the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Oct 2023
Reviews and Syntheses: Evaluating the Potential Application of Ecohydrological Models for Northern Peatland Restoration: A Scoping Review
Mark G. Healy
Laurence Gill
Abstract. Peatland restoration and rehabilitation action has become more widely acknowledged as a necessary response to mitigating climate change risks and improving global carbon storage. Peatland ecosystems require restoration timespans on the order of decades and thus cannot be dependent upon the shorter-term monitoring often carried out in research projects. Hydrological assessments using geospatial tools provide the basis for planning restoration works as well as analysing associated environmental influences. “Restoration” encompasses applications to pre- and post-restoration scenarios for both bogs and fens, across a range of environmental impact fields. The aim of this scoping review is to identify, describe, and categorise current process-based modelling uses in peatlands in order to investigate the applicability and appropriateness of eco- and/or hydrological models for northern peatland restoration. Two literature searches were conducted using the Web of Science entire database in September 2022 and August 2023. Of the final 211 papers included in the review, models and their applications were categorised according to this review’s research interests in 7 distinct categories aggregating the papers’ research themes and model outputs. Restoration site context was added by identifying 234 unique study site locations from the full database which were catalogued and analysed against raster data for the Köppen-Geiger climate classification scheme. A majority of northern peatland sites were in temperate oceanic zones or humid continental zones experiencing snow. Over one in five models from the full database of papers was unnamed and likely single-use. The top three most-used of these models, based on the frequency of their use on distinct site locations, were LPJ, ecosys, and DigiBog, in that order. Key themes emerging from topics covered by papers in the database included: modelling restoration development from a bog growth perspective; the prioritisation of modelling GHG emissions dynamics as a part of policymaking; the importance of spatial connectivity within or alongside process-based models to represent heterogeneous systems; and the emerging prevalence of remote sensing and machine learning techniques to predict restoration progress with little physical site intervention. This review provides valuable context for the application of ecohydrological models in determining strategies for peatland restoration and evaluating post-intervention development over time.
- Preprint
(1517 KB) - Metadata XML
-
Supplement
(504 KB) - BibTeX
- EndNote
Mariana P. Silva et al.
Status: final response (author comments only)
-
RC1: 'Comment on bg-2023-167', Anonymous Referee #1, 28 Oct 2023
Considering the long timescales required for peatland restoration and rehabilitation, modeling is critical for planning restoration and analyzing potential environmental impacts. This topic of this review is highly relevant for Biogeosciences. This review provides a comprehensive assessment of the current status of application of ecohydrological models on restoration of bogs and fens in northern regions considering a wide variety of potential environmental impacts. The authors clearly state their approach to identifying suitable studies and synthesize the results of these studies in a very useful manner. The results are based on 234 unique study sites which represents a substantial dataset. The authors identify the most widely used models, including LPJ, ecosys, and DigiBog. Much of the emphasis of the modeling exercises is on GHG emissions. The review highlights the emergence of remote sensing and machine learning to assess the progress of restoration with minimal intervention at sites. The figures and summaries provided in the review are very informative. This is an important review to provide context for peatland restoration.
This review is very timely as there are increasing efforts to restore and rehabilitate peatlands. The review synthesizes the models that have been applied to assess peatland restoration using several different categories. The conclusions are significant and important to understand the current and evolving status of this field of process-based modeling applied to peatlands in northern regions. The interpretations and conclusions are supported by the results of the review. The authors include detailed descriptions of some representative studies. The title is informative and reflects what is presented in the review. The abstract is excellent and summarizes the study clearly. The paper is clearly written. The review is concise and should not be shortened.
Citation: https://doi.org/10.5194/bg-2023-167-RC1 -
AC1: 'Reply on RC1', Mariana Silva, 07 Dec 2023
Thank you very much for your positive interpretation of our work. In reference to your note that much of the modelling emphasis is on GHG emissions, we can hopefully reflect on this more not only in justifying the prevalence of GHG simulations in the context of restoration, but also in addressing potential biases.
Citation: https://doi.org/10.5194/bg-2023-167-AC1
-
AC1: 'Reply on RC1', Mariana Silva, 07 Dec 2023
-
RC2: 'Comment on bg-2023-167', Anonymous Referee #2, 19 Nov 2023
General comments:
The paper by Silva et al. provides a timely and much needed review of global modelling efforts as applied to peatland ecohydrology in a restoration context. This has the potential to be an excellent resource for future peatland modellers. While the abstract, introduction, methodology, and results are well-written, concise, and valuable, the discussion section offers little of value and is unfocused. The discussion section is rife with speculative statements, which as far as I can tell are unsupported assertions by the authors on the “likelihood” of future model developments and work that will be conducted. I personally do not feel that the discussion adds substance to the article, and should be reorganized and rewritten. Two of the six figures are related to the spatial relationship between published peatland restoration studies and the Köppen-Geiger climate regions, while this has the potential to be an interesting feature of this review, it is not mentioned at all in the discussion section. The authors do not attempt to explain the patterns that the see, which perhaps are a spatial reflection of peatland prevalence, historical degradation, and being located within countries with a legacy of environmental stewardship and scientific funding. However, it is unclear what meaningful relationships can be derived from this analysis when the authors state that they deliberately placed an emphasis on NW Europe, as it is of particular interest to them. The remaining four figures are not well-integrated into the manuscript and do not contribute anything of substance. I would recommend that they be removed and perhaps replaced with a compressed version of Table S1.
Arguably, the key themes that the authors identified as “emerging” reflect those that have already been established and firmly rooted in the contemporary scientific zeitgeist, and do not reflect truly emerging trends. The exception to this is the discussion on machine learning techniques. This does not represent a critical flaw but rather a missed opportunity to talk about where peatland science and modelling will go over the next decade.
My suggestion for the discussion section would be to categorize models by level of process complexity, and not restrict the discussion to (largely) ecosys, CoupModel, and DigiBog. The models HYDRUS, Hydrogeosphere, and MODFLOW (SURFACT) incorporate the equations governing unsaturated flow and transport. While this has the potential to reveal detailed process-based insights into peatland function and recovery, they tend to have prohibitive data requirements for most projects. In contrast, DigiBog has strengths that these complex finite-difference/element models lack being able to simulate peatland development on a far longer time scale (centuries-millennia). By organizing the discussion section along the spectrum of process complexity I think this review will be more valuable and integrate a larger number of studies and approaches, each with their own strengths and weaknesses.
While I feel that this paper aligns with the aims and scope of Biogeosciences, the discussion section should be improved before the manuscript is accepted.
Specific comments:
L33: What is a “governmental scale”? That covers everything from an individual municipality to a national effort.
L78: Two decades on from Belyea and Baird, the representation of feedbacks across spatiotemporal scales still proves a challenge – although some progress has been made, see Waddington et al. (2015)
Waddington, J. M., Morris, P. J., Kettridge, N., Granath, G., Thompson, D. K., & Moore, P. A. (2015). Hydrological feedbacks in northern peatlands. Ecohydrology 8(1), 113-127.
L232 The rationale for claiming some of these models have more applicability to northern peatland restoration is not clear to me. Furthermore, I’m not sure that I agree that some models have more or less relevance in this regard, they are simply used to understand different things.
L244 “most natural” as in the largest number of processes?
L245 According to S1, Putra et al. (2022) is a 2D model developed in DigiBog. Although I can see in Figure 7 that it is in fact 3D, however the model presented in Putra et al. (2022) is nearly axisymmetric and is not the most illustrative example of the value and capabilities of 3D modelling. Sutton and Price (2022), and Zi et al. (2016) are 3D.
L256 Although excellent models, the Melaku et al. (2022) paper seems to have only tangential relevance to restoration, similar to Hwang et al. (2018).
L261 Is direct coding required for anything other than process modification? Please expand on the “etc.”
L268 This is incorrect and not the definition of the acrotelm. Besides being a potentially outdated conceptual framework (see Morris et al., 2011), the acrotelm comprises more than the “living layer” in bogs and includes partially and moderately decomposed plant matter (although plant roots can traverse the boundary between acrotelm and catotelm - as there is often more than just Sphagnum mosses living in peatlands). Furthermore, while the catotelm may comprise more well-decomposed peat with typically lower hydraulic conductivity, it is not hydrologically inactive and performs crucial hydrological functions in peatlands.
Morris, P. J., Waddington, J. M., Benscoter, B. W., & Turetsky, M. R. (2011). Conceptual frameworks in peatland ecohydrology: looking beyond the two‐layered (acrotelm–catotelm) model. Ecohydrology 4(1), 1-11.
L272-274 This is just my opinion, but I think more restraint should be exercised when recommending the use of models to practitioners that may not be aware of the myriad assumptions, limitations, and caveats that can apply to a model. That has the potential to cause more harm than good.
L277 This may not be germane to the overall paper, but it should be noted that fen-bog transitions can also be instigated quickly due to sudden drops in water table.
L277-279 These two sentences are rather hypothetical/speculative and unnecessary.
L297-298 One of the consequences of the organization of the discussion section (and focus on the Digibog, CoupModel, and ecosys) is that important developments are portrayed as unanswered open questions. The work of McCarter and Price have investigated this in a peatland restoration context using models.
L390 This is written as though this is unknowable information, but surely this is made clear within the paper, if not perhaps it is not the best example to use.
L394 The concept of an HRU does not inherently limit the resolution that different wetlands could be represented, many semi-distributed hydrologic models (ex. Raven) can have an arbitrarily large number of HRU, such that they begin to more closely resemble a fully spatially-distributed model.
L414-415 What makes it likely that a 13 year old paper will have further modelling occur if it has not already? Why speculate in this manner? There may be better examples to draw upon to describe the use of 3D modelling in a peatland restoration context.
Technical corrections:
L142: Missing the word “is”
Citation: https://doi.org/10.5194/bg-2023-167-RC2 -
AC2: 'Reply on RC2', Mariana Silva, 07 Dec 2023
The paper by Silva et al. provides a timely and much needed review of global modelling efforts as applied to peatland ecohydrology in a restoration context. This has the potential to be an excellent resource for future peatland modellers. While the abstract, introduction, methodology, and results are well-written, concise, and valuable, the discussion section offers little of value and is unfocused. The discussion section is rife with speculative statements, which as far as I can tell are unsupported assertions by the authors on the “likelihood” of future model developments and work that will be conducted. I personally do not feel that the discussion adds substance to the article, and should be reorganized and rewritten.
[Response: Thank you for this suggestion. I do feel there is still value in the current presentation and content of the discussion section. Each of the subsections stand to represent commonalities in model outputs which link to different aspects of peatland restoration, be they mechanical, hydrological, or ecological, in the present or in long-term time scales. Indeed, it was in compiling such a repository of peatland and peatland restoration-related modelling research that these categories surfaced organically throughout the methods process. Knowledge gaps are identified for each of these themes, especially regarding potential further modifications of existing models.
But insofar as is possible, I will attempt to make all statements more definitive, and carefully evaluate the levels of speculation I bring to the table.]
Two of the six figures are related to the spatial relationship between published peatland restoration studies and the Köppen-Geiger climate regions, while this has the potential to be an interesting feature of this review, it is not mentioned at all in the discussion section. The authors do not attempt to explain the patterns that the see, which perhaps are a spatial reflection of peatland prevalence, historical degradation, and being located within countries with a legacy of environmental stewardship and scientific funding. However, it is unclear what meaningful relationships can be derived from this analysis when the authors state that they deliberately placed an emphasis on NW Europe, as it is of particular interest to them. The remaining four figures are not well-integrated into the manuscript and do not contribute anything of substance. I would recommend that they be removed and perhaps replaced with a compressed version of Table S1.[Response: I agree that less emphasis is placed on the geographical contextual information as the review progresses. The initial approach was to present climate regions as global context before “zooming in” to NW Europe. I will simultaneously briefly address the patterns of peatland research locations globally like you suggest while also paring down slightly the amount of information I include about geographical context in the results, so this section is more balanced.]
Arguably, the key themes that the authors identified as “emerging” reflect those that have already been established and firmly rooted in the contemporary scientific zeitgeist, and do not reflect truly emerging trends. The exception to this is the discussion on machine learning techniques. This does not represent a critical flaw but rather a missed opportunity to talk about where peatland science and modelling will go over the next decade.[Response: this is certainly something that could use additional reflection – and I will take more care as to the usage of terms like “emerging” to ensure the reflect my meaning throughout the paper, while also acknowledging that what is seen as “established” in modelling could still be valuable to engineers and planners because the approaches come with precedence and protocol – so what I may be calling “emerging” from the history of enacting restoration these past decades is more of a reflection of model usage tendencies.]
My suggestion for the discussion section would be to categorize models by level of process complexity, and not restrict the discussion to (largely) ecosys, CoupModel, and DigiBog. The models HYDRUS, Hydrogeosphere, and MODFLOW (SURFACT) incorporate the equations governing unsaturated flow and transport. While this has the potential to reveal detailed process-based insights into peatland function and recovery, they tend to have prohibitive data requirements for most projects. In contrast, DigiBog has strengths that these complex finite-difference/element models lack being able to simulate peatland development on a far longer time scale (centuries-millennia). By organizing the discussion section along the spectrum of process complexity I think this review will be more valuable and integrate a larger number of studies and approaches, each with their own strengths and weaknesses. While I feel that this paper aligns with the aims and scope of Biogeosciences, the discussion section should be improved before the manuscript is accepted.[Response: I'm happy to do this. I want to retain some element of the methodological process in the discussion (i.e., the different modelling ends which could be applied in a restoration context), but both can likely be done. Thanks so much for this thorough review and the very constructive criticisms.
All the best. -- MPS]
Specific comments:
L33: What is a “governmental scale”? That covers everything from an individual municipality to a national effort.
[Response: the word “scales” may be replaced with “spheres” to help clarify my point here. I am trying to distinguish between the “spheres” in which restoration is governed – chiefly, the legislation and policy allowing it to go forward (national and municipal alike), compared with the technical planning carried out by industry (consultants, engineering firms, etc.) for its direct enactment.]
L78: Two decades on from Belyea and Baird, the representation of feedbacks across spatiotemporal scales still proves a challenge – although some progress has been made, see Waddington et al. (2015)
Waddington, J. M., Morris, P. J., Kettridge, N., Granath, G., Thompson, D. K., & Moore, P. A. (2015). Hydrological feedbacks in northern peatlands. Ecohydrology 8(1), 113-127.
[Response: an additional note about progress will be added to ensure accurate context, thank you.]
L232 The rationale for claiming some of these models have more applicability to northern peatland restoration is not clear to me. Furthermore, I’m not sure that I agree that some models have more or less relevance in this regard, they are simply used to understand different things.
[Response: This comment is valuable, thank you. I will use this as guidance for restructuring the discussion with a less subjective focus.]
L244 “most natural” as in the largest number of processes?
[Response: I mean to say that these models incorporate the largest number of processes representing “life” in a peatland, which is a bit qualitative and not easily definable. I want to include processes like vegetation growth, organic/inorganic transformation, and microbial processes. However, it is also true that these models do use the largest number of processes. Grant et al. in 2017 call ecosys “process-rich”, or deterministic, which might be what I was going for.]
L245 According to S1, Putra et al. (2022) is a 2D model developed in DigiBog. Although I can see in Figure 7 that it is in fact 3D, however the model presented in Putra et al. (2022) is nearly axisymmetric and is not the most illustrative example of the value and capabilities of 3D modelling. Sutton and Price (2022), and Zi et al. (2016) are 3D.
[Response: this sentence indeed has an error. I will clarify that Putra et al.’s model result is pseudo-3D and in DigiBog, and doesn’t quite represent the full capabilities of 3D modelling. For the case of ecosys, I neglected to include Grant et al.’s paper in 2017 which uses the full 3D coupling capabilities of ecosys. I will also consider including your additional citations – especially Sutton and Price (2022) – to exemplify the capabilities of 3D modelling.]
L256 Although excellent models, the Melaku et al. (2022) paper seems to have only tangential relevance to restoration, similar to Hwang et al. (2018).
[Response: I had intended to bring attention to these tangential references, however, to explore the possibilities of these newer models being incorporated more into restoration research. For example: comparing scenarios of emissions is a certain percentage of area in a basin is “restored” (i.e., parameters forced in a shorter time to represent wetter and eventually more vegetated conditions). This could be of particular interest to fields outside of modelling science research, especially policy-making. I will expand my discussion of this point, and disclaim that their current applicability is only in the theoretical stage.]
L261 Is direct coding required for anything other than process modification? Please expand on the “etc.”
[Response: sentence will be restructured, especially to avoid the term “expectation”, and instead stress that a shared open-source coding language can benefit the wider use of the models or the stitching together of routines or subroutines from separate models can facilitate more seamless combinations – along with easing the modification/addition/removal of processes.]
L268 This is incorrect and not the definition of the acrotelm. Besides being a potentially outdated conceptual framework (see Morris et al., 2011), the acrotelm comprises more than the “living layer” in bogs and includes partially and moderately decomposed plant matter (although plant roots can traverse the boundary between acrotelm and catotelm - as there is often more than just Sphagnum mosses living in peatlands). Furthermore, while the catotelm may comprise more well-decomposed peat with typically lower hydraulic conductivity, it is not hydrologically inactive and performs crucial hydrological functions in peatlands.
Morris, P. J., Waddington, J. M., Benscoter, B. W., & Turetsky, M. R. (2011). Conceptual frameworks in peatland ecohydrology: looking beyond the two‐layered (acrotelm–catotelm) model. Ecohydrology 4(1), 1-11.
[Response: will remove contents of parentheses and instead cite your suggestion. I will reference the term as a way to show where models are progressing from (retaining the words “changes occurring in” and adding “past designations of …”, to introduce what Morris et al. address in their DigiBog paper also published in 2011.]
L272-274 This is just my opinion, but I think more restraint should be exercised when recommending the use of models to practitioners that may not be aware of the myriad assumptions, limitations, and caveats that can apply to a model. That has the potential to cause more harm than good.
[Response: this is a valid concern. Will rephrase the sentence to focus more on the ease of execution (with fewer inputs and access to source code for greater context rather than claiming no background knowledge is needed.]
L277 This may not be germane to the overall paper, but it should be noted that fen-bog transitions can also be instigated quickly due to sudden drops in water table.
[Response: see below.]
L277-279 These two sentences are rather hypothetical/speculative and unnecessary.
[Response: will simplify only to point out that intensive restoration works will also bring about abrupt peatland change like what you mention above, e.g., Malloy and Price, 2014, and these abrupt changes should be considered when considering model applicability.]
L297-298 One of the consequences of the organization of the discussion section (and focus on the Digibog, CoupModel, and ecosys) is that important developments are portrayed as unanswered open questions. The work of McCarter and Price have investigated this in a peatland restoration context using models.
[Response: I had not encountered Gauthier, McCarter, and Price’s use of HYDRUS 1-D in 2018, if this is what you are citing – however, it appears to be applicable and will be incorporated post-hoc]
L390 This is written as though this is unknowable information, but surely this is made clear within the paper, if not perhaps it is not the best example to use.
[Response: I was hoping to acknowledge the notable differences between oil sands mining and peat extraction, which could require further modification of the new wetlands subroutine for additional peatland management contexts. It will be simplest I think to remove this statement entirely.]
L394 The concept of an HRU does not inherently limit the resolution that different wetlands could be represented, many semi-distributed hydrologic models (ex. Raven) can have an arbitrarily large number of HRU, such that they begin to more closely resemble a fully spatially-distributed model.
[Response: will rephrase “HRUs” to “the number of currently incorporated HRUs”.]
L414-415 What makes it likely that a 13 year old paper will have further modelling occur if it has not already? Why speculate in this manner? There may be better examples to draw upon to describe the use of 3D modelling in a peatland restoration context.
[Response: will change the term “likely/will” to “possible/can” – or, I can perhaps remove this phrase and end the sentence after “effectiveness”. I don’t want to discount the potential for older models to still have relevance and usefulness in current applications, though I agree that I should not allude to future use of the model if it has not already happened.]
Technical corrections:
L142: Missing the word “is”
[Will be rectified, thanks.]Citation: https://doi.org/10.5194/bg-2023-167-AC2
-
AC2: 'Reply on RC2', Mariana Silva, 07 Dec 2023
Mariana P. Silva et al.
Mariana P. Silva et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 156 | 66 | 10 | 232 | 23 | 6 | 7 |
- HTML: 156
- PDF: 66
- XML: 10
- Total: 232
- Supplement: 23
- BibTeX: 6
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1