW networks. This is a valuable contribution to the user community relying on greenhouse gas observation data. However, the readability of the paper could be improved, and further explanations could provide more clarity.

I recommend publication after the following comments are addressed:

Page 1, Line 17:  In the sentence "We also compared the long time series with the marine boundary layer reference values in the Northern Hemisphere", it is unclear to which station or network this refers. I recommend providing additional context or specifying the station or network being referenced.

Page 2, Line 22-25: Referring to “Especially in situ measurements of greenhouse gas mixing ratios are needed for quantifying the long-term trends of the greenhouse gases, as well as annual and interannual variations. They are also crucial for top-down emission estimates using atmospheric inverse models, which aim to optimize fluxes based on measured mixing ratios (Peiro et al. (2022), Crowell et al. (2019))”, it would be helpful to elaborate on why in situ measurements in particular are needed for quantifying long-term trends, annual, and interannual variations in greenhouse gases. Remote sensing technologies have advanced and it is unclear from the text why they cannot also provide sufficient data for these applications. A more detailed explanation of the advantages or necessity of in situ measurements compared to remote sensing would strengthen this argument.

Page 2, Line 39: “Within the GAW network, the station is referred to as Pallas-Sammaltunturi (station id: PAL) and it reports data on CO₂ and CH₄. Meanwhile, under the ICOS network, the station is named Pallas (station id: PAL), and it provides data not only on CO₂ and CH₄ but also on CO and N₂O. This data is also available as GAW data, as ICOS is a contributing network to GAW”: It would be helpful to clarify the term “contributing network” and the relationship between the GAW and ICOS networks. This would make the expected comparability of data from both networks clearer.

Page 3, Line 54: The statement, “To ensure that the station’s measurements are compatible with the WMO/GAW goals, they must be compared against other instruments to ensure the differences are within acceptable limits” implies that direct comparisons with other instruments are necessary. However, round robin comparisons might already be sufficient for this purpose. Could you clarify why direct comparison with other instruments is deemed essential, or whether round robin exercises or other methods could achieve the same objective?

Page 3, line 84: “The mean wind speed is 6.9 m/s.” Please specify the averaging time period over which this mean wind speed was calculated to provide the corresponding standard deviation as an indication of the variability.

Page 5, line 100: The sentence mentions that both ICOS instruments have been tested at the ICOS Atmosphere Thematic Centre (ATC) before being set up at the station. What about the GAW instruments? Could you clarify their validation procedure?

Page 5, line 105: The phrase “the maximum bias tolerable when measuring well-mixed background air” is used in Table 1, but it is unclear what the bias refers to. Could you please specify what is meant by “bias” in this context?

Page 6, Table 1: Table 1 shows “the WMO Compatibility goals for CO₂ and CH₄ measurements.” Could you please clarify the specific requirements for the ICOS network in relation to these goals?

Page 6, Line 120: The sentence states, “Before that time, the air was measured as wet, and corrections to convert to mole fractions in dry air were applied.” It is not clear whether the correction refers only to the dilution factor or if corrections for cross-sensitivities have already been applied. Additionally, it is unclear what the remaining water concentration is after passing through the Nafion dryer and whether any further corrections are applied afterward. Please provide additional information on these points.

Page 6, line 125: “To monitor the quality of the measurements, long-term and short-term target cylinders are measured to identify any drift in the measurements between calibrations.” Could you please provide further information on why two target cylinders (LTT and STT) were used instead of just one?

Page 6, line 130: “The dry mole fractions can be obtained by sufficiently drying the sample […]”. Please specify what is meant by sufficiently.

Page 7, Figure 3: The abbreviations for the cylinders in the schematic flow diagram (STWS, C1, C2, C3, C4) are either not explained or differ from those used in the text (STT, LTT). Please clarify the terms and ensure consistency between the diagram and the text.

Page 8, line 134: The sentence states “[…] used at ICOS, the correction coefficients are determined for each instrument individually […].” Could you clarify why this approach is used instead of applying the standard instrument water correction?

Page 9, line 153: The sentence states, “The calibration standard cylinders used for the GAW analyzers are filled by NOAA, and the STT and LTT cylinders are filled by the FMI.” Could you clarify how the FMI concentration is connected to NOAA, and how the assigned concentrations for the STT and LTT cylinders are obtained? Additionally, could you provide more information about the quality assurance process for GAW, similar to your description of ICOS? For instance, NOAA serves as the central calibration laboratory (CCL) for CO₂ and CH₄ in the GAW framework.

Page 9, line 157: The paragraph “2.2.3 Flask sampling” provides further information on the ICOS sampling strategy, but it is unclear how this information contributes to the analysis in the main paper. Could you clarify its relevance in your analysis or consider to leave it out for better flow?

Page 10, Line 202: The sentence states “In order to account for a potential drift of the travelling standards, they are regularly compared to laboratory standards between the audit campaigns”. It is unclear what is meant by "laboratory standards". Are those provided by the CAL-FCL and how current are the assigned values? Could you please provide further clarification on these points?

Page 11, line 223: “The WCC-Empa analyzer sampled air from a location close to the ICOS Picarro and the ICOS Mobile Laboratory inlets […]”. Could you please quantify what is meant by “close”?

Page 11, line 229:  The sentence states, “The lower limit for the wind speed is 3 m/s during summertime (June-August) and 4 m/s during wintertime.” Could you please clarify why the difference is made between the wind speed limits for summertime and wintertime?

Page 12, line 261:” CO₂ sinks at Pallas are mostly vegetation, and the effect can be seen in the seasonal cycle (7, bottom).” How is this effect distinguished from the seasonal cycle observed in background CO₂ concentrations, such as those in the NOAA marine boundary layer data?

Page 13, Figure 5: In Figure 5, specific years (e.g., 2001) show larger deviations in the CO₂ growth rate at Pallas compared to the MBL growth rate. Could you elaborate on the potential causes of these deviations?

Page 14, line 271: I do not understand why the average hourly difference differs from the average daily difference. If I understand correctly, as described in page 11, line 229 ff, the data was filtered based on hourly averages using wind speed, the standard deviation of the hourly measurements, and a minimum of 60 minutes of data. Since the calculation of the mean is a linear operation, I would expect the averaged deviation between the two instruments to be the same over the same time period, whether calculated from hourly or daily averages (excluding the confidence intervals, which will naturally differ). Is there an additional filtering process applied? Please clarify and provide further details in the corresponding paragraphs.

Page 14, line 282: The statement “Results of the combined ICOS and GAW audit are presented in Fig. 8” refers to the results obtained by the mobile lab. However, it is not clear whether the results are based on the measurements from both instruments (Picarro G2401 and Spectronus) individually or if they represent the average of both. Please clarify to which instruments the numbers refer.

Page 14, line 284: “As the ICOS inlet is in a slightly different location than the GAW inlet […]”. Whilst there is information on the location of each inlet in sections 2.2.1 and 2.2.2, the relative location of both inlets is not provided. Please include this information to give the reader a better understanding of their spatial relationship.

Page 14, line 293: It is discussed that different flow rates might cause the larger spread in the differences observed between the co-located WCC-EMPA and ICOS Mobile Lab measurements. Have you considered correcting for the different flow rates to validate this hypothesis?

Page 14, line 298: Please provide information on the type of reference instrument used at HEI, as well as the instruments at CBW and OPE.

Page 15, line 312: “There is no significant difference between the cold and warm seasons as in CO₂, indicating little influence of the local vegetation to CH₄”. While it is stated that there is no significant difference between the cold and warm seasons for CH₄, background concentrations of methane are generally expected to show some seasonal variation, particularly due to changes in the OH sink. Figure 11 also shows higher mean values in the cold season compared to the warm season. However, it is unclear what the error bars represent and whether they could account for the lack of significant difference.

Page 15, line 314: “A seasonal cycle is visible in CH₄”. This seems contradictory to the previous statement on line 312. Could you clarify how the observed seasonal cycle aligns with the statement that there is no significant difference between the cold and warm seasons.

Page 16, Fig. 7: Could you please provide information on the error bars in the figure? What do they represent?

Page 19, line 326: “However, the increase measured at Pallas is significantly higher than on the average in the northern hemisphere, indicating a strong increase of local and regional emissions”. Do you have any ideas on the potential sources of these emissions?

Page 19, line 335: “As the drying process eliminates most of the moisture from the sample, the variation is reduced”. Please quantify the remaining moisture in the sample in section 2.2.1.

Page 19, Line 336: “The discrepancy could also be caused by better performance of water vapor correction of the GAW analyzer compared to the ICOS analyzer”. Is the procedure for determining the water vapour correction different for the GAW and ICOS analysers, which might explain why the GAW instrument performs better?

Page 20, Figure 10: In Figure 10, specific years (e.g., 2005-2007, 2010, 2011) show larger deviations in the CH4 growth rate at Pallas compared to the MBL growth rate, or even opposite trends. Could you elaborate on the potential causes of these deviations?

Page 21, Line 356: “Furthermore, all the instruments are calibrated using a separate set of calibration standards, which could cause differences in the calibrated values”. Have you considered performing cross-calibrations of the cylinders to confirm this hypothesis?

Page 25, line 391: “[…] and all the other systems are connected to the same inlet.” I understand that the inlet lines are closely located but distinct (e.g., section 3.2). Could you please clarify this point?

Minor changes:

Page 1, Line 13:  For clarity, I recommend changing "World Calibration Centre (WCC)" to "World Calibration Centre (WCC-EMPA)" as there are multiple World Calibration Centres (WCCs) within the GAW program.

Page 2, Line 21:  Please change “Accurate, long-term observations of the atmospheric greenhouse gas […] to “Accurate, long-term observations of atmospheric greenhouse gases […]” to reflect the plural of the gases being studied.

Page 2, Line 22: I recommend changing “[…] in the composition of the atmosphere […]” to “[…]  atmospheric composition […]” for better fluency and conciseness.

Page 2, Line 50: “[…] ICOS aims to capture the entire carbon cycle. This includes atmospheric mixing ratio observations of different greenhouse gases, as well as […]”: For clarity, I would recommend explicitly mentioning the key components of the carbon cycle, such as CO2 and CH4, as not all ICOS-related observations (e.g., SF6) are part of the carbon cycle: “[…] atmospheric mixing ratio observations of CO2 and CH4, as well as […]”

Page 3, line 58: “One of the central facilities of the ICOS ATC is the ICOS Mobile Laboratory, is tasked with this exact purpose: auditing the different atmosphere stations by means of parallel measurements and cross-comparisons.”: The term "central facilities" may cause confusion if it is not the official terminology used by ICOS. To avoid misinterpretation, I suggest rephrasing the sentence as follows: “The ICOS ATC is composed of various components, including the ICOS Mobile Laboratory, which is tasked with this exact purpose: auditing the different atmospheric stations through parallel measurements and cross-comparisons.”

Page 3, Line 83: For clarity, I suggest to rephrase the sentence “The prevailing wind direction atop the Sammaltunturi is in the West - South axis (Fig. 2 (B)), with very little wind coming in from North” to “The prevailing wind direction atop Sammaltunturi is along the west-south axis (Fig. 2B), with very little wind coming from the north.”

Page 5, line 90: I suggest to replace “in sense“ to “in terms”: “During the last 25 years, greenhouse gas instrumentation has undergone substantial improvements in terms of precision, measurement frequency, and user-friendliness."

Page 5, Line 93: Please remove “based” in the sentence: “Later, in January 2009, both instruments were replaced by a single cavity ring-down spectroscopy (CRDS) instrument capable of measuring both species simultaneously”

Page 5, Line 95: The sentence states, “These instruments were producing data for the GAW network, which was later supplemented by a separate CRDS-based instrument producing data for the ICOS network.” To improve clarity, please specify the time frame for "later".

Page 5, line 98: I suggest to use “dry mole fraction” instead of “mole fraction” to improve clarity: “These commercially available CRDS instruments are capable of measuring dry mole fractions of […]”

Page 6, line 113: Please remove  “  in the sentence “[…] directly after each calibration.”

Page 6, line 125: The sentence states, “To monitor the quality of the measurements, a long-term and short-term target cylinders are measured to identify any drift in the measurements.” Please add “between calibrations” at the end of the sentence, as the calibrations will correct for any drifts.

Page 6, line 128: For clarity I would reorder the sentences and add “with varying ambient water content” to: “The water vapor present in the sample air dilutes the mixing ratios of CO₂ and CH₄, as well as broadening the absorption peaks. In order to make the measured mixing ratios comparable between different stations with varying water content, the effect of water vapor in the sample must be removed. The resulting dry mole fraction is the comparable physical quantity to report.”

Page 8, line 135: The sentence states, “For the GAW analyzer at Pallas, the coefficients are determined by the FMI.” This information would be more appropriately placed in the GAW section for better clarity and structure.

Page 9, line 147: Typo. Please correct “ever” to “every”.

Page 11, line 252: The sentence states, “Consistent with the global trend, the CO₂ levels have risen […]”. For clarity, I would recommend adding “at Pallas”: “Consistent with the global trend, the CO₂ levels at Pallas have risen […]”

Page 12, line 254:  Does the “average in the Northern Hemisphere” refer to the average over the entire Northern Hemisphere or specifically the marine Northern Hemisphere? Please clarify.

Page 12, line 255 (and following paragraphs): The format”'95% CI: -8.0 ppm–11.8 ppm” may be misleading, as it could be interpreted differently by some readers. To clarify that the interval ranges from -8.0 ppm to 11.8 ppm, I suggest using a more explicit format, such as “95% CI: [-8.0 ppm, 11.8 ppm]” or “95% CI: (-8.0 ppm to 11.8 ppm)”. This suggestion also applies to the following paragraphs where a similar format is used to describe confidence intervals.

Page 12, Line 256: “[…] with a mean difference of 4.10 ppm uses an additional leading digit. Consider simplifying to “4.1 ppm” for consistency.

Page 13, line 280: Typo in “[…] is 0.01 ppm ppm […]”. Please remove one “ppm”.

Page 16, Line 314: Typo: “(Fig 11” with missing “)”.

Page 16, line 314: Typo: Change “Amplitude of the diurnal cycle” to “The amplitude […]”

Page 19, line 345: Please change “When the data is filtered […]” to “When the data are filtered […]”

Page 26, line 410: Typo. Change “WDCGH” to “WDCGG”.

The manuscript Laitinen et al. presents experimental evidence that allows to assess the compatibility of two independent measurement programmes at the Pallas measurement site. The provision of longterm quality control (QC) data giving evidence of the data quality of observations is crucial to evaluate the consistency of the global observational data set. The information the authors have compiled for their study is exemplary. The two independent set-ups have each independent methodological choices concerning aspects like drying, calibration approaches or the frequency of QC measurements. The offsets that are visible in the CO2 and CH4 comparisons document a very satisfactory performance. In additition, both programmes maintain each their proper external auditing entitities. The core data set of this manuscript is the result of a coordinated auditing campaign to have four parallel measurements. This data set provides a great opportunity to learn about causes for biases or methodological weaknesses. So first of all I highly welcome the publication of such a study. On the other hand I feel that currently the manuscript does not evaluate the data set sufficiently. It has drawn just few conclusions without strong ambitions to use the findings to learn about which methodological approaches are more successful than others. The manuscript would be of much more interest if it could give guidance what limitations might be cause for measurement offsets. Therefore, I would strongly encourage the authors to further explore the data available to them as suggested below.

General comments

• The four parallel measurements done during the audit campaign have very clear differences in their calibration approaches (no. of calibration points between 2 and 9; frequency between daily and every 3 months). They all also have there own internal quality control set-up (mostly using target standard gas analysis to monitor the success of the efforts to achieve maximum accuracy). Those internal QC records should contain information on what internal reproducibility the two observational programmes (GAW and ICOS) but also the two audit programmes (by the ICOS mobile laboratory and GAW WCC) deliver. From my point of view, those target data records should also be included into the manuscript and a thorough data uncertainty assessment for the individual measurement programmes be made (for the continuous station measurements as well as for the auditing programmes). An estimate of the expected measurement compatibility between the various systems based on such an assessment is needed to judge if the observed comparison mean offsets and their spread presented in Table 2 point to additional systematic analytical limitations.
• The focus of the manuscript could be sharpened. There is general site descriptions that have already been published (in a reference that is cited) and are not essential to the understanding of the main topic of the study.
• Throughout the document it remains a challenge to understand at which points different names for the site are used as synonyms and where those specific names cover a different meaning. Please try to harmonize and avoid terms that are used as synonyms.

Specific comments

• l. 25: It is not clear why exactly those two references have been selected for the use of observational data for atmospheric inverse models and the top-down emission estimates.
• l. 30: The term supersite does not have a clear definition. Either define it or omit it.
• l. 41: “This data is also available as GAW data” is a bit confusing. Consider rephrasing as:”This data is also submitted to the WMO World Data Centre for Greenhouse Gases...”
• l. 54: The following sentence does not fully meet the meaning of WMO Expert Group’s recommendations: “To ensure that the station’s measurements are compatible with the WMO/GAW goals, they must be compared against other instruments to ensure the differences are within acceptable limits.” I suggest to rephrase like “Assessing the compliance to the WMO network compatibility goals requires comparison of station measurements with other laboratory’s measurements“. Please add Andrews et al (www.atmos-meas-tech.net/7/647/2014/) as reference.
• l. 111: “Pallas was labelled as an ICOS Class 1 atmosphere station”: The meaning of “labelled as an ICOS class 1 station” ought to be explained and doi.org/10.5194/amt-14-89-2021 should be provided as reference.
• Figure 4a: The caption should indicate that LTT only is displayed. The trend in an increasing DCO2 for D348367 is substantial with respect to the mean bias numbers in Table 2. A discussion of this should be offered some part in the manuscript. As the LTT are analysed only every 15 days after the calibration the STT record would be of greater interest to compare it with the observed measurement system differences.
• l. 134-136: Are the water vapour correction coefficients assumed to be constant over time (i.e. established just once) or are those re-established according to a certain schedule?
• l. 145: The GAW Picarro is calibrated every 2.5-3 months which assumes response stability throughout that period. When was the calibration done during the auditing campaign?
• l. 154: “STT and LTT cylinders..” A similar figure as Figure 4 displaying the GAW programme’s analyzer QC record should be presented.
• l. 200 ff: Same comment for the Mobile Laboratory measurements: the target measurement results should be presented similarly to Fig. 4.
• l. 213: What are the results of the assessment the validity of the internal water vapor correction coefficient?
• l. 219: The WCC calibration approach is not fully clear: the description states that there is a two point calibration performed (zero plus one atmospheric reference standard). The function of the working standard to account for “drift” (analyzer response drift or CO₂ drift?) is unclear. Is the working standard effectively being used as a second calibration standard?
• l. 221: Same comment as above for the WCC audit measurements: the target measurement results should be presented similarly to Fig. 4.
• Figure 5 caption: “the MBL data” and “the NOAA MBL data” refer to the same data set. Rather use only one term – I think “the NOAA mean MBL data” describes it best.
• l. 275-276: I cannot judge if a change from 94.1 % to 95.3 % is significant. The authors should consider adding a figure displaying the inter-instrument bias vs. the measured water content to reveal if there is any influence of the different drying and water vapour correction functions. Likewise a figure displaying the bias dependence on the mole fraction would be appreciated that might give an indication of a potential calibration offset contributing to the offset.
• l. 279: For readability please do not abbreviate “growth rates” (“GRs”).
• Figure 6 / Figure 12: Please state explicitly which of the two data is subtracted from which (ICOS-GAW or GAW-ICOS) either in the figure caption or in the y-axis label.
• Figure 8: In the presentation of three records in one graph the visibility of the blue data points is too limited. Three rows of graphs should be considered (e.g. grouping GAW-ICOS and WCC-MobLab / ICOS-WCC and ICOS-MobLab / GAW-WCC and GAW-Moblab.
• I do not see the additional value of Fig. 9 compared to Table 2. I would see additional information, though, if the offset between the measurement result pairs would be depicted vs. the mole fraction. The same applies to Fig. 14 and Table 3.
• l. 356: The authors reflect on the source of the biases and suggest calibration being one cause. It is unclear why this is written in a bit speculative manner as the audit of the Mobile Laboratory includes a cross-comparison of the reference gases by the station instrument and the Mobile Laboratory (see l. 195). The results of this is calibration gas assignment bias is not presented - why? Were the reference standards also exchanged between GAW, WCC and Mobile Laboratory? This is data that is necessary information and should be provided in an additional table (or in a supplement).
• l. 383-384: “In the CO2 data the effect [of the Nafion dryer] is less clear, with the difference before adding the Nafion being 0.02 ppm and -0.02 ppm afterward”. Fig. 6a clearly shows that a seasonal cycle of ca 0.1 ppm in the offset was much reduced after using the Nafion on the ICOS system. This has been described for CH4 (l. 332) but should also be mentioned for CO2. It is clear evidence that a problem has been remedied even if the absolute number of the offset has not changed.
• l. 389-391: “the largest spread (CI range) in the differences was between ICOS and GAW for CO2 and between GAW and the ICOS Mobile Lab for CH4. This is expected, as the GAW system is the only one measuring from its own inlet”. This conclusion is not fully convincing as long as the insignificance of other factors is not proven. Factors that might result in GAW measurements being slightly different could be for instance: 1) GAW measurements are also the only ones that are not made with dry air. So the application of the water vapour correction makes GAW different to all others. 2) Respective assignment errors (or instable CO2) of reference standards might result in mole fraction dependent offsets. 3) The reproducibility of the GAW system might be inferior due to less frequent calibrations. This could be easily checked by an evaluation of the target records.
• l. 393-394: “Compared to the differences between the ICOS and GAW analyzers over the whole period, the CO2 difference during the audit is slightly higher.” It seems more reasonable to compare the audit mean difference to the mean difference since December 2020 when the Nafion has been added to the ICOS system. Else this aspects needs to be mentioned in the discussion here.
• l. 401-402: “The audit period also indicates that the ICOS system is performing slightly better”. Phrase like: “The better measurement agreement with the two auditing units suggest a better performance of the ICOS system”. It would be very useful to add if this corresponded to a superior reproducibility estimate for the ICOS measurements based on the record of its short term target compared to the respective target records of the GAW system.
• l. 403: “but the effect of theNafion dryer on the differences between the two systems is still unclear”. As pointed out above (comment to l. 383) this statement is not justified. The Nafion dryer removed a stronger systematic seasonality in the offset. There might be a smaller seasonality remaining after December 2020 with the phase having shifted but it is difficult to capture this just visually from Fig. 6a. This could point to a remaining smaller humidity related offset now caused by the wet air analysis of the GAW system and also a non-perfect water vapour correction. This is something the authors should check.