the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Feb 2025
Process-based modelling of multi-decade carbon dynamics of a cool temperate swamp
Abstract. Swamps are important wetlands globally, but temperate swamps have been understudied even though they store substantial quantity of carbon (C) in their biomass and can accumulate peat. This stored C supports their role as nature-based solutions in climate change mitigation efforts. In particular, Southern Ontario swamps are estimated to store ~1.1 Pg C under distinct hydroclimatic conditions. Previous studies on temperate swamp C fluxes are mostly based on short-term (<5 years) field measurements that limit our understanding of the multi-decadal dynamics that exist between this ecosystem’s C flux and biophysical conditions. To elucidate the long-term interactions and feedbacks that are important to temperate swamp C dynamics, we adopted a process-based model (CoupModel) to simulate daily plant processes, energy, water and C fluxes in one of the most well-preserved swamps in Southern Ontario over a 40-year period (1983–2023). CoupModel reasonably simulated the C flux and controlling variables with coefficient of determination (R2) values of 0.60, 0.95 & 0.61 for soil respiration, surface soil temperature (0–5cm) and water table level respectively when validated with field measurements. Over the simulation period, the swamp’s C uptake capacity as net ecosystem exchange declined but it maintained a net C sink in most years. This declining trend can be attributed to a consistent rise in soil respiration (11 % per decade) that is likely to continue with future climate change predictions. Overall, the study shows that processed-based models are effective tools for improving our understanding of long-term C dynamics of temperate swamps.
- Preprint
(3050 KB) - Metadata XML
-
Supplement
(367 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-4049', Anonymous Referee #1, 18 Apr 2025
-
AC1: 'Reply on RC1', Oluwabamise Afolabi, 26 May 2025
GENERAL REMARKS
We thank the reviewers for taking out the time to read our submitted manuscript and sharing important feedback on it.
Even though we have responded to the reviewer’s specific feedback below, we would like to highlight that our submitted manuscript is one of the first attempts to model the long-term carbon dynamics of a temperate swamp peatland and analyze its controls on soil respiration. Currently, temperate swamp peatlands systems are understudied and very few data exist on this ecosystem (Davidson et al., 2022). Our study site, Beverly Swamp included data from the 1980s, short studies from 1998 – 2000 by Davidson et al (2019), McCarter et al. (2024) and resumed measurements from 2022. Thus, Beverly Swamp to our knowledge, represents one of the most well-studied temperate swamp peatlands. The long-term coverage of empirical data makes long term modeling study as ours possible, however, these data collections were in some cases not continuous and were done by different groups with varying methods over a 40+ year period. Thus, there are considerable uncertainties associated with these measurements. This aspect was mentioned in the current version but probably not discussed in depth. We will add those uncertainties in our revision and discuss the extent that the model data discrepancies can be explained by the measured data, model structure and parameters.
However, we would also like to mention that the main purpose of this paper is not to discuss those uncertainties or model performance in details but rather to first test whether or not a processed-based model can simulate the coupled hydrological, heat and carbon dynamics of a temperate swamp peatland system that has not been simulated so far. Furthermore, we undertook a systematic uncertainty analysis by using the Generalized Likelihood Uncertainty Estimation (GLUE) approach to identify the errors associated with the different aspects (e.g., measurement, parameter and model structure) of the modelling exercise. This uncertainty analysis and other associated experiments have been submitted as a separate manuscript to the Geoscientific Model Development (GMD) journal. Therefore, this means that some of the Reviewer’s feedbacks were also addressed in the latter manuscript. Although this first manuscript lacks the global sensitivity analysis, our goal was to investigate the effectiveness of CoupModel for characterizing temperate swamp thermal, hydrological and carbon cycling processes and provide first estimates of modelled swamp carbon balance. We believe this is already an important contribution to literature and of interest to the readers of Biogeosciences Journal.
In addition, the reviewers highlighted some concerns about the presentation quality of the paper. Some of the comments (e.g., model performance) were partly due to the figure presentations (e.g., mistakes in uploading our soil respiration figure). We mistakenly repeated the water table level figure (figure 5) instead of soil respiration in Figure 9. The journal editors were notified of this mistake, but we were told to update it during the next review stage. We sincerely apologize for the mistake in our initial uploading of the manuscript. We will fix this error in our revision, but we have also attached the figure. Furthermore, we will revise the texts (clarification of model calibration, parameter determination, expansion of the result analysis) and improve all figures to substantially improve the presentation.
REVIEWER’S COMMENT
This manuscript presents a process-based modelling study of carbon dynamics in a temperate swamp, based on extensive field measurements collected over a 40-year period. If the model demonstrates successful calibration and validation against consistent observations of abiotic variables (soil temperature, soil moisture, water table level, and heat fluxes) and carbon fluxes (soil respiration and GPP), the study could make a substantial contribution to our understanding of the long-term responses of ecosystem functioning—specifically water and carbon exchange—to recent climate variability, including both seasonal and interannual changes. It also has the potential to identify the key environmental drivers involved.
However, the current version of the manuscript does not meet the quality standards required for Biogeosciences, primarily due to the lack of sufficient detail regarding model calibration of carbon fluxes and poor model performance. I recommend that the authors address the following aspects:
Presentation of Results Figure 1: Please provide more detailed information on the specific measurements collected at each sampling site.
AUTHOR'S RESPONSE
We will ensure the measurement stations are properly displayed on the map of Beverly Swamp (Figure 1), and revise the texts in the method section accordingly to reflect exact measurement point and period of data collection.
We should also include that the details of the datasets used for this study were extensively described in the manuscript. You will observe that the data sources used for driving CoupModel, calibration and validation purposes were thoroughly described on pages 4 to 6 (Lines 96 to 142) of the manuscript. The driving (climatic) variables namely, air temperature, precipitation, relative humidity, global radiation and windspeed were sourced from nearby weather stations and additional sources described on lines 102 to 109, while lateral water input data was collected from nearby gauging station (Line 110 and Figure 1). We measured hourly water table levels across the northern transect of the swamp (see figure 1) from June 2022 to July 2023 with Solinst leveloggers, while a Campbell data logger was installed to log hourly datasets of soil temperature and volumetric moisture contents at different depths for the same period. The specific measurements points are also highlighted in Figure 1 (Map of Beverly swamp). Also see Lines 126 to 142.
Furthermore, energy fluxes (net radiation, ground heat, sensible heat), soil temperate (0-5cm), water table level and snow depth measurements from August 1983 to July 1987 were sourced from Munro et al. (2000) for the calibration period of the modeling study. See lines 115 to 119 for detailed description and sample collection point. Soil respiration and soil organic carbon data used for model initialization, calibration (1998 – 2000) and validation (2022 – 2023) were sourced from Davidson et al., 2019, McCarter et al., 2024 and Schmidt and Strack, 2024 (See lines 120 to 125). Details of other measurements used for estimating model parameters values can be found in Appendix A (Table A1).
- Figure 2: Why is ground heat shown at the bottom of the soil profile? It should be included as part of the surface energy balance. Additionally, if the model includes two tree species, the authors should explain how multiple canopies are represented. I suggest showing parameters such as species-specific height, root depth, leaf area index (LAI), and litterfall for both trees and peatland vegetation.
AUTHOR'S RESPONSE
We realize that the presentation of Figure 2 might cause some misunderstanding but the ground heat flow should not be confused with surface soil heat flux. The surface soil heat flux is the residual energy flux of the surface energy balance that is explicitly simulated in CoupModel (see equations in the appendix), however, the ground heat flux illustrated in figure 2 delineates the lower boundary conditions of the model for heat conduction at Beverly swamp (see lines 174 and 175 for description). We will revise the figure and include additional caption to explain that the ground heat flux is for simulating soil temperature.
Beverly swamp’s forest cover consists of both coniferous and deciduous species (plant functional types, PFTs) with almost equal standing biomass (Munro 1979). Therefore, the reported lumped PTFs characteristics of existing studies (e.g., Munro et al, 2000) were used to parameterize plant properties in CoupModel(one-dimensional model). In addition, the vegetation of the swamp in the CoupModel was simulated by an “explicit bigleaf” and a single representative canopy layer characterized by the closed canopy structure (LAI of 5-6 that was simulated dynamically, which makes it vary with seasons) and root depth of ~30cm. We did not simulate understory vegetation since this category is low and thus insignificant for the simulated carbon and hydrology fluxes. The specific characteristics / parameter values of the vegetation cover at Beverly Swamp are listed in Table A.1 (Appendix A) of the submitted transcript.
If a revision is requested, we will clarify aspects of Figure 2 further in the caption and ensure that all model description is clear to address the reviewer’s comments.
- Figure 3: What does the red line represent? Is it a moving average?
AUTHOR'S RESPONSE
The red line in figure 3 shows the trend direction of precipitation and air temperature. In particular, it is a function (LOWESS) in the R statistical software (Kendall library) that smoothens out data through the fitting of local regression. Captions will be updated in the revision to state this clearly.
- Figures 4a, 4b, and 5a: Please clarify the meaning of the red line and the grey shaded area—do these represent a regression line and confidence interval, respectively?
AUTHOR'S RESPONSE
The red lines in figures 4a, 4b and 5a are regression lines added to the plot by a function (geom_smooth) in the R statistical software, while the grey band represents the confidence interval. Captions will be updated in the revision to state this clearly.
- Figures 8b and 9: The model appears to underestimate soil moisture and water table levels, while simulated evapotranspiration (ET) seems to be overestimated. I recommend revisiting the water balance and interpreting these model biases.
AUTHOR'S RESPONSE
We apologize for the mistake in our initial uploading of the manuscript. We think the problem in the referenced Figures (8b and 9) is not really about water balance. The main error originates from the mistake of repeating the water table level figure (figure 5) instead of soil respiration in Figure 9. We would fix this error in our next revision.
Furthermore, the limitations and uncertainties associated with simulated water balance components were extensively discussed on lines 380 to 404 of the manuscript. Please note that our study relied on water table level measurement for both the calibration (July 1983 – August 1987) and validation periods (2022 – 2023) but only had access to volumetric moisture content measurements (2022 – 2023) and ET (1983 – 1987) for calibration period. Consequently, we limited our water balance evaluation to available datasets. There is no clear evidence in Figure 8 that ET is overestimated by the model and in fact, the comparison of observed and simulated ET patterns was consistent with evaluated water table levels during the calibration period (1983 – 1987) when both ET and WTD measurements were readily available (please see attached figures 5 and 8 ). Nevertheless, we will elucidate this in the next revision to make it clearer.
- Missing Calibration for Ecosystem Respiration. The manuscript appears to lack a figure or discussion on the model calibration results for ecosystem respiration. This is a critical component for evaluating model performance and should be included.
AUTHOR'S RESPONSE
Again, we apologize for mistakenly repeating the water table level figure (figure 5) that is meant for soil respiration in Figure 9 (please see attached). We will correct the error by updating the soil respiration figure. However, we believe the reviewer is referring to soil respiration here and not ecosystem respiration since no eddy covariance flux measurement were done at Beverly Swamp.
Kindly note that the model calibration results of soil respiration were discussed on lines 414 to 426 (Page 18) of the submitted transcript. However, we can also add some sentences to further discuss the model data discrepancies of the soil respiration in our revision.
- Interannual Variability in Soil Respiration Soil respiration from June to September shows a positive trend over time, while GPP and air temperature do not show any clear trends. Can the authors explain what might be driving the increased soil respiration during the growing season?
AUTHOR'S RESPONSE
Even though the increasing air temperature trend between June and September was not as pronounced as those reported for winter and fall seasons, moderate rise in air temperature was also observed during this period (see Figure B1). A moderate increase in air-soil temperature would accelerate soil microbial activities, and drier conditions with extended oxic zone during the growing season may have supported a spike in aerobic decomposition of swamp peat at both surface and deep layers. Consequently, soil respiration followed an increasing trend (11%) over the study period. The warmer conditions may not have supported a spike in GPP because of other limiting factors that can affect productivity. For instance, excessive dryness during this period and nutrient limitations may have affected GPP rates in the swamp. In addition, our study did not consider the effect of CO2 fertilization which has been reported to increase GPP rates in the last decades. We will make sure that this discussion of trends is clear in the discussion if a revision of the manuscript is requested.
- Minor Comments
Abstract, line 17: If negative NEE is defined as a carbon sink, then a decline in NEE (i.e., becoming more negative) suggests increased carbon uptake. Please clarify this wording.
AUHTOR'S RESPONSE
Yes, we followed the convention of attributing a negative NEE to more carbon intake by the swamp and the vice versa. For the study period, the swamp’s net carbon intake dwindled as more carbon was lost via soil respiration and with little intake via GPP. We will make sure that the text is clear with the trend in any revision.
- Line 35: A comma is missing after “Meanwhile”.
AUHTOR'S RESPONSE
Thank you for the observation. A comma will be included next to the word.
- Line 45: "> 5 years" should be corrected to "< 5 years".
AUHTOR'S RESPONSE
Thank you, the observation is valid. “>5 years” would be rewritten as “<5 years”.
- Line 48: The phrase “capture the impact of regional multidecadal disturbances (e.g., climate change)” could be improved to “capture regional ecosystem responses to multidecadal climate change”.
AUHTOR'S RESPONSE
Statement will be rejigged in the revision.
- Line 67: Consider rephrasing to clarify: “simulate key plant processes that regulate water, energy, and carbon fluxes in temperate swamps.”
AUHTOR'S RESPONSE
Statement will be clarified.
- Line 385: Could this result be related to a bias in simulated evapotranspiration?
AUHTOR'S RESPONSE
There is little or no evidence to show that the high mismatch between simulated and observed water table level during the validation period comes from the bias in simulated evapotranspiration. Instead, the difference can be linked to very dry conditions (annual precipitation record of 616 mm) during the validation period. Because the model was mostly trained with wet years during the calibration period (1983 – 1986), it was not able to capture the extremely dry conditions presented in the validation period. We further confirmed this with an uncertainty analysis that was completed in another manuscript. We will more clearly highlight the abnormally dry conditions of the validation period in the revised manuscript.
- Lines 446–454: I suggest presenting the modelled LAI and linking it to the discussion in this section.
AUHTOR'S RESPONSE
It is possible to present the simulated LAI in the manuscript and further link it to the discussion. The comparison of our simulated and observed LAI produced a R2 of 0.81 but we deliberately did not dwell on this in the submitted manuscript since LAI is not the focus of our study. Nevertheless, we discussed it more extensively in a later manuscript that was submitted to the Geoscientific Model Development (GMD) journal. We can add a comparison of simulated and observed in the revised manuscript.
- Line 495: The decoupling phenomenon typically occurs under very dry or very wet conditions. The authors should support this discussion with simulated GPP or stomatal conductance results.
AUHTOR'S RESPONSE
Our result shows that the decoupling occurs under very dry and warm conditions. Simulated GPP can be found in Figure B5 (Appendix B). We will make sure this link is clear in the discussion as suggested by the reviewer.
- Lines 755–760: A reference appears to be duplicated
AUHTOR'S RESPONSE
The publications by Metzger et al. were not duplicated. We cited three different papers by Metzger et al. (2015, 2016a, 2016b) and a PhD thesis by Metzger 2015.
-
AC1: 'Reply on RC1', Oluwabamise Afolabi, 26 May 2025
-
RC2: 'Comment on egusphere-2024-4049', Anonymous Referee #2, 23 Apr 2025
The manuscript by Afolabi et al. presents a study evaluating the performance of the process-based model CoupModel V6.2.2 in simulating soil hydrothermal dynamics and soil CO₂ fluxes in a swamp ecosystem affected by seasonal reservoir flooding. This is an interesting topic, as modeling biogeochemical processes in wetland systems remains a significant challenge. I particularly appreciate the authors' effort to simulate such dynamics in a peatland disturbed by reservoir flooding, which is very helpful to the modeling community and offers a potentially valuable example of how process-based models can be applied to such condition. However, I believe the manuscript requires further improvement before it can be considered for publication. These improvements primarily relate to clarification of the methods, expansion of the result analysis, and enhanced data presentation. Below are my main suggestions and concerns:
Major Comments:
1. The swamp site is notably influenced by seasonal reservoir flooding. It would indeed be valuable if CoupModel could adequately capture the hydrological, biogeochemical, and biophysical processes under such disturbances. However, the model appears to fail in simulating key hydrological variables such as water table depth and deep soil moisture—despite the use of observed lateral water input. I recommend the authors conduct a further hydrological parameter sensitivity analysis. This may help identify additional factors contributing to model underperformance beyond the underestimation of latent heat flux.
Furthermore, while I am interested in the model's potential to simulate carbon dynamics under seasonal flooding, the current presentation of results (particularly Figure 9) fail to the assessment of soil respiration dynamics. Please consider revising Figure B8 to present these processes at daily or monthly resolution, which could offer a clearer picture of the model’s capacity to capture seasonal variations. An extended analysis on the impact of reservoir flooding on carbon fluxes and other processes would be especially valuable.
2. Previous studies, such as those by Metzger et al., have conducted extensive modeling work on peatlands using CoupModel. To better situate your work within this context, it would be helpful to clarify how your study builds on or differs from these earlier efforts. If the goal is only to emphasize modeling performance in temperate swamp condition, a more detailed discussion comparing model performance and setup across peatland types is encouraged.
3. Model Structure and Version-Specific Updates. Sections 2.3 and 2.4 describe CoupModel’s structure too much. However, given that CoupModel has been described extensively in prior literature (e.g., Jansson et al., 2013; He et al., 2020), I suggest focusing more on what is unique or updated in version 6.2.2, especially in relation to this study. For example, have any components or parameterizations been modified specifically for temperate swamp simulation? Please also include specific parameter values used for S1, such as van Genuchten parameters, and clarify any deviations from default settings.
4. Dataset Description and Temporal Resolution. The manuscript mentions the use of multiple datasets with varying frequencies and durations, but the descriptions lack clarity. I strongly recommend adding a summary table that categorizes all datasets (e.g., input, calibration, validation), including their temporal resolution, measurement duration, methodology, and target variables. Also, please clarify how the authors standardized temporal resolution where dataset frequencies differed.
5. Clarification of Model Calibration Process. Given that observation periods differ among soil thermal data, respiration, and energy fluxes, more detail is needed on the model calibration and prioritization process. For instance, which variables were given higher priority in the calibration? What parameter values were adjusted, and how do they compare to defaults or values used in other peatland studies?
6. Soil Water Flow Modeling Approaches. The manuscript mentions the use of both Darcy’s Law and a two-domain approach, but does not provide comparison results. The authors state that the two-domain method performed better for water table and moisture simulation. If this approach was ultimately used in the full simulation, please clarify how it influenced other processes, such as NEE and soil temperature. Including comparison plots would enhance transparency and understanding.
7. Figure 4 shows that the model fails to simulate soil temperatures below 0°C at both 5 cm and 30 cm, possibly due to forced zeroing. Could the authors provide a technical explanation for this outcome?
8. The current manuscript focuses mainly on observational trends over time. However, it would be insightful to evaluate how well the model captures interannual variability in key variables through direct model–observation comparison.
9. The clarity of Figures 4–9 should be improved. It is currently difficult to interpret the data—please specify whether these are daily, monthly, or annual values, and adjust formatting for readability.
10. Peatlands are also known methane sources. I am curious that has CoupModel been evaluated or configured for methane flux simulation in peatlands?
Minor Suggestions:1. Consider including “CoupModel” in the title, as it is the central modeling tool used.
2. Integrate Figures 6 and 7 into one composite figure for clarity.
3. There seems to be a misplaced plot in Figure 9; please check and correct.
4. Provide exact p-values alongside R² values for all statistical comparisons.
5. Clarify the default parameter values used in S2, and highlight any key changes made.
Citation: https://doi.org/10.5194/egusphere-2024-4049-RC2 -
AC2: 'Reply on RC2', Oluwabamise Afolabi, 26 May 2025
GENERAL REMARKS
We thank the reviewers for taking out the time to read our submitted manuscript and sharing important feedback on it. Even though we have responded to the reviewer’s specific feedback below, we would like to highlight that our submitted manuscript is one of the first attempts to model the long-term carbon dynamics of a temperate swamp peatland and analyze its controls on soil respiration. Currently, temperate swamp peatlands systems are understudied and very few data exist on this ecosystem (Davidson et al., 2022). Our study site, Beverly Swamp included data from the 1980s, short studies from 1998 – 2000 by Davidson et al (2019), McCarter et al. (2024) and resumed measurements from 2022. Thus, Beverly Swamp to our knowledge, represents one of the most well-studied temperate swamp peatlands. The long-term coverage of empirical data makes long term modeling study as ours possible, however, these data collections were in some cases not continuous and were done by different groups with varying methods over a 40+ year period. Thus, there are considerable uncertainties associated with these measurements. This aspect was mentioned in the current version but probably not discussed in depth. We will add those uncertainties in our revision and discuss the extent that the model data discrepancies can be explained by the measured data, model structure and parameters. However, we would also like to mention that the main purpose of this paper is not to discuss those uncertainties or model performance in details but rather to first test whether or not a processed-based model can simulate the coupled hydrological, heat and carbon dynamics of a temperate swamp peatland system that has not been simulated so far. Furthermore, we undertook a systematic uncertainty analysis by using the Generalized Likelihood Uncertainty Estimation (GLUE) approach to identify the errors associated with the different aspects (e.g., measurement, parameter and model structure) of the modelling exercise. This uncertainty analysis and other associated experiments have been submitted as a separate manuscript to the Geoscientific Model Development (GMD) journal. Therefore, this means that some of the Reviewer’s feedbacks were also addressed in the latter manuscript. Although this first manuscript lacks the global sensitivity analysis, our goal was to investigate the effectiveness of CoupModel for characterizing temperate swamp thermal, hydrological and carbon cycling processes and provide first estimates of modelled swamp carbon balance. We believe this is already an important contribution to literature and of interest to the readers of Biogeosciences Journal.
In addition, the reviewers highlighted some concerns about the presentation quality of the paper. Some of the comments (e.g., model performance) were partly due to the figure presentations (e.g., mistakes in uploading our soil respiration figure). We mistakenly repeated the water table level figure (figure 5) instead of soil respiration in Figure 9. The journal editors were notified of this mistake, but we were told to update it during the next review stage. We sincerely apologize for the mistake in our initial uploading of the manuscript. We will fix this error in our revision, but we have also attached the figure. Furthermore, we will revise the texts (clarification of model calibration, parameter determination, expansion of the result analysis) and improve all figures to substantially improve the presentation.
REVIEWER'S COMMENT
The manuscript by Afolabi et al. presents a study evaluating the performance of the process-based model CoupModel V6.2.2 in simulating soil hydrothermal dynamics and soil CO₂ fluxes in a swamp ecosystem affected by seasonal reservoir flooding. This is an interesting topic, as modeling biogeochemical processes in wetland systems remains a significant challenge. I particularly appreciate the authors' effort to simulate such dynamics in a peatland disturbed by reservoir flooding, which is very helpful to the modeling community and offers a potentially valuable example of how process-based models can be applied to such condition. However, I believe the manuscript requires further improvement before it can be considered for publication. These improvements primarily relate to clarification of the methods, expansion of the result analysis, and enhanced data presentation. Below are my main suggestions and concerns:
AUTHOR'S RESPONSE
Many thanks for the positive comments on the significance of our study. In our revision, we will substantially increase the presentation of the paper. Please also see our general remarks and specific responses below.
- Major Comments:
- The swamp site is notably influenced by seasonal reservoir flooding. It would indeed be valuable if CoupModel could adequately capture the hydrological, biogeochemical, and biophysical processes under such disturbances. However, the model appears to fail in simulating key hydrological variables such as water table depth and deep soil moisture—despite the use of observed lateral water input. I recommend the authors conduct a further hydrological parameter sensitivity analysis. This may help identify additional factors contributing to model underperformance beyond the underestimation of latent heat flux.
AUTHOR'S RESPONSE
We appreciate the comments, however, we disagree that the model failed to simulate key hydrological variables. The performance indices (e.g., R square statistics) of the simulated hydrological variables (lines 296 to 335) and evaluation discussions (lines 379 to 408) showed a reasonably good model – observation agreement despite all the uncertainty and limitations associated with the modelling experiment. Please see attached figure.
Nevertheless, we agree with the Reviewer that a robust sensitivity analysis may help improve our modelling. Consequently, we undertook an uncertainty of the modelling experiment using the Generalized Likelihood Uncertainty Estimation (GLUE) approach to identify the errors associated with the different aspects (measurement, parameter and model structure) of the modelling exercise. This uncertainty analysis and other associated modelling experiments has been submitted as a separate manuscript to the Geoscientific Model Development (GMD) journal. Part of the lower model fit for water table in the validation period also results from this being an abnormally dry year, a condition over which the model was not trained. We will highlight this further in the discussion in a revised version of the manuscript.
- Furthermore, while I am interested in the model's potential to simulate carbon dynamics under seasonal flooding, the current presentation of results (particularly Figure 9) fail to the assessment of soil respiration dynamics. Please consider revising Figure B8 to present these processes at daily or monthly resolution, which could offer a clearer picture of the model’s capacity to capture seasonal variations. An extended analysis on the impact of reservoir flooding on carbon fluxes and other processes would be especially valuable.
AUTHOR'S RESPONSE
We apologize for mistakenly repeating the water table level figure (figure 5) that is meant for soil respiration in Figure 9 (please see attached). We will correct the error by updating the soil respiration figure. Furthermore, we agree with the Reviewer that is important to undertake an extended analysis of how the reservoir flooding will impact carbon fluxes. In addition to the uncertainty analysis completed for the study, we also undertook a sensitivity experiment of the swamp’s carbon fluxes to different reservoir flooded conditions in the second manuscript that was submitted to the Geoscientific Model Development (GMD) journal.
- Previous studies, such as those by Metzger et al., have conducted extensive modeling work on peatlands using CoupModel. To better situate your work within this context, it would be helpful to clarify how your study builds on or differs from these earlier efforts. If the goal is only to emphasize modeling performance in temperate swamp condition, a more detailed discussion comparing model performance and setup across peatland types is encouraged.
AUTHOR'S RESPONSE
The agree with the suggestion. As highlighted above, this is the first time that the CoupModel will be tested on temperate swamp peatland. Existing modeling studies with CoupModel (e.g. Metzger et al.) focused on bogs and fens, and were mostly in Europe and not North America. This means the models set-up process (e.g parameterization) at Beverly Swamp will be very different from this existing studies. Nevertheless, the aforementioned studies were consulted and well cited in our manuscript where applicable. We will make sure that these differences and the advances resulting from the present study are clearly highlighted in the revised manuscript.
- Model Structure and Version-Specific Updates. Sections 2.3 and 2.4 describe CoupModel’s structure too much. However, given that CoupModel has been described extensively in prior literature (e.g., Jansson et al., 2013; He et al., 2020), I suggest focusing more on what is unique or updated in version 6.2.2, especially in relation to this study. For example, have any components or parameterizations been modified specifically for temperate swamp simulation? Please also include specific parameter values used for S1, such as van Genuchten parameters, and clarify any deviations from default settings.
AUTHOR'S RESPONSE
The suggestions are well received. We will elucidate further the uniqueness of temperate swamp parameterization, parameter value assignment and other relevant aspects when revising the manuscript in the next stage.
- Dataset Description and Temporal Resolution. The manuscript mentions the use of multiple datasets with varying frequencies and durations, but the descriptions lack clarity. I strongly recommend adding a summary table that categorizes all datasets (e.g., input, calibration, validation), including their temporal resolution, measurement duration, methodology, and target variables. Also, please clarify how the authors standardized temporal resolution where dataset frequencies differed.
AUTHOR'S RESPONSE
The suggestions are well received. We will take this into consideration when revising our manuscript. For example, the data categories, their timescales and other important information will be tabulated. However, we should mention that all the measurements and simulated variables were on a daily timescale.
- Clarification of Model Calibration Process. Given that observation periods differ among soil thermal data, respiration, and energy fluxes, more detail is needed on the model calibration and prioritization process. For instance, which variables were given higher priority in the calibration? What parameter values were adjusted, and how do they compare to defaults or values used in other peatland studies?
AUTHOR'S RESPONSE
We will ensure that the calibration process is clearly described in the revised manuscript and that the reader is pointed to the appropriate details in the supplementary material. You will observe from the manuscript that soil respiration and simulated hydrological variables were given priority in the calibration. The parameters that were changed for the modelling experiment are listed in supplementary 2. Furthermore, we undertook a global sensitivity analysis and multiple variable calibration with the GLUE approach for this modelling experiment in a later manuscript that is submitted to the Geoscientific Model Development (GMD) journal.
- Soil Water Flow Modeling Approaches. The manuscript mentions the use of both Darcy’s Law and a two-domain approach, but does not provide comparison results. The authors state that the two-domain method performed better for water table and moisture simulation. If this approach was ultimately used in the full simulation, please clarify how it influenced other processes, such as NEE and soil temperature. Including comparison plots would enhance transparency and understanding.
AUTHOR'S RESPONSE
In addition to the ordinary Darcy’s matrix flow, we tested a two-domain approach that takes into consideration a bypass of the micropore soil matrix flow system at Beverly Swamp. You will notice on lines 332 to 335 that we discussed the results of the two-domain approach. It improved the simulation of the hydrological conditions but produced a mixed response for soil respiration during the calibration and validation periods. Therefore, we focused on the one-domain matrix flow for this manuscript. We will make sure that this discussion is clear in the revised version and that our decision to continue with the one-domain approach is clearly justified (i.e., due to our focus on carbon cycling estimation and the response soil respiration to the choice of approach). In addition, we will include the results of the two-domain approach in the supplementary material during the next review stage.
- Figure 4 shows that the model fails to simulate soil temperatures below 0°C at both 5 cm and 30 cm, possibly due to forced zeroing. Could the authors provide a technical explanation for this outcome?
AUTHOR'S RESPONSE
Yes, CoupModel overestimated soil temperature during the winter season because it was not able to properly capture the freeze-thaw process during this season. In addition, the moisture deficit from the fall season also extended into the winter season, thus lowering the specific heat capacity of the soil during this season. We will make sure this is clear in the discussion in the revised version.
- The current manuscript focuses mainly on observational trends over time. However, it would be insightful to evaluate how well the model captures interannual variability in key variables through direct model–observation comparison.
AUTHOR'S RESPONSE
We realistically captured the interannual variability of key variables (e.g. soil respiration) (please see attached figure 9), as the available measured data permitted in the discussion section. However, because there is a wide data gap in measurements, it was not realistic adopting a direct model-observation comparison to analyze the interannual variability for the whole study period (40 years). Nevertheless, we will also elucidate this in our discussion section during the next review stage.
- The clarity of Figures 4–9 should be improved. It is currently difficult to interpret the data—please specify whether these are daily, monthly, or annual values, and adjust formatting for readability.
AUTHOR'S RESPONSE
We will work to improve the figures and reflect in their caption that the presented data are on a daily resolution during the next review stage.
- Peatlands are also known methane sources. I am curious that has CoupModel been evaluated or configured for methane flux simulation in peatlands?
AUTHOR'S RESPONSE
Our study only focused on the CO2 fluxes at Beverly Swamp, so we did not simulate methane fluxes. Nevertheless, see e.g., Zhao et al (2025) Agricultural and Forest Meteorology 362 110359, for CoupModel applications in CH4 flux simulations in subarctic wetlands.
- Minor Suggestions:
- Consider including “CoupModel” in the title, as it is the central modeling tool used.
AUTHOR'S RESPONSE
Many thanks for this. We will give the suggestion a strong consideration
- Integrate Figures 6 and 7 into one composite figure for clarity.
AUTHOR'S RESPONSE
We will do this in the next review stage.
- There seems to be a misplaced plot in Figure 9; please check and correct.
AUTHOR'S RESPONSE
Again, we apologize for mistakenly repeating the water table level figure (figure 5) that is meant for soil respiration in Figure 9 (please find attached). We will correct the error by updating the soil respiration figure.
-
AC2: 'Reply on RC2', Oluwabamise Afolabi, 26 May 2025
Status: closed
-
RC1: 'Comment on egusphere-2024-4049', Anonymous Referee #1, 18 Apr 2025
This manuscript presents a process-based modelling study of carbon dynamics in a temperate swamp, based on extensive field measurements collected over a 40-year period. If the model demonstrates successful calibration and validation against consistent observations of abiotic variables (soil temperature, soil moisture, water table level, and heat fluxes) and carbon fluxes (soil respiration and GPP), the study could make a substantial contribution to our understanding of the long-term responses of ecosystem functioning—specifically water and carbon exchange—to recent climate variability, including both seasonal and interannual changes. It also has the potential to identify the key environmental drivers involved.
However, the current version of the manuscript does not meet the quality standards required for Biogeosciences, primarily due to the lack of sufficient detail regarding model calibration of carbon fluxes and poor model performance. I recommend that the authors address the following aspects:
(1) Presentation of Results
Figure 1: Please provide more detailed information on the specific measurements collected at each sampling site.Figure 2: Why is ground heat shown at the bottom of the soil profile? It should be included as part of the surface energy balance. Additionally, if the model includes two tree species, the authors should explain how multiple canopies are represented. I suggest showing parameters such as species-specific height, root depth, leaf area index (LAI), and litterfall for both trees and peatland vegetation.
Figure 3: What does the red line represent? Is it a moving average?
Figures 4a, 4b, and 5a: Please clarify the meaning of the red line and the grey shaded area—do these represent a regression line and confidence interval, respectively?
Figures 8b and 9: The model appears to underestimate soil moisture and water table levels, while simulated evapotranspiration (ET) seems to be overestimated. I recommend revisiting the water balance and interpreting these model biases.
(2) Missing Calibration for Ecosystem Respiration
The manuscript appears to lack a figure or discussion on the model calibration results for ecosystem respiration. This is a critical component for evaluating model performance and should be included.(3) Interannual Variability in Soil Respiration
Soil respiration from June to September shows a positive trend over time, while GPP and air temperature do not show any clear trends. Can the authors explain what might be driving the increased soil respiration during the growing season?Minor Comments
Abstract, line 17: If negative NEE is defined as a carbon sink, then a decline in NEE (i.e., becoming more negative) suggests increased carbon uptake. Please clarify this wording.Line 35: A comma is missing after “Meanwhile”.
Line 45: "> 5 years" should be corrected to "< 5 years".
Line 48: The phrase “capture the impact of regional multidecadal disturbances (e.g., climate change)” could be improved to “capture regional ecosystem responses to multidecadal climate change”.
Line 67: Consider rephrasing to clarify: “simulate key plant processes that regulate water, energy, and carbon fluxes in temperate swamps.”
Line 385: Could this result be related to a bias in simulated evapotranspiration?
Lines 446–454: I suggest presenting the modelled LAI and linking it to the discussion in this section.
Line 495: The decoupling phenomenon typically occurs under very dry or very wet conditions. The authors should support this discussion with simulated GPP or stomatal conductance results.
Lines 755–760: A reference appears to be duplicated
Citation: https://doi.org/10.5194/egusphere-2024-4049-RC1 -
AC1: 'Reply on RC1', Oluwabamise Afolabi, 26 May 2025
GENERAL REMARKS
We thank the reviewers for taking out the time to read our submitted manuscript and sharing important feedback on it.
Even though we have responded to the reviewer’s specific feedback below, we would like to highlight that our submitted manuscript is one of the first attempts to model the long-term carbon dynamics of a temperate swamp peatland and analyze its controls on soil respiration. Currently, temperate swamp peatlands systems are understudied and very few data exist on this ecosystem (Davidson et al., 2022). Our study site, Beverly Swamp included data from the 1980s, short studies from 1998 – 2000 by Davidson et al (2019), McCarter et al. (2024) and resumed measurements from 2022. Thus, Beverly Swamp to our knowledge, represents one of the most well-studied temperate swamp peatlands. The long-term coverage of empirical data makes long term modeling study as ours possible, however, these data collections were in some cases not continuous and were done by different groups with varying methods over a 40+ year period. Thus, there are considerable uncertainties associated with these measurements. This aspect was mentioned in the current version but probably not discussed in depth. We will add those uncertainties in our revision and discuss the extent that the model data discrepancies can be explained by the measured data, model structure and parameters.
However, we would also like to mention that the main purpose of this paper is not to discuss those uncertainties or model performance in details but rather to first test whether or not a processed-based model can simulate the coupled hydrological, heat and carbon dynamics of a temperate swamp peatland system that has not been simulated so far. Furthermore, we undertook a systematic uncertainty analysis by using the Generalized Likelihood Uncertainty Estimation (GLUE) approach to identify the errors associated with the different aspects (e.g., measurement, parameter and model structure) of the modelling exercise. This uncertainty analysis and other associated experiments have been submitted as a separate manuscript to the Geoscientific Model Development (GMD) journal. Therefore, this means that some of the Reviewer’s feedbacks were also addressed in the latter manuscript. Although this first manuscript lacks the global sensitivity analysis, our goal was to investigate the effectiveness of CoupModel for characterizing temperate swamp thermal, hydrological and carbon cycling processes and provide first estimates of modelled swamp carbon balance. We believe this is already an important contribution to literature and of interest to the readers of Biogeosciences Journal.
In addition, the reviewers highlighted some concerns about the presentation quality of the paper. Some of the comments (e.g., model performance) were partly due to the figure presentations (e.g., mistakes in uploading our soil respiration figure). We mistakenly repeated the water table level figure (figure 5) instead of soil respiration in Figure 9. The journal editors were notified of this mistake, but we were told to update it during the next review stage. We sincerely apologize for the mistake in our initial uploading of the manuscript. We will fix this error in our revision, but we have also attached the figure. Furthermore, we will revise the texts (clarification of model calibration, parameter determination, expansion of the result analysis) and improve all figures to substantially improve the presentation.
REVIEWER’S COMMENT
This manuscript presents a process-based modelling study of carbon dynamics in a temperate swamp, based on extensive field measurements collected over a 40-year period. If the model demonstrates successful calibration and validation against consistent observations of abiotic variables (soil temperature, soil moisture, water table level, and heat fluxes) and carbon fluxes (soil respiration and GPP), the study could make a substantial contribution to our understanding of the long-term responses of ecosystem functioning—specifically water and carbon exchange—to recent climate variability, including both seasonal and interannual changes. It also has the potential to identify the key environmental drivers involved.
However, the current version of the manuscript does not meet the quality standards required for Biogeosciences, primarily due to the lack of sufficient detail regarding model calibration of carbon fluxes and poor model performance. I recommend that the authors address the following aspects:
Presentation of Results Figure 1: Please provide more detailed information on the specific measurements collected at each sampling site.
AUTHOR'S RESPONSE
We will ensure the measurement stations are properly displayed on the map of Beverly Swamp (Figure 1), and revise the texts in the method section accordingly to reflect exact measurement point and period of data collection.
We should also include that the details of the datasets used for this study were extensively described in the manuscript. You will observe that the data sources used for driving CoupModel, calibration and validation purposes were thoroughly described on pages 4 to 6 (Lines 96 to 142) of the manuscript. The driving (climatic) variables namely, air temperature, precipitation, relative humidity, global radiation and windspeed were sourced from nearby weather stations and additional sources described on lines 102 to 109, while lateral water input data was collected from nearby gauging station (Line 110 and Figure 1). We measured hourly water table levels across the northern transect of the swamp (see figure 1) from June 2022 to July 2023 with Solinst leveloggers, while a Campbell data logger was installed to log hourly datasets of soil temperature and volumetric moisture contents at different depths for the same period. The specific measurements points are also highlighted in Figure 1 (Map of Beverly swamp). Also see Lines 126 to 142.
Furthermore, energy fluxes (net radiation, ground heat, sensible heat), soil temperate (0-5cm), water table level and snow depth measurements from August 1983 to July 1987 were sourced from Munro et al. (2000) for the calibration period of the modeling study. See lines 115 to 119 for detailed description and sample collection point. Soil respiration and soil organic carbon data used for model initialization, calibration (1998 – 2000) and validation (2022 – 2023) were sourced from Davidson et al., 2019, McCarter et al., 2024 and Schmidt and Strack, 2024 (See lines 120 to 125). Details of other measurements used for estimating model parameters values can be found in Appendix A (Table A1).
- Figure 2: Why is ground heat shown at the bottom of the soil profile? It should be included as part of the surface energy balance. Additionally, if the model includes two tree species, the authors should explain how multiple canopies are represented. I suggest showing parameters such as species-specific height, root depth, leaf area index (LAI), and litterfall for both trees and peatland vegetation.
AUTHOR'S RESPONSE
We realize that the presentation of Figure 2 might cause some misunderstanding but the ground heat flow should not be confused with surface soil heat flux. The surface soil heat flux is the residual energy flux of the surface energy balance that is explicitly simulated in CoupModel (see equations in the appendix), however, the ground heat flux illustrated in figure 2 delineates the lower boundary conditions of the model for heat conduction at Beverly swamp (see lines 174 and 175 for description). We will revise the figure and include additional caption to explain that the ground heat flux is for simulating soil temperature.
Beverly swamp’s forest cover consists of both coniferous and deciduous species (plant functional types, PFTs) with almost equal standing biomass (Munro 1979). Therefore, the reported lumped PTFs characteristics of existing studies (e.g., Munro et al, 2000) were used to parameterize plant properties in CoupModel(one-dimensional model). In addition, the vegetation of the swamp in the CoupModel was simulated by an “explicit bigleaf” and a single representative canopy layer characterized by the closed canopy structure (LAI of 5-6 that was simulated dynamically, which makes it vary with seasons) and root depth of ~30cm. We did not simulate understory vegetation since this category is low and thus insignificant for the simulated carbon and hydrology fluxes. The specific characteristics / parameter values of the vegetation cover at Beverly Swamp are listed in Table A.1 (Appendix A) of the submitted transcript.
If a revision is requested, we will clarify aspects of Figure 2 further in the caption and ensure that all model description is clear to address the reviewer’s comments.
- Figure 3: What does the red line represent? Is it a moving average?
AUTHOR'S RESPONSE
The red line in figure 3 shows the trend direction of precipitation and air temperature. In particular, it is a function (LOWESS) in the R statistical software (Kendall library) that smoothens out data through the fitting of local regression. Captions will be updated in the revision to state this clearly.
- Figures 4a, 4b, and 5a: Please clarify the meaning of the red line and the grey shaded area—do these represent a regression line and confidence interval, respectively?
AUTHOR'S RESPONSE
The red lines in figures 4a, 4b and 5a are regression lines added to the plot by a function (geom_smooth) in the R statistical software, while the grey band represents the confidence interval. Captions will be updated in the revision to state this clearly.
- Figures 8b and 9: The model appears to underestimate soil moisture and water table levels, while simulated evapotranspiration (ET) seems to be overestimated. I recommend revisiting the water balance and interpreting these model biases.
AUTHOR'S RESPONSE
We apologize for the mistake in our initial uploading of the manuscript. We think the problem in the referenced Figures (8b and 9) is not really about water balance. The main error originates from the mistake of repeating the water table level figure (figure 5) instead of soil respiration in Figure 9. We would fix this error in our next revision.
Furthermore, the limitations and uncertainties associated with simulated water balance components were extensively discussed on lines 380 to 404 of the manuscript. Please note that our study relied on water table level measurement for both the calibration (July 1983 – August 1987) and validation periods (2022 – 2023) but only had access to volumetric moisture content measurements (2022 – 2023) and ET (1983 – 1987) for calibration period. Consequently, we limited our water balance evaluation to available datasets. There is no clear evidence in Figure 8 that ET is overestimated by the model and in fact, the comparison of observed and simulated ET patterns was consistent with evaluated water table levels during the calibration period (1983 – 1987) when both ET and WTD measurements were readily available (please see attached figures 5 and 8 ). Nevertheless, we will elucidate this in the next revision to make it clearer.
- Missing Calibration for Ecosystem Respiration. The manuscript appears to lack a figure or discussion on the model calibration results for ecosystem respiration. This is a critical component for evaluating model performance and should be included.
AUTHOR'S RESPONSE
Again, we apologize for mistakenly repeating the water table level figure (figure 5) that is meant for soil respiration in Figure 9 (please see attached). We will correct the error by updating the soil respiration figure. However, we believe the reviewer is referring to soil respiration here and not ecosystem respiration since no eddy covariance flux measurement were done at Beverly Swamp.
Kindly note that the model calibration results of soil respiration were discussed on lines 414 to 426 (Page 18) of the submitted transcript. However, we can also add some sentences to further discuss the model data discrepancies of the soil respiration in our revision.
- Interannual Variability in Soil Respiration Soil respiration from June to September shows a positive trend over time, while GPP and air temperature do not show any clear trends. Can the authors explain what might be driving the increased soil respiration during the growing season?
AUTHOR'S RESPONSE
Even though the increasing air temperature trend between June and September was not as pronounced as those reported for winter and fall seasons, moderate rise in air temperature was also observed during this period (see Figure B1). A moderate increase in air-soil temperature would accelerate soil microbial activities, and drier conditions with extended oxic zone during the growing season may have supported a spike in aerobic decomposition of swamp peat at both surface and deep layers. Consequently, soil respiration followed an increasing trend (11%) over the study period. The warmer conditions may not have supported a spike in GPP because of other limiting factors that can affect productivity. For instance, excessive dryness during this period and nutrient limitations may have affected GPP rates in the swamp. In addition, our study did not consider the effect of CO2 fertilization which has been reported to increase GPP rates in the last decades. We will make sure that this discussion of trends is clear in the discussion if a revision of the manuscript is requested.
- Minor Comments
Abstract, line 17: If negative NEE is defined as a carbon sink, then a decline in NEE (i.e., becoming more negative) suggests increased carbon uptake. Please clarify this wording.
AUHTOR'S RESPONSE
Yes, we followed the convention of attributing a negative NEE to more carbon intake by the swamp and the vice versa. For the study period, the swamp’s net carbon intake dwindled as more carbon was lost via soil respiration and with little intake via GPP. We will make sure that the text is clear with the trend in any revision.
- Line 35: A comma is missing after “Meanwhile”.
AUHTOR'S RESPONSE
Thank you for the observation. A comma will be included next to the word.
- Line 45: "> 5 years" should be corrected to "< 5 years".
AUHTOR'S RESPONSE
Thank you, the observation is valid. “>5 years” would be rewritten as “<5 years”.
- Line 48: The phrase “capture the impact of regional multidecadal disturbances (e.g., climate change)” could be improved to “capture regional ecosystem responses to multidecadal climate change”.
AUHTOR'S RESPONSE
Statement will be rejigged in the revision.
- Line 67: Consider rephrasing to clarify: “simulate key plant processes that regulate water, energy, and carbon fluxes in temperate swamps.”
AUHTOR'S RESPONSE
Statement will be clarified.
- Line 385: Could this result be related to a bias in simulated evapotranspiration?
AUHTOR'S RESPONSE
There is little or no evidence to show that the high mismatch between simulated and observed water table level during the validation period comes from the bias in simulated evapotranspiration. Instead, the difference can be linked to very dry conditions (annual precipitation record of 616 mm) during the validation period. Because the model was mostly trained with wet years during the calibration period (1983 – 1986), it was not able to capture the extremely dry conditions presented in the validation period. We further confirmed this with an uncertainty analysis that was completed in another manuscript. We will more clearly highlight the abnormally dry conditions of the validation period in the revised manuscript.
- Lines 446–454: I suggest presenting the modelled LAI and linking it to the discussion in this section.
AUHTOR'S RESPONSE
It is possible to present the simulated LAI in the manuscript and further link it to the discussion. The comparison of our simulated and observed LAI produced a R2 of 0.81 but we deliberately did not dwell on this in the submitted manuscript since LAI is not the focus of our study. Nevertheless, we discussed it more extensively in a later manuscript that was submitted to the Geoscientific Model Development (GMD) journal. We can add a comparison of simulated and observed in the revised manuscript.
- Line 495: The decoupling phenomenon typically occurs under very dry or very wet conditions. The authors should support this discussion with simulated GPP or stomatal conductance results.
AUHTOR'S RESPONSE
Our result shows that the decoupling occurs under very dry and warm conditions. Simulated GPP can be found in Figure B5 (Appendix B). We will make sure this link is clear in the discussion as suggested by the reviewer.
- Lines 755–760: A reference appears to be duplicated
AUHTOR'S RESPONSE
The publications by Metzger et al. were not duplicated. We cited three different papers by Metzger et al. (2015, 2016a, 2016b) and a PhD thesis by Metzger 2015.
-
AC1: 'Reply on RC1', Oluwabamise Afolabi, 26 May 2025
-
RC2: 'Comment on egusphere-2024-4049', Anonymous Referee #2, 23 Apr 2025
The manuscript by Afolabi et al. presents a study evaluating the performance of the process-based model CoupModel V6.2.2 in simulating soil hydrothermal dynamics and soil CO₂ fluxes in a swamp ecosystem affected by seasonal reservoir flooding. This is an interesting topic, as modeling biogeochemical processes in wetland systems remains a significant challenge. I particularly appreciate the authors' effort to simulate such dynamics in a peatland disturbed by reservoir flooding, which is very helpful to the modeling community and offers a potentially valuable example of how process-based models can be applied to such condition. However, I believe the manuscript requires further improvement before it can be considered for publication. These improvements primarily relate to clarification of the methods, expansion of the result analysis, and enhanced data presentation. Below are my main suggestions and concerns:
Major Comments:
1. The swamp site is notably influenced by seasonal reservoir flooding. It would indeed be valuable if CoupModel could adequately capture the hydrological, biogeochemical, and biophysical processes under such disturbances. However, the model appears to fail in simulating key hydrological variables such as water table depth and deep soil moisture—despite the use of observed lateral water input. I recommend the authors conduct a further hydrological parameter sensitivity analysis. This may help identify additional factors contributing to model underperformance beyond the underestimation of latent heat flux.
Furthermore, while I am interested in the model's potential to simulate carbon dynamics under seasonal flooding, the current presentation of results (particularly Figure 9) fail to the assessment of soil respiration dynamics. Please consider revising Figure B8 to present these processes at daily or monthly resolution, which could offer a clearer picture of the model’s capacity to capture seasonal variations. An extended analysis on the impact of reservoir flooding on carbon fluxes and other processes would be especially valuable.
2. Previous studies, such as those by Metzger et al., have conducted extensive modeling work on peatlands using CoupModel. To better situate your work within this context, it would be helpful to clarify how your study builds on or differs from these earlier efforts. If the goal is only to emphasize modeling performance in temperate swamp condition, a more detailed discussion comparing model performance and setup across peatland types is encouraged.
3. Model Structure and Version-Specific Updates. Sections 2.3 and 2.4 describe CoupModel’s structure too much. However, given that CoupModel has been described extensively in prior literature (e.g., Jansson et al., 2013; He et al., 2020), I suggest focusing more on what is unique or updated in version 6.2.2, especially in relation to this study. For example, have any components or parameterizations been modified specifically for temperate swamp simulation? Please also include specific parameter values used for S1, such as van Genuchten parameters, and clarify any deviations from default settings.
4. Dataset Description and Temporal Resolution. The manuscript mentions the use of multiple datasets with varying frequencies and durations, but the descriptions lack clarity. I strongly recommend adding a summary table that categorizes all datasets (e.g., input, calibration, validation), including their temporal resolution, measurement duration, methodology, and target variables. Also, please clarify how the authors standardized temporal resolution where dataset frequencies differed.
5. Clarification of Model Calibration Process. Given that observation periods differ among soil thermal data, respiration, and energy fluxes, more detail is needed on the model calibration and prioritization process. For instance, which variables were given higher priority in the calibration? What parameter values were adjusted, and how do they compare to defaults or values used in other peatland studies?
6. Soil Water Flow Modeling Approaches. The manuscript mentions the use of both Darcy’s Law and a two-domain approach, but does not provide comparison results. The authors state that the two-domain method performed better for water table and moisture simulation. If this approach was ultimately used in the full simulation, please clarify how it influenced other processes, such as NEE and soil temperature. Including comparison plots would enhance transparency and understanding.
7. Figure 4 shows that the model fails to simulate soil temperatures below 0°C at both 5 cm and 30 cm, possibly due to forced zeroing. Could the authors provide a technical explanation for this outcome?
8. The current manuscript focuses mainly on observational trends over time. However, it would be insightful to evaluate how well the model captures interannual variability in key variables through direct model–observation comparison.
9. The clarity of Figures 4–9 should be improved. It is currently difficult to interpret the data—please specify whether these are daily, monthly, or annual values, and adjust formatting for readability.
10. Peatlands are also known methane sources. I am curious that has CoupModel been evaluated or configured for methane flux simulation in peatlands?
Minor Suggestions:1. Consider including “CoupModel” in the title, as it is the central modeling tool used.
2. Integrate Figures 6 and 7 into one composite figure for clarity.
3. There seems to be a misplaced plot in Figure 9; please check and correct.
4. Provide exact p-values alongside R² values for all statistical comparisons.
5. Clarify the default parameter values used in S2, and highlight any key changes made.
Citation: https://doi.org/10.5194/egusphere-2024-4049-RC2 -
AC2: 'Reply on RC2', Oluwabamise Afolabi, 26 May 2025
GENERAL REMARKS
We thank the reviewers for taking out the time to read our submitted manuscript and sharing important feedback on it. Even though we have responded to the reviewer’s specific feedback below, we would like to highlight that our submitted manuscript is one of the first attempts to model the long-term carbon dynamics of a temperate swamp peatland and analyze its controls on soil respiration. Currently, temperate swamp peatlands systems are understudied and very few data exist on this ecosystem (Davidson et al., 2022). Our study site, Beverly Swamp included data from the 1980s, short studies from 1998 – 2000 by Davidson et al (2019), McCarter et al. (2024) and resumed measurements from 2022. Thus, Beverly Swamp to our knowledge, represents one of the most well-studied temperate swamp peatlands. The long-term coverage of empirical data makes long term modeling study as ours possible, however, these data collections were in some cases not continuous and were done by different groups with varying methods over a 40+ year period. Thus, there are considerable uncertainties associated with these measurements. This aspect was mentioned in the current version but probably not discussed in depth. We will add those uncertainties in our revision and discuss the extent that the model data discrepancies can be explained by the measured data, model structure and parameters. However, we would also like to mention that the main purpose of this paper is not to discuss those uncertainties or model performance in details but rather to first test whether or not a processed-based model can simulate the coupled hydrological, heat and carbon dynamics of a temperate swamp peatland system that has not been simulated so far. Furthermore, we undertook a systematic uncertainty analysis by using the Generalized Likelihood Uncertainty Estimation (GLUE) approach to identify the errors associated with the different aspects (e.g., measurement, parameter and model structure) of the modelling exercise. This uncertainty analysis and other associated experiments have been submitted as a separate manuscript to the Geoscientific Model Development (GMD) journal. Therefore, this means that some of the Reviewer’s feedbacks were also addressed in the latter manuscript. Although this first manuscript lacks the global sensitivity analysis, our goal was to investigate the effectiveness of CoupModel for characterizing temperate swamp thermal, hydrological and carbon cycling processes and provide first estimates of modelled swamp carbon balance. We believe this is already an important contribution to literature and of interest to the readers of Biogeosciences Journal.
In addition, the reviewers highlighted some concerns about the presentation quality of the paper. Some of the comments (e.g., model performance) were partly due to the figure presentations (e.g., mistakes in uploading our soil respiration figure). We mistakenly repeated the water table level figure (figure 5) instead of soil respiration in Figure 9. The journal editors were notified of this mistake, but we were told to update it during the next review stage. We sincerely apologize for the mistake in our initial uploading of the manuscript. We will fix this error in our revision, but we have also attached the figure. Furthermore, we will revise the texts (clarification of model calibration, parameter determination, expansion of the result analysis) and improve all figures to substantially improve the presentation.
REVIEWER'S COMMENT
The manuscript by Afolabi et al. presents a study evaluating the performance of the process-based model CoupModel V6.2.2 in simulating soil hydrothermal dynamics and soil CO₂ fluxes in a swamp ecosystem affected by seasonal reservoir flooding. This is an interesting topic, as modeling biogeochemical processes in wetland systems remains a significant challenge. I particularly appreciate the authors' effort to simulate such dynamics in a peatland disturbed by reservoir flooding, which is very helpful to the modeling community and offers a potentially valuable example of how process-based models can be applied to such condition. However, I believe the manuscript requires further improvement before it can be considered for publication. These improvements primarily relate to clarification of the methods, expansion of the result analysis, and enhanced data presentation. Below are my main suggestions and concerns:
AUTHOR'S RESPONSE
Many thanks for the positive comments on the significance of our study. In our revision, we will substantially increase the presentation of the paper. Please also see our general remarks and specific responses below.
- Major Comments:
- The swamp site is notably influenced by seasonal reservoir flooding. It would indeed be valuable if CoupModel could adequately capture the hydrological, biogeochemical, and biophysical processes under such disturbances. However, the model appears to fail in simulating key hydrological variables such as water table depth and deep soil moisture—despite the use of observed lateral water input. I recommend the authors conduct a further hydrological parameter sensitivity analysis. This may help identify additional factors contributing to model underperformance beyond the underestimation of latent heat flux.
AUTHOR'S RESPONSE
We appreciate the comments, however, we disagree that the model failed to simulate key hydrological variables. The performance indices (e.g., R square statistics) of the simulated hydrological variables (lines 296 to 335) and evaluation discussions (lines 379 to 408) showed a reasonably good model – observation agreement despite all the uncertainty and limitations associated with the modelling experiment. Please see attached figure.
Nevertheless, we agree with the Reviewer that a robust sensitivity analysis may help improve our modelling. Consequently, we undertook an uncertainty of the modelling experiment using the Generalized Likelihood Uncertainty Estimation (GLUE) approach to identify the errors associated with the different aspects (measurement, parameter and model structure) of the modelling exercise. This uncertainty analysis and other associated modelling experiments has been submitted as a separate manuscript to the Geoscientific Model Development (GMD) journal. Part of the lower model fit for water table in the validation period also results from this being an abnormally dry year, a condition over which the model was not trained. We will highlight this further in the discussion in a revised version of the manuscript.
- Furthermore, while I am interested in the model's potential to simulate carbon dynamics under seasonal flooding, the current presentation of results (particularly Figure 9) fail to the assessment of soil respiration dynamics. Please consider revising Figure B8 to present these processes at daily or monthly resolution, which could offer a clearer picture of the model’s capacity to capture seasonal variations. An extended analysis on the impact of reservoir flooding on carbon fluxes and other processes would be especially valuable.
AUTHOR'S RESPONSE
We apologize for mistakenly repeating the water table level figure (figure 5) that is meant for soil respiration in Figure 9 (please see attached). We will correct the error by updating the soil respiration figure. Furthermore, we agree with the Reviewer that is important to undertake an extended analysis of how the reservoir flooding will impact carbon fluxes. In addition to the uncertainty analysis completed for the study, we also undertook a sensitivity experiment of the swamp’s carbon fluxes to different reservoir flooded conditions in the second manuscript that was submitted to the Geoscientific Model Development (GMD) journal.
- Previous studies, such as those by Metzger et al., have conducted extensive modeling work on peatlands using CoupModel. To better situate your work within this context, it would be helpful to clarify how your study builds on or differs from these earlier efforts. If the goal is only to emphasize modeling performance in temperate swamp condition, a more detailed discussion comparing model performance and setup across peatland types is encouraged.
AUTHOR'S RESPONSE
The agree with the suggestion. As highlighted above, this is the first time that the CoupModel will be tested on temperate swamp peatland. Existing modeling studies with CoupModel (e.g. Metzger et al.) focused on bogs and fens, and were mostly in Europe and not North America. This means the models set-up process (e.g parameterization) at Beverly Swamp will be very different from this existing studies. Nevertheless, the aforementioned studies were consulted and well cited in our manuscript where applicable. We will make sure that these differences and the advances resulting from the present study are clearly highlighted in the revised manuscript.
- Model Structure and Version-Specific Updates. Sections 2.3 and 2.4 describe CoupModel’s structure too much. However, given that CoupModel has been described extensively in prior literature (e.g., Jansson et al., 2013; He et al., 2020), I suggest focusing more on what is unique or updated in version 6.2.2, especially in relation to this study. For example, have any components or parameterizations been modified specifically for temperate swamp simulation? Please also include specific parameter values used for S1, such as van Genuchten parameters, and clarify any deviations from default settings.
AUTHOR'S RESPONSE
The suggestions are well received. We will elucidate further the uniqueness of temperate swamp parameterization, parameter value assignment and other relevant aspects when revising the manuscript in the next stage.
- Dataset Description and Temporal Resolution. The manuscript mentions the use of multiple datasets with varying frequencies and durations, but the descriptions lack clarity. I strongly recommend adding a summary table that categorizes all datasets (e.g., input, calibration, validation), including their temporal resolution, measurement duration, methodology, and target variables. Also, please clarify how the authors standardized temporal resolution where dataset frequencies differed.
AUTHOR'S RESPONSE
The suggestions are well received. We will take this into consideration when revising our manuscript. For example, the data categories, their timescales and other important information will be tabulated. However, we should mention that all the measurements and simulated variables were on a daily timescale.
- Clarification of Model Calibration Process. Given that observation periods differ among soil thermal data, respiration, and energy fluxes, more detail is needed on the model calibration and prioritization process. For instance, which variables were given higher priority in the calibration? What parameter values were adjusted, and how do they compare to defaults or values used in other peatland studies?
AUTHOR'S RESPONSE
We will ensure that the calibration process is clearly described in the revised manuscript and that the reader is pointed to the appropriate details in the supplementary material. You will observe from the manuscript that soil respiration and simulated hydrological variables were given priority in the calibration. The parameters that were changed for the modelling experiment are listed in supplementary 2. Furthermore, we undertook a global sensitivity analysis and multiple variable calibration with the GLUE approach for this modelling experiment in a later manuscript that is submitted to the Geoscientific Model Development (GMD) journal.
- Soil Water Flow Modeling Approaches. The manuscript mentions the use of both Darcy’s Law and a two-domain approach, but does not provide comparison results. The authors state that the two-domain method performed better for water table and moisture simulation. If this approach was ultimately used in the full simulation, please clarify how it influenced other processes, such as NEE and soil temperature. Including comparison plots would enhance transparency and understanding.
AUTHOR'S RESPONSE
In addition to the ordinary Darcy’s matrix flow, we tested a two-domain approach that takes into consideration a bypass of the micropore soil matrix flow system at Beverly Swamp. You will notice on lines 332 to 335 that we discussed the results of the two-domain approach. It improved the simulation of the hydrological conditions but produced a mixed response for soil respiration during the calibration and validation periods. Therefore, we focused on the one-domain matrix flow for this manuscript. We will make sure that this discussion is clear in the revised version and that our decision to continue with the one-domain approach is clearly justified (i.e., due to our focus on carbon cycling estimation and the response soil respiration to the choice of approach). In addition, we will include the results of the two-domain approach in the supplementary material during the next review stage.
- Figure 4 shows that the model fails to simulate soil temperatures below 0°C at both 5 cm and 30 cm, possibly due to forced zeroing. Could the authors provide a technical explanation for this outcome?
AUTHOR'S RESPONSE
Yes, CoupModel overestimated soil temperature during the winter season because it was not able to properly capture the freeze-thaw process during this season. In addition, the moisture deficit from the fall season also extended into the winter season, thus lowering the specific heat capacity of the soil during this season. We will make sure this is clear in the discussion in the revised version.
- The current manuscript focuses mainly on observational trends over time. However, it would be insightful to evaluate how well the model captures interannual variability in key variables through direct model–observation comparison.
AUTHOR'S RESPONSE
We realistically captured the interannual variability of key variables (e.g. soil respiration) (please see attached figure 9), as the available measured data permitted in the discussion section. However, because there is a wide data gap in measurements, it was not realistic adopting a direct model-observation comparison to analyze the interannual variability for the whole study period (40 years). Nevertheless, we will also elucidate this in our discussion section during the next review stage.
- The clarity of Figures 4–9 should be improved. It is currently difficult to interpret the data—please specify whether these are daily, monthly, or annual values, and adjust formatting for readability.
AUTHOR'S RESPONSE
We will work to improve the figures and reflect in their caption that the presented data are on a daily resolution during the next review stage.
- Peatlands are also known methane sources. I am curious that has CoupModel been evaluated or configured for methane flux simulation in peatlands?
AUTHOR'S RESPONSE
Our study only focused on the CO2 fluxes at Beverly Swamp, so we did not simulate methane fluxes. Nevertheless, see e.g., Zhao et al (2025) Agricultural and Forest Meteorology 362 110359, for CoupModel applications in CH4 flux simulations in subarctic wetlands.
- Minor Suggestions:
- Consider including “CoupModel” in the title, as it is the central modeling tool used.
AUTHOR'S RESPONSE
Many thanks for this. We will give the suggestion a strong consideration
- Integrate Figures 6 and 7 into one composite figure for clarity.
AUTHOR'S RESPONSE
We will do this in the next review stage.
- There seems to be a misplaced plot in Figure 9; please check and correct.
AUTHOR'S RESPONSE
Again, we apologize for mistakenly repeating the water table level figure (figure 5) that is meant for soil respiration in Figure 9 (please find attached). We will correct the error by updating the soil respiration figure.
-
AC2: 'Reply on RC2', Oluwabamise Afolabi, 26 May 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 236 | 77 | 21 | 334 | 27 | 13 | 28 |
- HTML: 236
- PDF: 77
- XML: 21
- Total: 334
- Supplement: 27
- BibTeX: 13
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a process-based modelling study of carbon dynamics in a temperate swamp, based on extensive field measurements collected over a 40-year period. If the model demonstrates successful calibration and validation against consistent observations of abiotic variables (soil temperature, soil moisture, water table level, and heat fluxes) and carbon fluxes (soil respiration and GPP), the study could make a substantial contribution to our understanding of the long-term responses of ecosystem functioning—specifically water and carbon exchange—to recent climate variability, including both seasonal and interannual changes. It also has the potential to identify the key environmental drivers involved.
However, the current version of the manuscript does not meet the quality standards required for Biogeosciences, primarily due to the lack of sufficient detail regarding model calibration of carbon fluxes and poor model performance. I recommend that the authors address the following aspects:
(1) Presentation of Results
Figure 1: Please provide more detailed information on the specific measurements collected at each sampling site.
Figure 2: Why is ground heat shown at the bottom of the soil profile? It should be included as part of the surface energy balance. Additionally, if the model includes two tree species, the authors should explain how multiple canopies are represented. I suggest showing parameters such as species-specific height, root depth, leaf area index (LAI), and litterfall for both trees and peatland vegetation.
Figure 3: What does the red line represent? Is it a moving average?
Figures 4a, 4b, and 5a: Please clarify the meaning of the red line and the grey shaded area—do these represent a regression line and confidence interval, respectively?
Figures 8b and 9: The model appears to underestimate soil moisture and water table levels, while simulated evapotranspiration (ET) seems to be overestimated. I recommend revisiting the water balance and interpreting these model biases.
(2) Missing Calibration for Ecosystem Respiration
The manuscript appears to lack a figure or discussion on the model calibration results for ecosystem respiration. This is a critical component for evaluating model performance and should be included.
(3) Interannual Variability in Soil Respiration
Soil respiration from June to September shows a positive trend over time, while GPP and air temperature do not show any clear trends. Can the authors explain what might be driving the increased soil respiration during the growing season?
Minor Comments
Abstract, line 17: If negative NEE is defined as a carbon sink, then a decline in NEE (i.e., becoming more negative) suggests increased carbon uptake. Please clarify this wording.
Line 35: A comma is missing after “Meanwhile”.
Line 45: "> 5 years" should be corrected to "< 5 years".
Line 48: The phrase “capture the impact of regional multidecadal disturbances (e.g., climate change)” could be improved to “capture regional ecosystem responses to multidecadal climate change”.
Line 67: Consider rephrasing to clarify: “simulate key plant processes that regulate water, energy, and carbon fluxes in temperate swamps.”
Line 385: Could this result be related to a bias in simulated evapotranspiration?
Lines 446–454: I suggest presenting the modelled LAI and linking it to the discussion in this section.
Line 495: The decoupling phenomenon typically occurs under very dry or very wet conditions. The authors should support this discussion with simulated GPP or stomatal conductance results.
Lines 755–760: A reference appears to be duplicated