General:

The paper under my reviewing addresses an important topic and contributes to improving the understanding of soil N₂O emissions from global land ecosystems. Overall, the authors have done a great job. They integrated soil nitrification-denitrification processes into the DGVM LPJ-GUESS model, clearly explaining and documenting the added mathematical formulations and parameters, although some complex processes were simplified during the development due to the “large-scale” model structure. Additionally, they examined the model’s performance using a valuable site-level data set and quantified the environmental factors driving changes in global N₂O emissions among terrestrial ecosystems through model-based scenario analyses. I think this topic is interesting and of great relevance to GMD. The manuscript is in general well-written and easy to follow. However, a few pieces of information are missing, and some sections require minor adjustments before it can be accepted for publication. Below, I provide specific comments, which I hope will help further improve the manuscript.

comments and suggestions:

The introduction is well stated. The description is sufficient to understand what has been developed in previous DGVM studies in terms of N₂O-related processes. However, the organization in some parts can be improved.

LN65-66: 0.2-1.8% of what?  Is it the proportion of N fertilization lost as N₂O?

LN97-100: What specific cropland management practices are included in your model? How are these practices set up in the global simulations? Since conservation agriculture can significantly influence N₂O fluxes in the fields, even though most practices mitigate soil CO₂ emissions by enhancing soil C storage. It is better the specify the management setups somewhere in the methodology section.

Fig.1: In nitrification processes, where is the heterotrophic pathway? How do you consider this in the model depending on DON? This is an important process in acidic soils or environments where autotrophic nitrification is less dominant. It should be mentioned in the model description.

LN191-195: Maybe the 38℃ is  needed further discussion  to represent the entire nitrification process, as NOB bacteria significantly favor higher temperatures compared to AOB and AOA.

LN420-424: It would be valuable to compare crop yields (or N use efficiency). For instance, in the high N fertilization scenario, was the modelled yield lower than the observed yield? If so, this could suggest that less N was taken up, resulting in less N removal from the system and leaving more N to be emitted as N₂O.

LN450-453: I would suggest to add the observed WFPS to Fig. S3 to clearly compare the mismatch between the simulation and observation, if the reported data are available.

LN482-484 and Fig.5d: It could be interesting to identify which thin-dashed blue WFPS corresponds to each N treatment. Does the N250 scenario, with the highest N, use the most water in the soil?

LN545-547: I seems to find an explanation for the reduced N₂O emissions under elevated CO₂ concentrations.

LN651-660: For large-scale applications, fertilizer type-usually simplified as the ratio of NH4 to NO3 in the model-is one of the factors contributing to the simulated uncertainty in N₂O emissions. Does LPJ-GUESS actually account for this when performing the S5 runs ("Const_Nfert")?

Ma et al. describe a new parametrization for soil mineral N transformation in the widely used DGVM LPJ-GUESS model. They mainly take NH₃ volatilization, nitrification, and denitrification processes into account, with well explaining the mechanistic parametrization of N dynamics. They also provide a global evaluation of the model development in terms of simulated N₂O fluxes among land ecosystems. Overall this is a well written paper with a good structure in simulation experiment design and assessment results. Although modelling soil N₂O emissions is a challenging work with large uncertainties, authors do a good job by giving a solid discussion on model limitations of N transformation which are well referenced. I think this is an exciting and timely work that greatly advances existing modeling tool and will benefit to biogeochemical research community, it can be accepted by GMD for publication after minor adjustments.

Specifically:

The new model development section is clear and thorough; however, the model simulation protocol needs to be further clarified. For instance, in sections 2.3.1 and 2.3.2, are the forcing data (e.g., climate and N deposition) the same for the site-specific and global runs? Additionally, how does LPJG estimate the dynamics of soil pH during N transformation processes? A brief description of the model's representation of soil physical properties should also be included in section 2.3.

The site-level evaluation is convincing, but the global-scale results require further discussion. Based on Figures 6 and 7, the simulated N₂O emissions from pastures are not entirely convincing. It would be expected to see higher N₂O fluxes from pastures starting in the 1980s onward, given the increased use of manure and fertilizers during this period. Additionally, how does the model account for manure deposition from livestock waste on grazing pastures? This aspect should be clarified to better understand the model's representation of pasture emissions.

Line78: “... (DGVMs)”, a citation of DGVM would be appreciated.

Line 119 “...of a proportion of the aboveground biomass”, What is the proportion used in the model? It appears to be a dynamic variable that varies depending on the study region, but what is the default value applied in your global simulation? This information is important for understanding the baseline assumptions and how they influence the results.

Line347: Could you provide specific details on how the soil physical characteristics are set up at the study sites? Are these characteristics treated as constant throughout the experimental periods, or are they modeled as time-dynamic variables?

Figure 4c: Why is there such a large mismatch between the simulation and observations in April?

Sect. 3.1.2: What is the soil texture at the study sites? In most Dynamic Global Vegetation Models, the simulated WFPS is highly sensitive to soil texture, with very fine soils typically exhibiting higher WFPS. An accurate representation of hydrology in the model would significantly improve the simulation of all N₂O-related processes that depend on WFPS (see your Figure 1). Since different soil texture classes undoubtedly affect WFPS, a brief explanation of how WFPS is associated with soil texture in your model would be greatly appreciated.

Figure 7: Here again, the low N₂O emission on global grazing pasture makes non-sense to me due to the livestock waste inputs and N fertilizer use in the recent decades.

Line745-752: How do N₂O emissions in the model respond to different agricultural management practices, such as fertilizer application, tillage, or crop rotation? How closely do the simulated results align with real-world observations? Additionally, are these management practices explicitly included in your global simulations? If not, I think your discussion is not strongly supported.