Shahriyer et al. modified the water table calculation in the LandscapeDNDC model, reliably simulating changes in LAI, the water table, CH4 flux, CO2 flux and soil C storage along the forest development trajectory under different management regimes (continuous cover vs. rotational forestry) on drained peatlands. The model parameter setting and performance are well supported by field observations. As the first model being able to simulate such changes along the conversion of peatland to forests, this modification of LDNDC may provide important implications for forest management strategies in drained peatlands. However, I would suggest that the authors expand the scope of their discussion and explicitly introduce the pros and cons of different forest management strategies, as well as the limitations of the current simulations (see my specific comments). Other than this, I recommend accepting the paper for publication following minor revisions.  

Specific comments:

•    L30: I think you can be specific here. C:N is an indicator for the “nitrogen” status. 

•    L147: Why did you use RCP4.5 instead of the actual observations of atmospheric CO2 levels, given that no future climate scenario forecasting was conducted in this study?

•    Table 1: Some other publications may contain layer-specific C, N, C:N information, such as Peltoniemi et al. (2023). However, as the values may not vary much across layers, the current setting (where the same value is used for 0-90 cm) is probably okay.  

•    Fig. 1: There appears to be a mismatch in the seasonal patterns between the observations and the simulated variables here (especially in the autumn), as well as in Fig. 5. Is this my illusion or could there be an explanation for this?

•    L250: What do you mean by “during harvest and infect spruce LAI…”?

•    L259: the “modeled” 10cm…

•    Figs. 6&7: It seems that the high values of GPP are overestimated by the model, especially in postharvest rotational forests. This was explained in L372 that the discrepancy arose from the overestimated recovery speed of vegetation. However, I think exploring a bit more into potential processes would be helpful for future improvement of the model. For example, could high photosynthesis or low self-thinning be responsible for the overestimation of GPP?

•    Discussion: It is good that the authors compared their simulations with various field observations. However, I feel that the current discussion focuses only on the reliability of the simulated variables (which is good). Additionally, it would be valuable to see the practical implications of the model outputs, as well as the limitations/future perspective of LDNDC. 

(1)    Practical implications: The authors did interesting simulations of different management practices (i.e., continuous cover vs. rotational forestry). I believe a section/paragraph dedicated to summarizing the pros and cons (in terms of carbon and water cycles) of the two practices may provide valuable information for the management of drained peatland ecosystems. 

(2)    Model limitations: The authors noted the current model’s failure to capture the effect of drought, which may explain why the CO2 balance in 2018 was not accurately represented. This highlights the need for the model to better represent drought-induced tree mortality. Additionally, I wonder if mismatched seasonality (of LAI or CH4 flux, see my previous comment) could be due to the function used to simulate vegetation phenology. By examining the simulated output variables and intermediate parameters more closely, I believe the authors can identify the modules/functions that are underperforming. Discussing such limitations could have important implications for future model development.

References

Peltoniemi, M., Li, Q., Turunen, P., Tupek, B., Mäkiranta, P., Leppä, K., et al. (2023). Soil GHG dynamics after water level rise – Impacts of selection harvesting in peatland forests. Science of the Total Environment, 901, 165421. https://www.sciencedirect.com/science/article/pii/S0048969723040445

Evaluation of the LandscapeDNDC model for drained peatland forest managements, LDNDC v1.35.2 (revision 11434)

The manuscript by Ahmed Hasan Shahriyer et al. describes developments of the LandscapeDNDC model targeting the representation of forestry on drained peatlands in Finland. The authors present extensions to the LDNDC model to represent lateral water movement as a consequence of drainage, and they alter the behaviour of the recalcitrant humus pool to better capture the CO₂ dynamics. In addition, they provide a parameterization of the model to capture these conditions, and optimize a large set of parameters to capture soil processes.  

The manuscript provides an overall detailed assessment of the model's ability to capture the impacts of draining on the hydrology of the sites, as well as observed fluxes of CO₂ and CH₄ in the different sites. While the development of the new lateral water movement is well tested and illustrated in the manuscript (with a possibility to strengthen this even more by moving the simulated WT results from the Supplementary material to the main text), I miss a rationale and analysis of the changes in the recalcitrant humus pool: Why were these changes made, and how does this alter the model's ability to capture this system? 

The manuscript is written in a nice and clear manner, and provides a good overview of the changes. I would like to suggest to improve the description of the sensitivity tests (detailed comments below), and a general review of the figures would be welcome (variables and units on the axes, add sufficient information to the figure captions). Moreover, the current discussion largely repeats the results and does not address uncertainties and shortcomings in the model very well, nor does it provide a good comparison with the literature. 

Once these points are addressed, I expect that the manuscript will be acceptable for publication in Geoscientific Model Development and will be a good basis for future research with this constellation of the LDNDC model. 

Detailed comments

L. 27: The contrast between nutrient-rich and nutrient-poor forested peat soils is not entirely clear in this section - maybe in partly because of missing literature. I trust that the statement that drainage of nutrient-rich soils is a source of CO₂ would also apply to nutrient-poor soils? And do the authors have sources for the impacts on both types (both the effect of drainage, and the offsetting effect of forestry)?

L. 47: There seem to be two contradicting arguments in this paragraph: On the one hand, CCF maintains a higher WT  (L. 48), but the transpiration is large enough to keep a low WT that avoids hindering tree growth (L. 51). I think it would be nice to clarify this, and provide a good overview of the trade-offs affected by WT here (peat degradation, tree growth, methanogenesis).

L. 55: The Lettosuo site has not been introduced here yet - maybe provide a brief description of the site. 

L. 111 and elsewhere: To enhance understanding, I would recommend to add the units of the variables in the explanation of the equations. 

L. 112: Defining negative z_gw as below the surface seems to contradict with the definition in Eq. 1 (q will only be >0 if z_gw is more negative than the reference level Z_gw)

L. 114: Please explain the scaling of the recalcitrant old humus pool. I trust that the pool is computed dynamically, so where is the scaling happening, and how is excess C treated? And which shortcoming in the model is this change trying to resolve? 

Fig. A1: Please add a scale. Also, it would be interesting to add the date of the photo in the caption (for comparison with the management dates in section 2.2). 

L. 147: Please clarify the CO₂ driving data: RCP 4.5 does not start in 1963. Also, please add a reference. 

Section 2.3: It would be nice to distinguish here between data used to drive the LDNDC model, and data used for evaluation

Table A1: Please add units for the variables. The ranges are hard to interpret without knowing the unit in the model. 

L. 194: An NSE value for "good" model performance ought to be variable-specific: Some variables are much harder to capture in models than others. 

L. 225: The parameter optimization method needs to be described in more detail. It is unclear to me how the best parameter value for an individual parameter can be found while varying a set of parameter values at once. Is there any optimization involved, or are you simply treating all sets of parameters as samples? In the case of the latter, the number of simulations could be worryingly low, given the high dimensionality of the problem (3 sets of 9 parameters to vary at the same time). But I may misunderstand the exact method used here. 

Also, it is unclear whether NEE is the only variable used to optimize against in the sensitivity analysis. It would also be good to report the default value used in set 2 and 3 while varying set 1. 

Finally, it would be nice to not only provide the ranges of parameter values that you apply in the sensitivity tests, but also the final values that you end up using. 

Fig. 2: What is "soil moisture" here referring to? Volumetric moisture content or saturation of the pore volume? The modelled value for 20 cm depth seems to peak around 60%, but there is one episode in May 2018 where it suddenly rises to 80% - it would be nice if the authors could investigate why the model produces this peak. Also, It would be interesting to see the model performance outside the growing season: Does LDNDC capture the freeze-thaw dynamics, too?  And what causes the model to be so water-conserving in the 2018 drought?     

L. 269: I like Fig. S5 and would urge the authors to bring it into the main manuscript (and include a description of it). The figure displays the difference between the three setups, and hence illustrates both the lateral groundwater movement that was introduced in LDNDC in this manuscript, and the difference in influence of the forest practice on water dynamics, which is one of the key arguments in the introduction. 

Section 3.3: It would be nice to describe the two simulations with different water tables in more detail. How was the WT variation imposed (varying Z_gw?), and what was the rationale for these changes? 

Figure 5: It would be nice to have the legend outside the figure box, or with a box around it - to ensure that there is a clear difference between observation points and legend. 

L. 283: There is an interesting difference in the simulated and observed timing of CH₄ fluxes. The authors describe this to some extent, but do not comment on it or provide possible explanations. It would be nice to hear whether the authors have a possible explanation for it (this could also be a part in the Discussion).

Fig. 6: Is this based on the half-hourly data, or the daily averages? Please add this information to the figure caption. 

Fig. 7: "Uncertainties in the EC-based NEE [...] are shown with the error bars". Please explain how these uncertainties were derived.

Section 3.4: The analysis is based primarily on a comparison between daily (or subdaily?) model and observation estimates (Fig. 6) and annual sums (Fig. 7). It would be interesting to learn a bit more on the seasonality in the CO₂ fluxes: How well are these captured, and how well does the model capture differences between individual summers (similar to the analysis of the CH₄ flux in section 3.3)? 

Section 3.5: It would be nice to see an analysis that describes the impact of the changes made to the model. You have captured the description of the drainage nicely by showing differences in hydrology between the sites. What has been the impact of the changes in the recalcitrant humus pool? Has this been improving the seasonal behaviour, the long-term trend, or both? 

Section 4: The discussion is mainly used to highlight results, and to describe the comparison between simulation results and observations. It would be nice to strengthen it by (1) discussing the importance of the model developments (new functionality, altered parameterization) made in this study, (2) comparing the results more to other studies, and (3) highlighting potential use of the model with this new functionality. 

Language editing:

L. 93: replace "differentiated" with "are differentiated"

L. 95: "matter": should this read "organic matter"?

L. 97: what is meant with CH₄ deposition here? 

L. 97: Check sentence "and as well as CH₄ oxidation"

L. 138: Replace "were used" by "was used"

Table 1: Please finalise the sentence in the figure caption 

L. 205: replace "was also calculated" with "were also calculated"

L. 214: check spelling of "python"

Fig. 1: The difference between stars and circles is difficult to recognize. 

L. 339: replace "did not gave" with "did not give". Also, consider explaining what you mean with "did not give the same perspective".