Comment on "The effects of peat thickness and water table depth on CO2 and N2O emissions from agricultural peatlands - a process-based modelling approach" by Kajasilta et al., Biogeosciences

General comments

In this manuscript, authors applied LDNDC to simulate C and N dynamics in a single peatland field site following traditional grass-intensive crop rotation, and examined the change of net ecosystem carbon balance (NECB) and N2O emission under different prescribed peat thickness and water table depth. Although only one site is studied, this study also evaluated the capability of LDNDC to simulate C/N dynamics of peatland, which can be considered as a novel work. Though most works are solid, some key information is missing and shall be presented to fortify major findings. In addition, the description of the N cycle in LDNDC and its calibration is missing. 

Specific comments:

1. Authors calculate soil water movement but enforce water table depth to the model at the same time. Based on my understanding, soil below the water table shall be saturated and only the soil moisture calculation above the water table is meaningful. Authors shall provide more information to justify your model treatment and explain how the boundary conditions and calculations have been changed in LDNDC after overlapping prescribed water tables.
2. Since the EC tower is installed in the middle of the field, authors should provide more description to justify how they can obtain EC data only representing block 5 and 6. How authors conducted footprint analysis and what criteria used to filter EC observation contributed from block 1-4?
3. Authors mentioned air temperature measurement as an environmental variable, but there's no analysis related to air temperature in this article. Please explain why you need air temperature measurement.
4. Model description shall be informative. Authors at least shall provide enough information for readers to understand model structure and parameters. I would suggest authors to explain: 1) Major model processes, which authors may have already mentioned a bit, but need to consolidate and rephrase. 2) Model structure: How plant type is represented? How many organic C, organic N and mineral N pools are defined in different modules? How many layers of soil and the node depth of each soil layer? 3) Add description in table S1 to explain key parameters used in these key processes and 4) explain the reason why choosing parameters in Table S1 for calibration. I also have some minor suggestions/questions related to model description, so feel free to rephrase this part if necessary.
5. Though evaluation is performed, no calibration on the N cycle, especially mineral N, is done in this study. I understand the mineral N observation is hard to obtain, but authors shall explicitly explain why the calibration cannot be done?
6. Conclusions from this study are based on results from one model. Though the parameter uncertainty related to respiration and evapotranspiration is explored, uncertainties from other parameters and from model structure are not considered so it shall be discussed at least, e.g., certain simplification of model representation can cause under-/overestimation of NECB/N2O emission?

Technical correction:

Line 79: "chemical soil properties" shall be "soil chemical properties"

Line 95: If both blocks 5up and 6up have similar peat layer thickness to block 5 and 6, I suggest changing the color of block 5up and 6up in Figure 1.

Line 125: "following (Vekuri et al., 2025)" shall be "following Vekuri et al. (2025), "?

Line 128: "gaps of two hours" shall be "gaps within two hours"?

Line 130: What is the size of the window for your moving average?

Line 131: The cited paper Vira et al. (2025) does not describe how you process measured ET?

Line 148: "LAI was evaluated using the methods described in" Do you have in-situ measurements of LAI you did evaluate against S2 products?

Line 154: "layer-wise representation" How did you define the soil layer depth, node depth and top/bottom boundary conditions for your soil water movement/dynamics module to simulate WTD dynamics?

Line 160: It seems like "Plamox" is an abbreviation. Can you tell readers the full name?

Line 162: What does "CanopyECM" mean? 

Line 163: It seems like "MeTrx" is a module representing soil biogeochemical processes? Please provide the full name.

Line 169: "as soils may exhibit substantial annual losses" shall be "as soil organic carbon may exhibit substantial annual losses"

Line 172: "The module handles the dynamics of water within the soil profile" can be simplified "The module handles the soil water movement, and accounts for the amount of precipitation intercepted by foliage, infiltration, ..."

Line 173: "possible changes in snow cover and ice content in the soil" This requires air temperature and soil temperature profile. How does your model obtain this information? By calculation or use prescribed temperature profile?

Line 174: "Evapotranspiration follows the potential evapotranspiration and is limited either by the amount of surface water or remaining potential evapotranspiration, whichever is reached first." Actual evapotranspiration (AET) is different from potential evapotranspiration (PET), which can only be equivalent if you are calculating ET from a submerged surface. In your case however, you have drained peatland blocks during a certain time period, so they are different. Please clarify.

Line 180: "The default value of spinupdeltac is 0, corresponding to the original equilibrium assumption, but it can now be set to reflect user-defined annual changes." What is the value of this "user-defined annual changes" in your study?

Line 186: "we increased the decomposition rates" What method did you use to determine the increment of the decomposition rate?

Line 187: "model initialises most of the carbon and nitrogen in two pools that represent young and old organic matter," Should this be labile vs. recalcitrant? Also, how did you determine their relative proportion without observation?

Line 189: "resulted in spurious blockwise variability in soil respiration" can you describe how spurious it is? For example, respiration in a certain block becomes extremely high or low? Also, I am confused that authors change parameter values but claim "no differences in the parameterisation between the blocks", then why is there a blockwise variability in soil respiration?

Line 199: "we adjusted parameters handling the photosynthesis activity (H2OREF_A) and stomata closing (H2OREF_GS)" Can you show me the actual name of these parameters? If readers cannot get detailed information about these parameters, the rest of the description on how you adjust these parameters to "avoid underestimating the GPP" does not make sense.

Line 202: "The drought periods were not seen in EC measurements" If you don't have observation based evidence, how can you be sure that low GPP is problematic?

Line 226 - 236: This shall be part of calibration not initialization.

Line 254: How did you treat tillage in LDNDC? For example, change the soil porosity and the base decomposition rate of C/N pools after tillage? 

Line 256: The rest 1% of biomass is removed from the site or is kept alive?

Line 267: Do you also calculate CH4? If not, why is the ambient CH4 level mentioned in this study?

Line 268: Remove "the" from "similar to the those measured"

Line 314: Is the length of the vector equivalent to the length of the observed time series?

Line 336: Describe how you define and calculate N2O balances?

Line 348: "Soil measurements during the winter time were unreliable and should not be emphasized due to the measurement problems when the soil is frozen or close to that point." This is not your result. Shall move to section 2.

Figure 7: It seems like some of the modeled values are repeating, reflected by the almost same modeled values against completely different values from chamber measurement. Can you explain this phenomenon?

Line 463: "However, the simulated respiration differed between shallow and deep peat fields, which led the model to underestimate the ecosystem respiration for the shallow peat fields and overestimate it for deep peat fields." can be simplified to "However, the model underestimates the ecosystem respiration for the shallow peat fields and overestimates it for deep peat fields." Also, you can merge this one with the previous one paragraph.

Line 466: "the other years were either under or overestimated by up to a factor of two". Can you provide a certain evaluation about how this over-/underestimation of N2O can bring impact to your major conclusion of N2O balances under different WTD scenarios?

Line 482: "supports the hypothesis that increasing the water table can suppress nitrification and subsequently reduce the availability of nitrate for denitrification" Did LDNDC produce the similar phenomenon? I believe most of the popular models have these two fluxes as output so I would recommend authors to check model outputs.

Line 500: "This sensitivity analysis showed that our findings regarding the CO2 emissions were more robust to parametrisation than the absolute CO2 emissions." Can you rephrase this sentence? It is a bit tough for me to interpret?

Line 512: Incomplete sentence. "The scenario with a 50 cm average WTD required on average a 31 cm higher water table for deep peat blocks and a 44 cm higher water table in shallow peat blocks"

Line 519: "below the organic soil horizon in the shallow peat blocks". What is the depth of the O horizon for shallow and deep peat blocks, respectively? 

Table S1. Add explanation of all parameters you have calibrated in table S1.

This is a well-constructed and well-written paper which adapts an existing model (LDNDC) for agricultural peatlands based on detailed observation data from a Finnish experimental and monitoring study. The results showing clear CO2 and N2O emissions mitigation potential of both partial and full water table raising are broadly consistent with existing empirical data but provide new mechanistic insights. The study merits publication, subject to revisions based on the comments below.

My main concern with the paper relates to the use of a ‘traditional’ multiple pool turnover model, developed for mineral soils, to represent carbon cycling in a peatland system where organic matter protection via waterlogging is likely to be more important (at least pre-drainage) than the properties of the organic matter itself (represented as ‘young’ and ‘old’ carbon’). Given that the peat now appears to have a high mineral content, organic matter protection by mineral surfaces may also be important. See e.g. Schmidt et al., Nature (2011) on why protection mechanisms are more important than intrinsic properties of organic matter, and why pool-turnover based models may therefore fail to represent processes correctly.

The authors do seem to partly recognise these model limitations, and describe a process of parameter adjustment to try to reproduce observations. While the results look convincing, I would like to see a discussion of the potential limitations of the model conceptualisation (especially the application of a fixed pool turnover model applied to mineral-enriched drained peat) and the risk of equifinality resulting from adjustment of multiple parameters. In other words, can the authors provide a convincing case that the model is not getting the right answers for the wrong reasons?

Regarding N2O emissions the model struggled to reproduce observations and this aspect of the study is consequently somewhat downplayed. This is unsurprising – N2O is difficult to predict or even measure, and the EC dataset the authors use for this study is one of very few available for agricultural peatlands, so there may be scope to comment a little more on this in the paper. With apologies for recommending a paper I was involved in, a recent EC N2O study of a cropland in England (Cowan et al., GCB 2025) may help with the interpretation of data from the Finnish study site (I guess this probably appeared too late for this submission). We found highest emissions under conditions of high soil temperatures and intermediate high soil moisture (associated with irrigation) and little or no immediate response to fertilisation events. Would the LDNDC model reproduce this?

Whilst I am no expert on carbon modelling, my understanding is that models based on multiple pools with fixed turnover rates are not able to capture the key role of protection mechanisms (see e.g. Schmidt et al., Nature 2011). In the case of natural peatlands, the protection mechanism is waterlogging/anaerobicity. Is LDNDC able to represent this process adequately, and more generally is the pool-turnover approach justifiable for a peat soil? See also comment below about mineral association.

Detailed comments

L28 – Statement that Finland accounts for 1/3^rd of European peatland area may depend on the definition of Europe (EU? Everything west of the Urals? Something else?). Please clarify.

L39 – It is possible to either raise the water table, or to reduce the water table depth, but I don’t think that “Raising the water table depth” makes sense. Depth is positive downwards (a 2m deep swimming pool is deeper than a 1m deep swimming pool) so either refer to something like groundwater level (positive upwards) or amend the terminology around WTD.

L50 – See also Cowan et al (2025) which suggested that these intermittent emission pulses coincided with disturbance events during periods of low N demand, but not with fertiliser events when demand was high.

L68 – What is the justification for the statement that LDNDC provides a suitable basis for modelling peat soils, given that the preceding text states that it has been shown to work on mineral soils? Perhaps better to say ‘may provide a suitable basis’?

L225-240 – It is not entirely clear whether some of the parameter values (e.g. Ksat) were derived from actual measurements, literature, or expert judgement. If the latter, can the authors confirm that parameter values were not iteratively fitted in order to reproduce the observations (potential equifinality/over-parameterisation issue noted above).

L239 – The presence of underdrains greatly enhances the movement of water between the peat and the ditch network (in both directions, depending on relative water levels). This effectively leads to a much higher but more heterogeneous hydraulic conductivity within the field. How was this handled in the model? A 1d model that treats the soil simple as a vertical two-layer system with uniform Ksat may not be able to reproduce field-scale variations in water table (or therefore emissions) if the drains are not somehow represented.

 L243 and onwards – conventionally CO2 emissions from land are reported  (e.g. in IPCC guidance) in t CO2-C/ha/yr, or t CO2/ha/yr. I would suggest using one of these, and avoiding any ambiguity in the units used (kg/ha could be C or CO2).

L276 – How meaningful is the constant 15 cm WTD scenario? This is not environmentally realistic and appears an obvious outlier compared to the other scenarios in Fig 3. At the least, it would be helpful to also include a 15 cm variable scenario so that it is possible to make comparisons – at the moment it is hard to know how to interpret the difference in emissions between 15 cm fixed versus 30 or 50 cm variable scenarios.

L278 – ‘Squeezing’ is an odd term here – ‘rescaling’?

L331 – Strictly I would consider that NECB should also include aquatic C loss and the CH4 flux (hence why we have referred to NEE + harvest as ‘NEP’, although I appreciate that this term is not used consistently either). If NECB is retained then please acknowledge which terms are omitted.

L343 – Regarding drier measured vs simulation conditions in early summer, see my comment above about the role of subsurface drains.

L373 – The units here (kg C/m2/yr) are different to the ones used earlier. I recommend using either t CO2-C/ha/yr or g CO2-C/m2/yr throughout the paper. Similarly for N2O, you could use kg N2O-N/ha/yr for consistency with IPCC emissions reporting.

L378 – Delete ‘get’

Figures all look nice but some fonts (especially numbers along axes) are small, and will become very small if they are reduced in size for final publication - please enlarge.

Fig 7 – I think it is more conventional to plot observations on the x axis and simulated values on the y axis.

Fig 8 – Grey dots in lower plot (EC N2O) fade into the background a bit, but are really rare and important data so perhaps make them more prominent, e.g. black? The years in these plots should also be labelled with the crop (grass/barley) that was present in that year, to help with interpretation (also for LAI and ET)

L424 – See general comments about the validity of using a fixed-turnover C model for peatlands.

Fig 12 – Use ‘decrease’ and ‘increase’, not  ‘decreas.’ and ‘increas.’ As with other figures, fonts will be too small if this figure is smaller in a paper.

L460 – It is a significant overstatement to claim that the match between simulated and obsvered values (for a limited number of observations, following what seems to be a considerable amount of parameter adjustment) ‘confirmed’ that the model is suitable for agricultural peatlands. At best, I think the results maybe ‘supported’ the use of the model, subject to caveats.

L465-471 – See Cowan et al (2025) for some more recent EC measured N2O flux data from an agricultural peatlands.

L475 – For obvious reasons I would be interested to know how the results (observations and model outputs) compared to the Evans et al (2021) relationship between CO2 and WTDe (i.e. drained peat depth). In one scenario (WTD 50 cm, shallow peat ~ 38 cm) the water table was not raised into the peat layer (at least on average) so our simple function would predict no reduction in emissions. This is over-simplified of course, because with variable WTD will rise into the peat layer during wet periods, but still I would expect the emissions reduction to be small (also if a more sophisticated aerated C stock value is used). If the authors are  able to offer any further insights about this that would be a useful addition to the discussion.

[I see that this question is briefly discussed on L518, but without really commenting on why some mitigation is predicted in this situation. I guess it may be for the reasons above?]

L491 – The %C data in Table 1 indicate a high degree of mineral mixing with the peat in most or all of the fields. Have the authors considered whether organic matter stabilisation onto mineral surfaces might influence carbon turnover and emissions? It does not seem like LDNDC considers this possible mechanism. It seems potentially more important than (unspecified changes in) ‘peat quality’ as a possible  reason why CO2 fluxes and C stocks might become partly decoupled.

L505 – See earlier comment about the realism (or otherwise) of a flat 15 cm WTD scenario. Even a very well managed paludiculture field would struggle to maintain this uniformity, so I am not convinced that this scenario is meaningful in practical terms. A variable 15 cm scenario would be more enlightening (as the subsequent discussion here suggests) so why not include this? If a variable 15 cm scenario (with biomass harvesting) predicted that the system would be close to C-balanced that would be an interesting and important result.