The manuscript addresses the very current problem of CH4 emissions from the thawing sub-arctic permafrost regions. The problem is analyzed with novel methods and unique data, and its contribution to the current state of knowledge in biogeosciences is significant. The overall scientific quality of the manuscript is high. It is in general clear and well organized and should be published in the BG with only minor modifications. Below are a few specific comments that should be considered to improve finale version of the manuscript.

Specific comments.

The issue of water table level:

1) in Figure 2 WTL (water depth) is expressed in m a.s.l. It should be presented in relation to the ground level. In line 164 authors report that the EC system collects data from a height of 2.2 m a.g.l., which means that the ground level is somehow determined. The WTL should be related to this level to provide information if the WTL is above ground level or at a certain depth in the ground.

2) Figure 2 shows a jump in WTL between 2014 and the next two years by about 1 m for the western sector and several dozen cm (> 50 cm) for the eastern sector. It is surprising that the formation of such a thick peat aeration layer has not affected the CH4 flux.

3) In the case of such a WTL jump, its distribution should be bimodal (why in Figure S2 it is clearly visible for the eastern sector, and not visible for the western sector?) and the correlation between CH4 flux and WTL should be checked separately for each year.

4) Even for the eastern sector, for which WTL is considered representative (ln. 380), the soil moisture in summer 2014 was lower than in the following two summers, while WTL was much higher in 2014 (Figure 2). How is it possible?

Ln. 397: Please mark the contribution level (e.g. 80%, 50%) on at least a few selected lines in Fig.3.

Ln. 405: As seen in Figure S4, no diel cycle was observed – in my opinion Fig. S4 shows weak diel cycle, with 10-20% differences between nighttime and noon fluxes. Moreover, the potential diel cycle should be exanimated separately in the seasons. In summer, changes in solar radiation can cause a significant diel cycle of surface temperature (temperature impulse), which may affect methanogenesis, while in winter there is no such forcing.

Ln. 410-414: Information in Tab. 3 are a bit misleading. For example, for 2016 the coverage by a good data is 99% (sum for eastern and western sector), which seems quite unrealistic for EC method. In fact, the assumption that 10 good data over a full 24 hours is sufficient to calculate daily value (ln. 408) is a kind of gap-filling method and means that up to 58% (14/24) of data might be gap-filled by mean daily value.

Ln. 430-431: The peak season of the CH4 emission was defined as two weeks forward and backward from the day with the maximum daily emission in a given year– it is possible that a single high emission does not occur in the peak of the season, so why not use a 14-day moving average and next use the maximum of this function as the peak emission?

Ln. 436: Wintertime average emissions were 24 mg-CHm-2 d-1 for the eastern sector and 16 mg-CH4 m-2 d-1 for the western sector – but when we compare these values with Fig. 4, the 24 mg-CH4 m^-2 d^-1 level is clearly above the most of green tringles for wintertime (blue areas). Similarly, the 16 mg-CH4 m^-2 d^-1 level is above black dots at winter. It means that the quoted average values for the eastern and western sectors are amplified by gap-filled values, i.e., the gap-filled values on average are significantly higher than the measured once. Is that correct? Any reflection on this effect?

Ln. 450 and Table 7: Controlling factors were examined before and after temperature normalization (Table 7) – please be more specific about which normalization is concerned. The normalization described in lines 402-405 refers to diel cycle. Of course, it doesn't make sense to correlate such normalized values with other (non-normalized) variables. At this point, the authors are likely to use a different normalization (exponential function of temperature), the same as Rinne et al. (2018). However, this only becomes clear on line 559.

Ln.521-523: …the fen has the highest percentage of carbon emitted as CH4. The eastern and the western sectors emitted less of the carbon as CH4. – these sentences suggests that both ecosystems emit carbon also as CO2, while in the annual scale, they absorb CO2 (and total carbon).

Ln. 576: … small variation, without strong extreme conditions, in the WTL – can WTL changes in >0.5m (differences between 2014 and next two years) be considered small?

Ln. 619: editorial: “als oin” – also in;

Ln. 699-700:The seasonal cycles were furthermore characterized by a gentle increase in spring and a more rapid decrease in fall –  in my opinion, Figure 4 does not confirm this, or even suggest something quite the opposite.

Ln.: 706-707: the temperature at different depths seemed to control the CH4 fluxes for the two analyzed mire sectors – can the temperature profile measured at one location east of the tower be representative of the entire eastern (patched) sector? Is the temperature at the set depth the same for the entire eastern sector? The same for western sector. So the conclusion seems a bit too firm.

Łakomiec et al. presented a manuscript discussing CH4 fluxes from a sub-arctic permafrost mire. Experimental conception, data acquisition, and statistical analysis were conducted following current state-of-research recommendations. The authors obtained a data set of high quality and present a well done analysis with novel insights and aspects.

The manuscript is within the scope of "Biogeosciences" (BG) and improves the scientific knowledge in the context of thawing permafrost CH4 emissions. It is well written and structured. The manuscript is already of good quality, thus I just add a few remarks to consider below. I recommend the final manuscript for publication in BG after minor corrections.

Generally: in l. 58 you first introduce methane (CH4), but later you switch randomly between "CH4" and "methane" in the text. To provide consistency, please use always "CH4" in the text after first mentioning it in l. 58. Please check the same also for other abbreviations you introduced.

l. 99: This sentence might be difficult to understand. I recommend to divide it into two sentences for each first and second area.

l. 124 and others: there is a space character missing between value and unit. Please write "0 °C" instead of "0°C" and check also the other parts of the manuscript regarding that.

l. 126 and later: in many parts of the manuscript you give both air and peat temperatures with 2 decimal places. Is this really justified, considering the uncertainties of the sensors?

l. 153: The intake tube of the LGR analyzer had a length of almost 30 metres,  which is a relatively long tubing. Did you carefully check whether the measured CH4 signal was dampened due to the flow characteristics of the sampling tube? How does the co-spectra look like? Are there any signs for a dampening effect in the high-frequency range, and if possible, did you apply a suitable correction? Please provide a short statement on that in your manuscript.

l. 160: The LI-7200 ist an enclosed path analyzer. Additionally, the official notation of the manufacturer is "LI-COR". Please write it consistent in the manuscript.

In l. 257, 316 you write "global radiation", in l. 465, 467, 565 you name it "shortwave radiation". I recommend to write "shortwave incoming radiation" generally in the entire manuscript.

l. 436: You report an average emission of 24 mg-CH4 m-2 d-1 for the eastern sector in wintertime, which is in accordance with Table 6. However, refering to Fig. 4, wintertime emissions at the eastern sector seem to be substantially lower than 24 mg-CH4 m-2 d-1. Are the mean values, maybe, in Table 5 and 6 the gap-filled ones? a) If yes, please clarify in the table descriptions and in l. 433, l. 437. b) If yes, why does the gap-filled value seem to be substantially higher than the the non-gap-filled data? c) If no, what is the reason for this discrepancy?

l. 637, "Method...": is there a word missing at the beginning of the sentence?

l.699f: You conclude a "gentle increase" of CH4 fluxes in spring, and a "more rapid decrease in fall". Figure 4 somewhat differs to that finding: I see no difference in increase / decrease ratio for 2016, while for 2014 and 2015 there seems to be a more rapid increase in spring, followed by a less rapid decrease in fall? Am I wrong?

Fig. 1: change m/s => m s-1.

Fig. 2: The water table level (WTL) is given in metres above sea level. For what reason? I guess it could be more intuitive to give relative values referencing to the ground level. In l. 164 you introduced a ground level (a.g.l.) baseline - maybe you could do that also for WTL?

Fig. 3, upper panel: To avoid misunderstandings, I recommend to add the information that the red contour lines correspond to the 10 % to 90 % contributions of the flux.

Fig. 4: Shouldn't you change "temp" to "surface peat temperature" in the x-axis label? Additionally, you never use the term "breakout week" in neither text nor the figure itself. Please clarify the figure and/or figure description.

Table 1: Tables are always harder to understand than figures, especially when comparing different years and footprints. I suggest to replace table 1 by a figure with the DOY (1 - 366) on the x-axis, and the years (2014, 2015, 2016) on the y-axis. You then draw the unfrozen periods into the plot, using the color codes (grey = western, green = eastern) of Fig. 4. This makes it much simpler to compare different years and footprints.

Table 7: I guess, "normalization" means that you used temperature-based normalization approach following Rinne et al. (2018) as stated in l.559? This makes sense, however it remains somewhat unclear until reading l. 559 later. Please clarify in the table description and in l. 450 to avoid misunderstandings.

Table 9: Please write the unit "g-C m-2 yr-1" to be consistent with Table 8 and Table 10.

Table 10: The annual emissions of the type "thawing wet surface" is 11 +/- 2 g-CH4 m-2 yr-1? Do you mean +/- 2.0? Additionally: why is the cell in first column, fifth row, which refers to 28.3 +/- 1.7 g CH4 m-2 yr-1 empty? It is also "thawed fen", I guess?

General comments

The manuscript of Lakomiec et al. presents very valuable data and analyses of land-atmosphere fluxes of methane at a heterogeneous permafrost-affected mire. The study is particularly relevant because it allows conclusions about the effects of permafrost thaw-induced landscape transformation on methane emissions. The question of how methane fluxes of permafrost regions will change in the warming Arctic is of great importance for assessing the climate-carbon cycle feedbacks in our Earth system. It is very good that the study is able to present three complete annual methane balances including winter seasons; such data is still very scarce for the Arctic – but urgently needed. The topic fits well in the scope of Biogeosciences.

The study approach in general is sound, and the applied methods are state of the art. However, I have several questions and requests for clarifications, which I list in the list of specific comments.

I think that the introduction and discussion are a bit narrow when it comes to referencing related work. The manuscript mostly refers to studies previously performed at Stordalen mire, however, there is not much discussion about related work around the Arctic.

The manuscript is already well structured and mostly well written. However, I have some comments on some text parts that appear to have been missed by the internal review. These are given in the list of technical comments.

I recommend to accept the scientifically interesting and relevant manuscript of Lakomiec et al. for publication in Biogeociences after major revisions.

Specific comments

l. 62: Do you refer here to surface-near air temperature or soil temperature or permafrost temperature?

l. 104-109: Pleae refer here in the introduction to previous studies that have conducted similar comparisons, e.g., Hommeltenberg et al. (2014), Rößger et al. (2019), Kim et al. (2020). Rößger et al. (2020) investigated methane fluxes from a heterogenous tundra ecosystem; thus this article would be quite appropriate for comparison also in other regards.

l. 123: Specify if surface-near air temperature is meant.

l. 153: The tube length is very long. Can you assure that flow was turbulent throughout the tube? What ist he high-frequency attenuation oft the fluxes due to the tube transport effects?

l. 168-181: Please describe better the locations of the ancillary soil measurements. In a heterogenous mire landscape peat temperature can have large spatial variability. Particularly of interest is what site you choose as being representative for the heterogeneous eastern area composed of drier palsas and thawed wetter sites.

l. 202-203: I do not understand this approach of removing flux values when two consecutive data points originated from different wind direction sectors? Which flux values where then removed? Why was this done?

l. 213-215: Have you tried to model also 30 min fluxes? Why not modelling the 30 min flux data (as Rößger et al. (2019))?

l. 222-223: Please describe in more detail how the 30 min data were „aggregated“ to annual footprint climatologies.

l. 235-237: How is the weighting calculation exactly done? Which quantity of the „climatology“ was used for weighting the contribution of a mosaic pixel?

l. 245-246: The statement that “...methane emissions … do not show diel cycle” is too bold. Figure S4 shows that there is systematic diurnal variability – even for whole-year data. Indeed, it would be good to analyse diurnal variability month by month.

l. 274: Rößger et al. (2019) applied ANN for a heterogenous tundra; the paper might be interesting for comparison.

l. 323-333: I think that the equations (1) and (2) are only valid under rather strong assumptions that should be clearly stated. In my view equation (1) and equation (2) can be considered valid for the 30-min periods for which the footprint contributions of the two contrasting landcover types f_p and f_t are estimated. However, using the same form of the equation for annual averages is only valid if the time series of the footprint contributions fp and ft, respectively, are uncorrelated with the temporal development of the emission factors E_p and E_t, respectively. If they would show some correlation, the average of the product f*E would equal the product of the averages of f and E, respectively, plus the covariance of f and E. Therefore, one maybe important assumption is that f and E of the respective landcover types are uncorrelated. Other important assumptions are that the average methane fluxes of the palsa sites in the eastern area and in the western area are equal and that the average methane fluxes of the thawed sites in the eastern area and in the western area are equal. It would be good if this assumption could be backed by more comprehensive description of microtopography, hydrology and vegetation of palsa and thawed sites of the western and eastern areas, respectively.

l. 360-361: Sentence too vague: Where and when the western sector is colder? Which depth? Time scale?

l. 369, Figure 2: The lowest panel does not show water depth but water table height. However, it would be more suitable to show water table depth or height referenced to a reference point at the ground surface in the investigated mire. The jump in water table height between the summer of 2014 and the summer of 2015 appears unrealistic. Please check the water level times series for biases and inhomogeneities in the time series.

l. 387-390: Please write more specific, e.g. “…on average over all three years more than 90% to the fluxes measured at the eddy covariance tower.”

l.392, Caption figure 3: Please describe more precise what is shown in the bottom panel. How are these average contributions of contrasting landcover types calculated. Generally, I would prefer another diagram type that allows evaluation of the variability of footprint contributions of the two landcover types.

l. 419-420: How did you deal with the autocorrelation in the time series? Serial dependence of data points could lead to biased results of the Wilcoxon test.

l. 433-434: How were differences tested? See comment above.

l. 455: Unclear what “breakout week” means.

l. 565-572: I find the discussion of the explanatory strength of incoming shortwave radiation confusing. The GLM parameters in Table S2 for the explanatory variable incoming shortwave radiation are all negative, indicating that methane emissions were lowered under high incoming shortwave radiation. Thus, the GLM results do not suggest strong relations between shortwave radiation, photosynthesis, substrate supply and CH4 production. Or was the sign convention for incoming radiation different than I assumed?

l. 724-725: The hysteresis-like behaviour can be also explained by the phase lags between different temperatures in air, ground surface and different soil depths.

Figure S1: Specify in the caption if soil or air temperature is shown. At which height in the atmosphere?

Technical comments

l. 62: I recommend to always place a space between numerical value and unit, also for the unit „°C“. Please see „The International System of Units“ (BIPM, 2019, section 5.4.3).

l. 99: Please correct awkward sentence: „…while the second one is thawing, wetter areas.“ (singular-plural)

l. 102: Insert „the“ to „the palsa plateau“.

l. 123: Comma  before „and“ (independent clause)

l. 131: I suggest to insert a comma and use „which“ instead of „that“ here (unrestrictive clause).

l. 150: Correct „closed-path“

l. 160: Insert „a“ before  „LI-7200..“

l. 164: I do not think that „a.g.l.“ is a standard abbreviation directly clear for everybody.

l. 215: Remove „be“.

l. 219: Remove „a“.

l. 229-230: I suggest wrinting „50 cm x 50 cm“ (equal 2500 cm^2)

l. 236: Correct „climatology“.

l. 295: Insert comma before „and“.

l. 302: Hyphenate “gap-filled“.

l. 314: Hyphenate „gap-filling“.

l. 360: Remove comma before “was”.

l. 361: Remove commas before “between” and “at”.

l. 413: Correct: “after averaging”

l. 637: Remove comma before “because”

l. 637: Correct “have”.

l. 637-638: Improve awkward sentence.

l. 639: Remove “the”.

L. 675: Insert comma before “which”.

l. 696: Insert comma before “but”

L. 714: Correct “the gap distribution”.

l. 719: Insert comma before “further”

l. 720: Hyphenate “low-emissions wetlands”

References:

Rößger, N., Wille, C., Veh, G., Boike, J., & Kutzbach, L. (2019). Scaling and balancing methane fluxes in a heterogeneous tundra ecosystem of the Lena River Delta. Agricultural and Forest Meteorology, 266, 243-255.

Kim, Yeonuk, Mark S. Johnson, Sara H. Knox, T. Andrew Black, Higo J. Dalmagro, Minseok Kang, Joon Kim, and Dennis Baldocchi. "Gap‐filling approaches for eddy covariance methane fluxes: A comparison of three machine learning algorithms and a traditional method with principal component analysis." Global change biology 26, no. 3 (2020): 1499-1518.

Hommeltenberg, J., Mauder, M., Drösler, M., Heidbach, K., Werle, P., & Schmid, H. P. (2014). Ecosystem scale methane fluxes in a natural temperate bog-pine forest in southern Germany. Agricultural and Forest Meteorology, 198, 273-284.