the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Dec 2024
Review and Synthesis: Peatland and Wetland Models Simulating CH4 Production, CH4 Oxidation and CH4 Transport Pathways
Abstract. Peatlands play an important role in the global CH4 cycle and models are key tools to assess global change effects on CH4 processes. It remains unclear how well our existing wetland modelling frameworks are suited to peatland questions. Therefore, we reviewed 16 peatland or wetland models operating at different spatial (seconds-to-decadal) and temporal (soil core-to-global) scales, having different spin-up periods for carbon pool stabilization and various CH4 production, oxidation and transport processes. Through a literature review, model specific advantages and limitations, common and specific driving inputs of all models and critical inputs of individual models impacting CH4 plant-mediated transport, diffusion and ebullition were summarized. The 16 reviewed models were qualitatively ranked 0 to 4 (none-to-full process representations) with respect to CH4 production, oxidation and transport. The most common temporal and spatial scale for 14 models was daily time-step and field scale respectively, while the spin-up stabilization periods of different carbon pools (peat, litter, roots, exudates, microbial, humus, slow, fast) of all models ranged 7 to 90102 years. With regards to CH4 production and oxidation, 50 % of reviewed models (Ecosys, CLM-Microbe, ELM-Spruce, Peatland-VU, Wetland-DNDC, TRIPLEX-GHG, TEM, CLM4Me) exhibited full to adequate process representation. Meanwhile 44, 44 and 25 % models exhibited full to adequate process representation for plant mediated transport, diffusion and ebullition respectively. This meant there is ample scope to improve ebullition processes in the remaining 75 % models. We conclude that existing models are adequate for site-level CH4 flux assessments but may lack a predictive understanding of CH4 production pathways.
- Preprint
(1508 KB) - Metadata XML
-
Supplement
(246 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3852', Ashley Ballantyne, 14 Jan 2025
Summary: Here the authors provide a qualitative review of the current state-of-the-art peatland methane models and evaluate them based on the physical, chemical, and biological processes incorporated in order to simulate CH4 fluxes.
General Comments: As a researcher not directly involved in the development of methane models, but sometimes reliant on their simulations, I found this to be a helpful high-level ‘dummies’ guide to methane models. The authors identify key attributes and limitations of these models that should be considered before researchers select a particular model or rely on its output. Although I did find this qualitative review informative, the information could possibly be presented with greater clarity and a more illustrative way.
The authors do a decent job of summarizing across this diverse array of models, but I wonder if the information was better synthesized graphically it might be more useful. For instance, as much as I detest Venn-Diagrams perhaps this is a situation where differences and similarities among these models could be better illustrated. I also found some of the figures to be not that informative. For instance, maybe figure 1 could have a log-spatial scale on the x-axis using data in Table 2. Also figure 2 seems to have CH4 pathways on both axes and thus is more confusing than helpful. Perhaps a combined figure with your scores of transport (Fig. 4) vs. process (Fig. 3) or number of inputs would help researchers identify the groups of models best suited for their research questions. Alternatively, a dichotomous tree could be helpful for researchers in selecting a suitable model for their research specific research question, while providing focus in the conclusion section.
There was a detailed description of how methane is measured in the introduction using bottom-up vs. top-down approaches but this was not imperative for the rest of the paper.
There were a couple of key citations that I was expecting to find, such as (Evans et al. 2021) and others noted in PDF comments, but were not included in the references
Coupled diffusion of heat and CH4 is challenging! With my limited experience modeling CH4 in peatlands, it seemed that we spent 75% of the time trying to get the thermic properties of heat diffusion correct and then only 25% of the time actually simulating the diffusion of CH4. So these coupled diffusion processes are challenging, especially with complex hydrology and/or snowcover.
All in all, I think that this is an effective qualitative review. However, the attempts at making this review more quantitative, including some figures that were actually confusing, actually made this review less effective. I am not sure if this article is submitted as a research article or a review article but it seems to fall between the cracks- not quite a research assessment of models and not quite a review.
Specific Comments:
See PDF with editorial comments added
References
Evans, C. D., M. Peacock, A. J. Baird, R. R. E. Artz, A. Burden, N. Callaghan, P. J. Chapman, et al. 2021. “Overriding Water Table Control on Managed Peatland Greenhouse Gas Emissions.” Nature, April, 1–7.
-
AC1: 'Reply on RC1', Amey Tilak, 19 Feb 2025
Author responses to Referee 1 (RC1)
Referee comments: Summary: Here the authors provide a qualitative review of the current state-of-the-art Peatland methane models and evaluate them based on the physical, chemical, and biological processes incorporated to simulate CH4 fluxes. General Comments: As a researcher not directly involved in the development of methane models, but sometimes reliant on their simulations, I found this to be a helpful high-level ‘dummies’ guide to methane models. The authors identify key attributes and limitations of these models that should be considered before researchers select a particular model or rely on its output. Although I did find this qualitative review informative, the information could possibly be presented with greater clarity and a more illustrative way.
Author responses: Thank you for your thorough review. Please find below the detailed responses which hopefully bring more clarity into this review.Referee comments: The authors do a decent job of summarizing across this diverse array of models, but I wonder if the information was better synthesized graphically, it might be more useful. For instance, as much as I detest Venn-Diagrams, perhaps this is a situation where differences and similarities among these models could be better illustrated. I also found some of the figures to be not that informative. For instance, maybe figure 1 could have a log-spatial scale on the x-axis using data in Table 2. Also figure 2 seems to have CH4 pathways on both axes and thus is more confusing than helpful. Perhaps a combined figure with your scores of transport (Fig. 4) vs. process (Fig. 3) or number of inputs would help researchers identify the groups of models best suited for their research questions. Alternatively, a dichotomous tree could be helpful for researchers in selecting a suitable model for their research specific research question, while providing focus in the conclusion section.
Author response: Thank you for your constructive suggestions. We have added the following information in the manuscript text, which hopefully exhibits greater clarity.
- We combined previous Figure 1 (carbon pool stabilization), Table 2 (spatial and temporal scale of each model) into new Table 2. This new Table 2 now reports the model’s name, temporal and spatial scale, spin-up stabilization periods including details on accelerated spin-up periods and turnover times of different soil carbon and litter pools for each model.
- In addition, we report in the temporal scale column, whether each model runs at a single point scale or grid scale resolution and mention the grid scale resolution of each model.
- The new Table 2 should enable future model users to select the appropriate model for their intended use or at least serve as an initial screening point for model selection.
- We also combined previous Figure 3 (CH4 production & oxidation indices) and Figure 4 (CH4 transport indices) and into a new Figure 1 where processes of CH4 production, oxidation, plant-mediated transport, diffusion and ebullition within each model distinguished into five indices (0 to 4) using five radar plots (one radar plot for each CH4 process indices).
These five radar plots will enable future model users to select the appropriate model for their intended use. - We also combined previous Figure 2 into Figure 1 using the Venn This Venn diagram clearly shows the 12 models simulating all CH4 transport pathways, models like MEM simulating diffusion (DF) and ebullition (EB), WETMETH simulating no CH4 transport pathways, while CH4MOD only simulating plant transport (PL) and ebullition (EB), and PEPRMT only simulating plant transport (PL) and diffusion (DF).
Referee comments: There was a detailed description of how methane is measured in the introduction using bottom-up vs. top-down approaches but this was not imperative for the rest of the paper.
Author responses: The authors agree and delete the bottom-up vs. top-down approach section.Referee comments: There were a couple of key citations that I was expecting to find, such as (Evans et al. 2021) and others noted in PDF comments but were not included in the references.
Author response: All the suggested key references added in manuscript text and included in the list of references. New added references are as follows:- Basu, S., Lan, X., Dlugokencky, E., Michel, S., Schwietzke, S., Miller, J. B., Bruhwiler, L., Oh, Y., Tans, P. P., Apadula, F., Gatti, L. V., Jordan, A., Necki, J., Sasakawa, M., Morimoto, S., Di Iorio, T., Lee, H., Arduini, J., and Manca, G.: Estimating emissions of methane consistent with atmospheric measurements of methane and δ13C of methane, Atmos. Chem. Phys., 22, 15351–15377, https://doi.org/10.5194/acp-22-15351-2022, 2022.
- Evans, C.D., Peacock, M., Baird, A.J., Artz, R.R.E., Burden, A., Callaghan, N., Chapman, P.J., Cooper, H.M., Coyle, M., Craig, E., Cumming, A., Dixon, S., Gauci, V., Grayson, R.P., Helfter, C., Heppell, C.M., Holden, J., Jones, D.L., Kaduk, J., Levy, P., Matthews, R., McNamara, N.P., Misselbrook, T., Oakley, S., Page, S.E., Rayment, M., Ridley, L.M., Stanley, K.M., Williamson, J.L., Worrall, F., Morrison, R: Overriding water table control on managed peatland greenhouse gas emissions, Nature 593, 548-552, https://doi.org/10.1038/s41586-021-03523-1, 2021.
- Kwon, M. J., Ballantyne, A., Ciais, P., Qiu, C., Salmon, E., Raoult, N., Guenet, B., Göckede, M., Euskirchen, E. S., Nykänen, H., Schuur, E. A. G., Turetsky, M. R., Dieleman, C. M., Kane, E. S., Zona, D.: Lowering water table reduces carbon sink strength and carbon stocks in northern peatlands, Global Change Biology, 28, 6752-6770, https://doi.org/10.1111/gcb.16394, 2022.
- Watts, J.D., Kimball, J.S., Bartsch, A., McDonald, K.C.: Surface water inundation in the boreal-Arctic: potential impacts on regional methane emissions, Environmental Research Letters, 9 075001, http://dx.doi.org/10.1088/1748-9326/9/7/075001, 2014.
Referee comments: All in all, I think that this is an effective qualitative review. However, the attempts at making this review more quantitative, including some figures that were confusing, made this review less effective. I am not sure if this article is submitted as a research article or a review article, but it seems to fall between the cracks- not quite a research assessment of models and not quite a review.
Author responses: Thank you for the constructive suggestions.- We agree that our article is more effective as a qualitative review, since we reviewed the published literature on 16 peatland or wetland models but did not simulate any of the reviewed models using common assessment criterion or compared any published model outputs. Thus, we have revised tables and figures to focus on the qualitative review aspects and removed any confusing quantitative aspects.
- The fully revised Table 2 and Figure 1 will provide the future model users with information to select the appropriate model for their intended use or at least serve as an initial screening point for the model selection. These revised tables and figures will be helpful to future model users that otherwise would be conducting their own desk studies and putting a great amount of effort into gathering this information.
Referee comments: Maybe a figure showing your rank of how mechanistic a model is with the number of inputs required.
Author responses: We like the reviewer’s suggestion and already report on the inputs of each model in Table 4. In the revised manuscript text, we include a brief discussion of how mechanistic a model is based on its numbers of inputs. We did not make a new figure as we thought it would be redundant with the richer information of model inputs provided in Table 4.Referee comments: Author responses to Referee’s comments posted in PDF manuscript file.
Referee comments: Add “from” on line 28
Author responses: Added in the manuscript.
Referee comments: Do all these estimates converge on the same number?
Maybe report the range.
Author responses: Added in the manuscript the range of carbon stored in peatlands.Referee comments: Is this citation for peatlands or wetlands?
Author responses: Reviewer is correct in pointing out that the overall paper is focused on wetlands (Fluet-Chouinard et al., 2023). However, the paper also reports peatland loss estimates and thus we cite it for peatlands. In the revised text we say, “wetlands and peatlands across the globe have been drained.”
Referee comments: Make two sentences here: Although chamber measurements quantify CH4 fluxes from specific source areas, they require multiple replications to capture spatial and temporal variations and often do not provide continuous flux data, while the EC method provides continuous temporal CH4 flux data, but these measurements are not easily attributable to a specific microsite type (Erland et al., 2022).
Author responses: Based on a previous comment by the referee, we have removed/deleted the entire section on bottom-up and top-down measurements containing the above two sentences.Referee comments: Add atmospheric transport in line 64.
Author responses: Added “atmosphere transport” in the manuscript.Referee comments: Add “and” in line 80.
Author responses: Added “and” in the manuscript.Referee comments: Strikethrough lines in line 97 and rewrite “so that model users can decide”
Author responses: The suggested lines have been removed and rewritten in the manuscript.Referee comments: Published in English, Fortran is not English.
Author responses: Mentioned in the manuscript.Referee comments: How many models were found that did not meet the criteria?
Author responses: We have now add in the manuscript text that the peatland models that did not meet the criteria were MPeat and Mpeat2D (Mahdiyasa et al., 2021), Digibog and Digibog_Hydro (Morris et al., 2012; Young et al., 2017), Soil and Water Assessment Tool (SWAT) utilized at the watershed scale to simulate riparian wetlands (Rahman et al., 2016) but rarely utilized as a peatland model. Although recently, Melaku et al. (2022) developed wetland specific algorithms consisting of groundwater table, CO2 emissions, and net ecosystem exchange. But all these above-mentioned models did not simulate CH4 production, oxidation and transport processes.Referee comments: maybe add the spatial resolution of the models, especially global.
Author responses: The revised Table 2 in the manuscript now reports whether the model is simulated at the single point scale or grid scale resolution and mentions the grid scale resolution of each model.Referee comments: Maybe a figure showing your rank of how mechanistic a model is with the number of inputs required.
Author response: We did not make a new figure as we thought it would be redundant with the richer information on individual model inputs already reported in Table 4.Referee comments: Instead of WTDs mention “Water Table Depth”
Author responses: Instead of WTDs, we now mention “Water Table depth” in the manuscript.Referee comments: So, is there any diffusion in HIMMELI model, or is it all ebullition?
Author responses: HIMMELI simulates diffusion, ebullition and plant-mediated transport. The revised Venn diagram in Figure 1 now shows HIMMELI simulating all three CH4 transport pathways.Referee comments: This is an important conclusion “With regards to model inputs, WETMETH (Nzotungicimpaye et al., 2021), MEM (Lai, 2009) and PEPRMT (Oikawa et al., 2017) are less input intensive, but the modelled CH4 fluxes simulated the measured CH4 fluxes from different peatland and wetland sites located in pan-arctic, tropical, temperate and boreal regions (Lai, 2009; Oikawa et al., 2017; Fertitta-Roberts et al., 2019; Nzotungicimpaye et al., 2021)”. How accurately do these models simulate CH4 fluxes and at what scales? Could these models be identified graphically?
Author responses: We added a new Table 7 in the manuscript where we report based on the published literature, the goodness of fit (R2, normalized root mean square error (NRMSE), normalized root mean square difference (NRMSD) and Nash Sutcliffe Efficiency Index) between measured and simulated CH4 fluxes for all the reviewed models. The scale of all models is reported in Table 2.Referee comments: Include the lines “measured CO2 and CH4 are required to test these model simulations”.
Author responses: Included the above lines in the manuscript.Referee comments: Include these words “and among different sites”.
Author responses: Included these words “and among different sites” in the manuscript.Referee comments: Include these words “are parametrized”
Author responses: Included the words “are parametrized” in the manuscript.Referee comments: Add “scales”
Author responses: Included “scales” in the manuscript.Referee comments: Add “ ORCHIDEE is part of the IPSL ESM so some inputs are prognostic variables”
Author responses: Included the above in the manuscript.Referee comments: This paragraph has a lot of information to assimilate; perhaps these could be presented in a table with all models and R2 and RMSE or NSE if reported.
Authors responses: We added a new Table 7 in the manuscript that reports based on the published literature, the goodness of fit (normalized root mean square error (NRMSE), normalized root mean square difference (NRMSD), R2, Nash Sutcliffe Efficiency Index) between measured and simulated CH4 fluxes for all the reviewed models.Referee comments: Perhaps data assimilation approaches could help here see work by Kwon et al. on peatland fluxes in response to WTD.
Author responses: We added in the manuscript text “Utilizing a data assimilation approach to optimize key model parameters related to carbon, hydrology, methane for predicting changes in the C stock and quantifying CO2 and CH4 fluxes under variable water table depths as in the case of ORCHIDEE land surface model (Kwon et al., 2022). For details refer ORCHIDAS (https://orchidas.lsce.ipsl.fr/overview/) (Bastrikov et al., 2018).Referee comments: See work by Basu suggesting tropical sources of recent increase in atm CH4: https://acp.copernicus.org/articles/22/15351/2022/acp-22-15351-2022.html
Author responses: We added in the manuscript text “To improve CH4 predictions from different earth system models (ESMs), continuous flux and environmental in-situ data are required from data scarce regions namely Congo, Amazon and Southeast Asia rainforests, rice crop areas in India and Bangladesh, Savannah, Africa, Central America, South America and microbial emissions from tropical areas (Basu et al., 2022; McNicol et al., 2023; Zhao et al., 2024).Referee comments: Don't most models simulating CH4 also simulate CO2? However, very few of these models consider lateral transport of C.
Author responses: Yes, most models simulating CH4 also simulate CO2. We have now added in the manuscript text on models that simulate lateral transport of C.Referee comments: maybe peer reviewed citation here https://www.nature.com/articles/s41586-019-0912-1
Author responses: Added the reference of Reichstein et al. (2022) in the manuscript https://www.nature.com/articles/s41586-019-0912-1 before USDOE, 2024 reference.Referee comments: This section reads like a laundry list of current machine learning approaches. How about other approaches such as data assimilation for parameter optimization.
Author responses: We have reduced this text substantially and now only briefly mention the potential of machine learning, hybrid modeling and data assimilation for parameter optimization.Referee comments: Add and see work by Watts in the Remote sensing section 4.5. https://iopscience.iop.org/article/10.1088/1748-9326/9/7/075001/meta
Author responses: Added the reference of Watts et al. (2014) in remote sensing section 4.5 in the manuscript https://iopscience.iop.org/article/10.1088/1748-9326/9/7/075001/metaReferee comments: Lines 805-809 on page 26, will this help with future CH4 projections?
Author responses: We agree that there is limited evidence of microbial processes improving future CH4 predictions. However, this modeling advance is necessary for an improved understanding of processes. We now add the following text in the manuscript to give examples of instances where microbial representation has decreased data-model disagreements and emphasized the process relevance. “Availability of genomic data from different peatland and wetland sites across different biomes could enable efficient parameterization of microbial based models such as CLM-Microbe, Ecosys and ELM-SPRUCE that will deepen the mechanistic understanding of CH4 processes and reduce data-model disagreements (Zuo et al., 2024)”.Referee comments: It is not clear how AI enabled platforms or edge computing would be used to measure ebullition.
Author responses: Our revised manuscript text now includes the following “We were referring to the use of AI enabled platforms of edge computing such as edge computing-based AI models, e.g. RNNPool (Saha et al., 2020), and other smart sensors that can be utilized to continuously measure hot spots and hot moments of CH4 ebullition such as the iAMES (An inexpensive, Automated Methane Ebullition Sensor) (Maher et al., 2019), Smart AI-based sensors (EnviTrace) and low cost automated sensor placed inside a floating chamber (Sø et al., 2024)”.Referee comments: no need to add new acronyms in discussion.
Author responses: removed the acronyms in discussion from the manuscript.Referee comments: for global scale predictions? It seems like you are saying that small scale process models are adequate, but not global models
Author responses: Our revised manuscript text now includes the following “ In conclusion, based on the information derived from the published literature of the individual models, the reviewed models exhibited a wide array of different scales and process representations that can be utilized to test different hypotheses using a common assessment criterion to quantitatively rank model processes of CH4 production, oxidation and transport”.Citation: https://doi.org/10.5194/egusphere-2024-3852-AC1
-
AC1: 'Reply on RC1', Amey Tilak, 19 Feb 2025
-
RC2: 'Review of egusphere-2024-3852', Anonymous Referee #2, 16 Jan 2025
In their manuscript, Tilak and co-authors summarize features of a number of models of wetland methane production and transport. Analysis, however, remains superficial and mainly consists of summarizing the statements made by the authors of the original model description papers without deeper analysis. As a result, the authors quite often compare apples with oranges – likely unknowingly -- and the value of the manuscript is questionable.
I suggest rejecting the manuscript.
I was really looking forward to reviewing this manuscript. Having worked in both peatland and methane modelling, I believe a review and synthesis paper is overdue. However, reading the manuscript, I was quickly disappointed. It appears the authors have very little experience with modelling approaches. As a result, the paper does not go beyond the superficial analysis of the model description papers.
Unfortunately there are too many issues with the manuscript to do a complete review. Instead I will list a few of the issues I noticed as examples and to support my suggestion to not publish the manuscript.
However, I very much encourage the authors to add an experienced modeller to their team and submit a rewritten manuscript, as I believe a manuscript on this subject could be interesting to a wider audience.
Here is a collection of issues with the manuscript, ordered by order of appearance in the manuscript:
1) The discussion of top-down and bottom-up approaches (lines 51 ff) appears to be superfluous, as both measurements and the models discussed in the manuscript are usually considered as bottom-up approaches.
2) Spatial scales (lines 145 ff and table 2): I assume the authors collected what appeared in the model description papers? The scales reported (plot/field/regional etc. scale) are, however, rather arbitrary in the modelling context. Most models follow a “big-leaf” approach, where processes are calculated for a single representative specimen, usually a leaf or a plant, and then upscaled to the scale of interest. Much more interesting is whether models only run on a single point, or whether they calculate values for a grid of points, thus allowing the coverage of larger areas and not just single sites. This is completely neglected, though.
3) Spinup time (lines 155 ff): Here many apples are compared with a few oranges. Generally, carbon pools have turnover timescales of years for litter pools, decades for reactive soil carbon, centuries for non-reactive / recalcitrant soil carbon and millennia for specific peatland carbon pools. Thus, the different spinup timescales reported for the different model are determined by the carbon pools considered in the model. Unfortunately, this is not discussed in the manuscript, but the authors only report the different spinup times given by the model authors. This is highly misleading: As an example, the HIMMELI model only reported the equilibration times for the methane specific components of the model. The HIMMELI model, however, needs to be run within a land-surface / carbon cycle model, as HIMMELI doesn’t deal with any of the relevant land surface physics or carbon calculations. Spinup times will therefore be determined by the host carbon cycle model, not by HIMMELI itself.
Much more interesting is the question whether an accelerated spinup procedure is available – this is not reported, though.
4) Process representation evaluations, lines 383 – 417: The manuscript mentions “full to adequate process representation” for plant mediated transport, diffusion, and ebullition. These are, however, not clarified, and to my knowledge there is no community consensus what might constitute “full” or “adequate” process representation. Thus, the manuscript passes judgement on models (“only 25% of the models exhibited full…”) without adequately defining the criteria.
5) Advantages and disadvantages of reviewed models (lines 446 – 494): In these sections, the authors seem to summarize whatever the authors of the original model description papers gave as advantages and disadvantages, it is unclear whether these sections contain significant analysis by the manuscript authors. It is also doubtful whether the same criteria were applied to all models. One example from the disadvantages section: “… while the HIMMELI does not simulate any electron acceptors , except O2.” What is not mentioned in this sentence (or anywhere else) is that non-O2 electron acceptors are mostly handled by an unspecific bulk term, if at all, and very rarely, if ever, explicitly. Thus the “disadvantage” of the HIMMELI model is mainly that they actually reported that their treatment is not perfect, while other authors glossed over this fact. Thus this analysis is incomplete at best and potentially misleading.
Citation: https://doi.org/10.5194/egusphere-2024-3852-RC2 -
AC2: 'Reply on RC2', Amey Tilak, 19 Feb 2025
Authors responses to Referee 2 (RC2):
Referee suggestions: In their manuscript, Tilak and co-authors summarize features of a number of models of wetland methane production and transport. Analysis, however, remains superficial and mainly consists of summarizing the statements made by the authors of the original model description papers without deeper analysis. As a result, the authors quite often compare apples with oranges – likely unknowingly -- and the value of the manuscript is questionable.
Author responses: We thank you for your suggestions and thoughtful feedback. We have now thoroughly revised the manuscript emphasizing its goal as a resource and not as a quantitative comparison of models (for example, using a common assessment criteria). Furthermore, to address the reviewer's 2 comments, we have rectified their concerns and deepened the synthesis in places where it seemed like a superficial analysis. Specifically, we have done the following:- Provided new detailed information on whether models operate at a single point scale or grid scale resolution and mentioned the grid scale resolution of each model.
- Clarified the discussion on the turnover timescales of various soil carbon and litter pools and reported information on the availability and details of accelerated spin up procedures for each model.
- Better defined what is meant by “full” or “adequate” process representation in the context of production, oxidation, plant-mediated transport, diffusion, and ebullition and changed from ranking of models to model categories.
- Based on reviewer 1 comments, we also improved visuals depicting the process representation of models and added more information on how models have been benchmarked so far (details on which sites were used, goodness of fit values, etc.)
Referee suggestions: I was really looking forward to reviewing this manuscript. Having worked in both peatland and methane modelling, I believe the review and synthesis paper is overdue. However, reading the manuscript, I was quickly disappointed. It appears the authors have very little experience with modelling approaches. As a result, the paper does not go beyond the superficial analysis of the model description papers.
Author response: The following points mentioned below are added in the manuscript.- The intent of our review study is to serve as a starting point or initial screening point for scientists and non-expert modelers seeking information on available methane models simulating CH4 production, oxidation and transport processes.
- This review study is conducted to assist future model users who may be performing similar desk studies to narrow down the selection of potentially appropriate models that they want to further analyze. So, we have compiled the information on each model at one single place rather than looking at several published papers of the individual models.
- This review study did not conduct any in-depth analysis using a common assessment criterion across all the 16 reviewed models, but all the information on each model was derived based on the existing published literature.
- We emphasize that the model-assessments solely based on the published literature review may lack a comprehensive model evaluation review, stressing the need for conducting an
in-depth model assessment using a common assessment criterion across all reviewed models that will complement the existing literature review. - Since this review study did not simulate any models using a common assessment criterion, the word “ranking of models” was replaced with model categories with respect to processes of CH4 production, oxidation and transport.
- Importantly, we emphasize that findings from this modelling review study do not reflect the actual capabilities or limitations of the reviewed models, since this information was solely based on the published literature.
- We also emphasize that the reviewed models are constantly evolving and updated by the model developers from time to time, with the updated versions that might or might not be reported.
- Importantly, we revised our wording on the previously reported model advantages and limitations and instead utilized “benefits” and “challenges” of each model and discussed in detail which settings each model may be most appropriate for using.
Referee suggestions: Unfortunately, there are too many issues with the manuscript to do a complete review. Instead, I will list a few of the issues I noticed as examples and to support my suggestion to not publish the manuscript. However, I very much encourage the authors to add an experienced modeler to their team and submit a rewritten manuscript, as I believe a manuscript on this subject could be interesting to a wider audience.
Author’s responses: We have addressed your suggestions and worked closely with those authors who have modelling experience in our team to enhance the manuscript, making it clearer and more robust. We appreciate your encouragement and believe the revised version will engage a wider audience.Referee suggestions: Here is a collection of issues with the manuscript, ordered by order of appearance in the manuscript:
Referee suggestions: The discussion of top-down and bottom-up approaches (lines 51 ff) appears to be superfluous, as both measurements and the models discussed in the manuscript are usually considered as bottom-up approaches.
Author’s responses: We agree and have deleted this from the Introduction section.Referee suggestions: Spatial scales (lines 145 and table 2): I assume the authors collected what appeared in the model description papers? The scales reported (plot/field/regional etc. scale) are, however, rather arbitrary in the modelling context. Most models follow a “big-leaf” approach, where processes are calculated for a single representative specimen, usually a leaf or a plant, and then upscaled to the scale of interest. Much more interesting is whether models only run on a single point, or whether they calculate values for a grid of points, thus allowing the coverage of larger areas and not just single sites. This is completely neglected, though.
Author’s responses: In addition to the previously reported information, we now include further details in Table 2 on whether each model runs at a single point scale or grid scale resolution, and we report the grid scale resolution of each model.Referee suggestions: Spinup time (lines 155): Here many apples are compared with a few oranges. Generally, carbon pools have turnover timescales of years for litter pools, decades for reactive soil carbon, centuries for non-reactive / recalcitrant soil carbon and millennia for specific peatland carbon pools. Thus, the different spinup timescales reported for the different model are determined by the carbon pools considered in the model. Unfortunately, this is not discussed in the manuscript, but the authors only report the different spinup times given by the model authors. This is highly misleading: As an example, the HIMMELI model only reported the equilibration times for the methane specific components of the model. The HIMMELI model, however, needs to be run within a land-surface / carbon cycle model, as HIMMELI doesn’t deal with any of the relevant land surface physics or carbon calculations. Spinup times will therefore be determined by the host carbon cycle model, not by HIMMELI itself.
Author’s responses: Thank you very much for the constructive suggestions.- In addition to the previously reported information in Table 2, we add new information for each model on the turnover times of different soil carbon and litter pools. For example, in the case of the TRIPLEX-GHG (Zhu et al., 2014), the soil biogeochemistry module is derived from the Integrated Biosphere Simulator (IBIS), where the turnover time of the active carbon pool ranges weeks to months; slow carbon 10-30 years and recalcitrant carbon pool has 1000 years (Delire et al., 2003). Meanwhile CLM-Microbe has three litter pools: litter 1 (labile), litter 2 (cellulose), and litter 3 (lignin) having a turnover time of 20 hours to 71 days and four soil organic matter pools ranging from low to high recalcitrance having a turnover time of 14 days to 27 years (He et al., 2024). Such information is reported for each model in Table 2. We think that this kind of information is a valuable resource to assist future model users, who otherwise would be conducting their own desk studies and putting a great amount of effort into gathering this information.
- In Table 2, we add that if HIMMELI is run independently, then it takes 7 years to stabilize CH4 concentrations, however the actual time required for carbon pool stabilization will depend upon the type of land surface model coupled to HIMMELI.
Referee suggestions: Much more interesting is the question whether an accelerated spin up procedure is available – this is not reported, though.
Author’s responses: Thank you for these constructive suggestions. In Table 2, we now report in detail, based on published literature, the accelerated decomposition spin-up times for each model. For example, the accelerated decomposition spin-up time in LPJWhyMe (Wania et al., 2009a; Wania et al., 2009b; Wania et al., 2010) takes 10 years of climatology data which is repeated 100 times to obtain 1000 years for the C stock to attain equilibrium for non-peatlands and permafrost. But for peatlands, a spin-up of 90000 years plus 102-year transient runs required to stabilize peat C plus acceleration of 0.5 mm d−1 required to avoid large fluctuations in vegetation due to climatic oscillations.Referee suggestions: Process representation evaluations, lines 383 – 417: The manuscript mentions “full to adequate process representation” for plant mediated transport, diffusion, and ebullition. These are, however, not clarified, and to my knowledge there is no community consensus what might constitute “full” or “adequate” process representation. Thus, the manuscript passes judgement on models (“only 25% of the models exhibited full…”) without adequately defining the criteria.
Author’s responses: Thank you for the constructive suggestions.- We now clarify further in the text that we set these designations simply for the sake of discussing the extent of process representation in each model. These designations are not intended to rank or evaluate the models, but rather to help a new user decide which model would be best for them based on the number of processes represented in each model.
- We now add the detailed description of what the different categories mean in the manuscript text, and it reads as follows “ The “no process representation (category 0)” implies that the specific peatland or wetland model does not incorporate any processes or mechanisms simulating CH4 production, oxidation and transport. In case of the minimal process representation (category 1), a specific peatland or wetland model exhibits simplified representation of CH4 fluxes without quantifying in detail the different CH4 production, and oxidation pathways, while the transport process is only described using rate coefficients. The intermediate process representation (category 2) incorporates some degree of CH4 production and oxidation, while CH4 transport is described based on rate coefficients and CH4 concentrations supporting bubbling, minimum and threshold CH4 concentrations and vegetation specific CH4 transport and oxidation factors. The adequate process representation (category 3) quantifies in detail the different CH4 production, oxidation and transport pathways (Zhuang et al., 2004; Zhu et al., 2014; Salmon et al., 2022), while the full process representation (category 4) quantifies detailed microbial CH4 production and oxidation processes (Grant and Roulet, 2002; Xu et al., 2015; Wang et al., 2019; Ricciuto et al., 2021).
Referee suggestions: Advantages and disadvantages of reviewed models (lines 446 – 494): In these sections, the authors seem to summarize whatever the authors of the original model description papers gave as advantages and disadvantages, it is unclear whether these sections contain significant analysis by the manuscript authors. It is also doubtful whether the same criteria were applied to all models. One example from the disadvantages section: “… while the HIMMELI does not simulate any electron acceptors , except O2.” What is not mentioned in this sentence (or anywhere else) is that non-O2 electron acceptors are mostly handled by an unspecific bulk term, if at all, and very rarely, if ever, explicitly. Thus the “disadvantage” of the HIMMELI model is mainly that they actually reported that their treatment is not perfect, while other authors glossed over this fact. Thus, this analysis is incomplete at best and potentially misleading.
Author responses: The points mentioned below are now added in the manuscript.- Since the advantages and disadvantages of each model are solely derived from the published literature, this does not reflect the actual capabilities or limitations of each model and lacks a comprehensive model evaluation emphasizing the need for an in-depth model review using a common assessment criterion.
- Also, model developers are constantly testing their models via new data sets or conducting sensitivity analysis from time to time and updating to new versions that might be on the verge of being reported publicly or may be in the peer-review process. So, considering all the above-mentioned points, we re-named the model advantages as “model benefits” and model disadvantages or limitations as “model challenges” with the goal that these can be utilized by future users as the initial model screening points.
Citation: https://doi.org/10.5194/egusphere-2024-3852-AC2
-
AC2: 'Reply on RC2', Amey Tilak, 19 Feb 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 429 | 77 | 13 | 519 | 73 | 13 | 11 |
- HTML: 429
- PDF: 77
- XML: 13
- Total: 519
- Supplement: 73
- BibTeX: 13
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 139 | 27 |
| United Kingdom | 2 | 103 | 20 |
| Ireland | 3 | 57 | 11 |
| Germany | 4 | 49 | 9 |
| China | 5 | 36 | 7 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 139