the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Jan 2024
Optimizing the terrestrial ecosystem gross primary productivity using carbonyl sulfide (COS) within a “two-leaf” modeling framework
Abstract. Accurately modeling gross primary productivity (GPP) is of great importance in diagnosing terrestrial carbon-climate feedbacks. Process-based terrestrial ecosystem models are often subject to substantial uncertainties, primarily attributed to inadequately calibrated parameters. Recent attention has identified carbonyl sulfide (COS) as a promising proxy of GPP, due to the close linkage between leaf exchange of COS and carbon dioxide (CO2) through their shared pathway of stomatal diffusion. However, most of the current modeling approaches for COS and CO2 did not explicitly consider the vegetation structural impacts, i.e. the differences between the sun-shade and sunlit leaves in COS uptake. This study used ecosystem COS fluxes data from 7 sites to optimize GPP estimation across various ecosystems with the Boreal Ecosystem Productivity Simulator (BEPS), which was further developed for simulating the leaf COS uptake under its state-of-the-art ‘two-leaf’ framework. Our results demonstrated the substantial improvement in GPP simulation across various ecosystems through the fusion of COS data into the ‘two-leaf’ model, with the ensemble mean of root mean square error (RMSE) for simulated GPP reduced by 18.99 % to 66.64 %. Notably, we also shed light on the remarkable identifiability of key parameters within the BEPS model, including the maximum carboxylation rate of Rubisco at 25 °C (Vcmax25), minimum stomatal conductance (bH2O), and leaf nitrogen content (Nleaf), despite intricate interactions among COS-related parameters. Furthermore, our global sensitivity analysis delineated both shared and disparate sensitivities of COS and GPP to model parameters and suggested the unique treatment of parameters for each site in COS and GPP modeling. In summary, our study deepened insights into the sensitivity, identifiability, and interactions of parameters related to COS, and showcased the efficacy of COS in reducing uncertainty in GPP simulations.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(4028 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4028 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-3032', Anonymous Referee #1, 19 Feb 2024
- AC1: 'Reply on RC1', Huajie Zhu, 10 Apr 2024
-
RC2: 'Comment on egusphere-2023-3032', Anonymous Referee #2, 23 Feb 2024
The authors have modelled vegetation and soil COS fluxes within the “two-leaf” version of the BEPS (Boreal Ecosystem Productivity Simulator) model. Then, they used observations of COS fluxes at seven sites and a Monte-Carlo approach to reduce parameter uncertainty in BEPS. They further evaluate the impact on GPP, and discuss parameter identifiability.
The paper is well built and very neat, the results are clearly presented. Some further explanations are however needed, and a few outlooks would be welcome.Main comments
Abstract
L14-15: “However, most of the current modeling approaches for COS and CO2 did not explicitly consider the vegetation structural impacts, i.e. the differences between the sun-shade and sunlit leaves in COS uptake” -> It is a bit misleading that the authors bring forward such an argument, because they did not demonstrate in this paper the advantage of distinguishing between sunlit and shaded leaves. Why did not they show the impact of having a two-leaf model compared to a one flux model, as they were the first ones (to my knowledge) to use such a model? This would indeed have been a great achievement.
2.4.1 Parameter selection and sampling strategy
L154-155: “9 parameters were selected to be calibrated in this study” -> Why didn’t the authors perform a sensitivity analysis to select the most important parameters for COS and GPP? We are left with the impression that the selection was arbitrary, and we may fear that they have missed some important parameter.
L155: Table B1 should be placed in the main manuscript, it’s important to see here the detailed description of the parameters.
2.4.2 Selection of behavioral simulations
“Behavioral simulation” is not an expression I’ve seen before. Could the authors use simpler terms like “selected” and “rejected” (for “non-behavioral”)?
L168-169: “Thus, the deterministic model prediction is given by the ensemble mean of the 100 behavioral simulations.” -> The authors could explain that the “100” comes from 0.5%*20,000.
2.6 Parameter uncertainty
L183: “Due to the complexity of ecosystem” -> Could the authors be more specific: “Due to the functional and structural complexity of ecosystems”?
L193-194: “Taking into account the influence of the prior distribution to the behavioral parameter sets, the PI is defined as the reduction of the parameter range width. -> This means that if the initial range is overestimated, the PI may be artificially high. This could be the case for the bH2O parameter, where the max value (1) is 57 times larger than the initial value. Plus, the authors later write, citing Miner et al. (2017), that “83 % of the 𝑏𝐻2O values are located between 0 and 0.15 mol m−2 s−1, and about half are located between 0 and 0.04 mol m−2 s−1” (L236-237).
3.2 Posterior parameter distributions
L252: The authors should explain what they call “the grouping value”.
Figure2. The authors should add ‘COS’ somewhere in the legend, document the boxplot (say it describes the posterior distribution), and explain axes, colours, title (PI).
3.3 The optimization performance in COS fluxes
L300-301: “despite remarkable improvement is attached by the posterior simulations” -> This is a weird formulation, to be rephrased.
Figure 3/Figure 4: “The means and uncertainties of these observations and simulations are calculated and plotted on a daily or monthly scale” -> Do the authors compute the standard deviation of hourly values for daily means and over daily means for monthly means? Do they compute the standard error of the mean (SEM), defined as the standard deviation (SD) divided by the square root of the number of observations, and which would be more appropriate than SD to estimate the uncertainty of the mean?
4.2 Parameter interactions
L407: “their weak equivalence” -> What do the authors mean? Equivalence to what?
Figure 6 is a bit difficult to interpret, I’m not sure it brings something, could it be moved to the Supplementary part?
4.3 Parameter identifiability
L442: “the sensitivity of the input data to the parameter” -> This should rather be “the sensitivity of the modeled output to the parameter”.
L446-447: “However, our findings indicate that the sensitivity of 𝑉𝑐max25, 𝑁𝑙eaf is much greater than that of 𝑏𝐻2O, yet the latter is much more identifiable” -> An alternative explanation is once again the overestimated prior range of 𝑏𝐻2O.
L448: “as parameter interaction is a primary contributor to parameter unidentifiability” -> But then, this should also apply to 𝑏𝐻2O, as it is highly correlated to fleaf and mH2O, as shown in Figure 5.
L456-457: “It has been previously demonstrated that soil hydrology-related parameters exert a minimal impact on COS simulations and cannot be effectively constrained through COS assimilation” -> That would depend on whether soil water stress conditions are present or not.
4.4 Relationship between COS and GPP simulation performance -> performances
L464: “respond to RMSE” -> This seems awkward, to be rephrased.
Figure 7: “Each data point represents a parameter set, with color indicating data density” -> That does not seem possible, some binning has to be made to get a density.
5 Conclusions
L485-486: “within the Monte Carlo-based methodology base on the coupling of COS modeling and the BEPS model” -> “with a Monte Carlo approach using COS modeling within BEPS”
L486-487: “Global parameter sensitivity analysis was conducted to identify the sensitive parameters” -> “A global parameter sensitivity analysis was conducted to identify the most sensitive ones among a set of 9 pre-selected parameters.”
The conclusion is a bit abrupt. The authors should develop some outlooks. What are the consequences of this study? Is there a need to acquire more COS fluxes observations, or a need to improve the COS vegetation model? What will be the next steps with BEPS?
A2 BEPS leaf COS modeling approach
L568: “where COS𝑎 represents the COS mole fraction in the bulk air” -> Did the authors use a variable atmospheric COS mole fraction as it has been shown important (Kooijmans et al., 2021; Abadie et al., 2022)?
L572: How did the authors derive the empirical relationship expressed in equation (A19)?
Minor comments
L15: “i.e.” -> “i.e.,”
L15: I could not find information on the “sun-shade” expression, would not the simpler “shaded” be more appropriate?
L31: “(GPP), is” -> “(GPP) is”
L33: “the modeling of GPP are affected” -> “the modeling of GPP is affected”
L60: “Ecosystem carbon, water and energy processes are interacted” -> “Ecosystem carbon, water and energy processes are interacting”
L64: “e.g.” -> “e.g.,”
L67: “Which parameters the COS simulation is sensitive to” -> “To which parameters is the COS simulation sensitive”
L84: “calculated” -> “calculates” (harmonize verb tenses.)
L111: “locations” -> “Locations” (capital letter L)
L116: “three” -> “two” (I see only GLOBMAP and GLASS.)
L163: “in in” -> “in”
L168, 387: “In specific” -> “Specifically”
L181: “if all model parameters is considered” -> “if all model parameters are considered”
L184: “compensating with each other” -> “compensating each other”
L193-194: “Taking into account the influence of the prior distribution to the behavioral parameter sets” -> “Taking into account the influence of the prior distribution of the behavioral parameter sets”
L220 (twice): Fig. 1 -> Fig. 2
L225: “parameters related energy balance” -> “parameters related to energy balance’
L281: “Fig. 2” -> “Fig. 3”
L287: “e.g. Fig. 2d” -> “e.g., Fig. 3d”
L297: “a further underestimate of the” -> “a further underestimation of the”
L307: “the ensemble mean deviate remarkable from observations” -> “the ensemble mean strongly deviates from the observations”
L320, 352: “posterior (green)” -> It seems purple.
L322, 354: “blue dots” -> -> It seems gray.
L334: “Fig. 3” -> “Fig. 4”
L363-364: “by influence the modeling of stomatal conductance” -> “by influencing the modeling of the stomatal conductance”
L384: “in photosynthetic machinery” -> “in the photosynthetic machinery”
L397: “confident levels” -> “confidence levels”
L399: “A total of 14 parameter combinations demonstrate significantly correlated” -> “A total of 14 parameter combinations demonstrate significant correlations”
L411, 596: The red font looks weird, like mixed with a black one, could the authors improve that?
L419: “e.g.” -> “e.g.,”
L450: “exhibits low sensitivity” -> “exhibits a low sensitivity”
L468: “for COS simulation” -> “for COS simulations”
L475: “such as that” -> “for example considering that”
L478: “e.g.” -> “e.g.,”
L495: “interactions exists” -> “interactions exist”
L495-496: “In particularly” -> “Particularly” or “In particular”
L519: “according the” -> “according to the”
L525: In the first exponential of equation (A7), “Kn” should be “kn”.
L530: “is the is the” -> “is the”
L535: “(𝑔𝑠w in)” -> The unit is missing.
L538: “is intercept” -> “is the intercept”
L549: “the number of soil layer” -> “the number of soil layers”
L572: In equation (A19), shouldn’t “LAI” be “L” as in equations (A6/7)?
L590: “of the 9 parameters were” -> “of the 9 parameters that were”
L591: “to the parameter dependent” -> “to the parameter dependency”
L620: “Reference" -> “References”Citation: https://doi.org/10.5194/egusphere-2023-3032-RC2 - AC2: 'Reply on RC2', Huajie Zhu, 10 Apr 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-3032', Anonymous Referee #1, 19 Feb 2024
- AC1: 'Reply on RC1', Huajie Zhu, 10 Apr 2024
-
RC2: 'Comment on egusphere-2023-3032', Anonymous Referee #2, 23 Feb 2024
The authors have modelled vegetation and soil COS fluxes within the “two-leaf” version of the BEPS (Boreal Ecosystem Productivity Simulator) model. Then, they used observations of COS fluxes at seven sites and a Monte-Carlo approach to reduce parameter uncertainty in BEPS. They further evaluate the impact on GPP, and discuss parameter identifiability.
The paper is well built and very neat, the results are clearly presented. Some further explanations are however needed, and a few outlooks would be welcome.Main comments
Abstract
L14-15: “However, most of the current modeling approaches for COS and CO2 did not explicitly consider the vegetation structural impacts, i.e. the differences between the sun-shade and sunlit leaves in COS uptake” -> It is a bit misleading that the authors bring forward such an argument, because they did not demonstrate in this paper the advantage of distinguishing between sunlit and shaded leaves. Why did not they show the impact of having a two-leaf model compared to a one flux model, as they were the first ones (to my knowledge) to use such a model? This would indeed have been a great achievement.
2.4.1 Parameter selection and sampling strategy
L154-155: “9 parameters were selected to be calibrated in this study” -> Why didn’t the authors perform a sensitivity analysis to select the most important parameters for COS and GPP? We are left with the impression that the selection was arbitrary, and we may fear that they have missed some important parameter.
L155: Table B1 should be placed in the main manuscript, it’s important to see here the detailed description of the parameters.
2.4.2 Selection of behavioral simulations
“Behavioral simulation” is not an expression I’ve seen before. Could the authors use simpler terms like “selected” and “rejected” (for “non-behavioral”)?
L168-169: “Thus, the deterministic model prediction is given by the ensemble mean of the 100 behavioral simulations.” -> The authors could explain that the “100” comes from 0.5%*20,000.
2.6 Parameter uncertainty
L183: “Due to the complexity of ecosystem” -> Could the authors be more specific: “Due to the functional and structural complexity of ecosystems”?
L193-194: “Taking into account the influence of the prior distribution to the behavioral parameter sets, the PI is defined as the reduction of the parameter range width. -> This means that if the initial range is overestimated, the PI may be artificially high. This could be the case for the bH2O parameter, where the max value (1) is 57 times larger than the initial value. Plus, the authors later write, citing Miner et al. (2017), that “83 % of the 𝑏𝐻2O values are located between 0 and 0.15 mol m−2 s−1, and about half are located between 0 and 0.04 mol m−2 s−1” (L236-237).
3.2 Posterior parameter distributions
L252: The authors should explain what they call “the grouping value”.
Figure2. The authors should add ‘COS’ somewhere in the legend, document the boxplot (say it describes the posterior distribution), and explain axes, colours, title (PI).
3.3 The optimization performance in COS fluxes
L300-301: “despite remarkable improvement is attached by the posterior simulations” -> This is a weird formulation, to be rephrased.
Figure 3/Figure 4: “The means and uncertainties of these observations and simulations are calculated and plotted on a daily or monthly scale” -> Do the authors compute the standard deviation of hourly values for daily means and over daily means for monthly means? Do they compute the standard error of the mean (SEM), defined as the standard deviation (SD) divided by the square root of the number of observations, and which would be more appropriate than SD to estimate the uncertainty of the mean?
4.2 Parameter interactions
L407: “their weak equivalence” -> What do the authors mean? Equivalence to what?
Figure 6 is a bit difficult to interpret, I’m not sure it brings something, could it be moved to the Supplementary part?
4.3 Parameter identifiability
L442: “the sensitivity of the input data to the parameter” -> This should rather be “the sensitivity of the modeled output to the parameter”.
L446-447: “However, our findings indicate that the sensitivity of 𝑉𝑐max25, 𝑁𝑙eaf is much greater than that of 𝑏𝐻2O, yet the latter is much more identifiable” -> An alternative explanation is once again the overestimated prior range of 𝑏𝐻2O.
L448: “as parameter interaction is a primary contributor to parameter unidentifiability” -> But then, this should also apply to 𝑏𝐻2O, as it is highly correlated to fleaf and mH2O, as shown in Figure 5.
L456-457: “It has been previously demonstrated that soil hydrology-related parameters exert a minimal impact on COS simulations and cannot be effectively constrained through COS assimilation” -> That would depend on whether soil water stress conditions are present or not.
4.4 Relationship between COS and GPP simulation performance -> performances
L464: “respond to RMSE” -> This seems awkward, to be rephrased.
Figure 7: “Each data point represents a parameter set, with color indicating data density” -> That does not seem possible, some binning has to be made to get a density.
5 Conclusions
L485-486: “within the Monte Carlo-based methodology base on the coupling of COS modeling and the BEPS model” -> “with a Monte Carlo approach using COS modeling within BEPS”
L486-487: “Global parameter sensitivity analysis was conducted to identify the sensitive parameters” -> “A global parameter sensitivity analysis was conducted to identify the most sensitive ones among a set of 9 pre-selected parameters.”
The conclusion is a bit abrupt. The authors should develop some outlooks. What are the consequences of this study? Is there a need to acquire more COS fluxes observations, or a need to improve the COS vegetation model? What will be the next steps with BEPS?
A2 BEPS leaf COS modeling approach
L568: “where COS𝑎 represents the COS mole fraction in the bulk air” -> Did the authors use a variable atmospheric COS mole fraction as it has been shown important (Kooijmans et al., 2021; Abadie et al., 2022)?
L572: How did the authors derive the empirical relationship expressed in equation (A19)?
Minor comments
L15: “i.e.” -> “i.e.,”
L15: I could not find information on the “sun-shade” expression, would not the simpler “shaded” be more appropriate?
L31: “(GPP), is” -> “(GPP) is”
L33: “the modeling of GPP are affected” -> “the modeling of GPP is affected”
L60: “Ecosystem carbon, water and energy processes are interacted” -> “Ecosystem carbon, water and energy processes are interacting”
L64: “e.g.” -> “e.g.,”
L67: “Which parameters the COS simulation is sensitive to” -> “To which parameters is the COS simulation sensitive”
L84: “calculated” -> “calculates” (harmonize verb tenses.)
L111: “locations” -> “Locations” (capital letter L)
L116: “three” -> “two” (I see only GLOBMAP and GLASS.)
L163: “in in” -> “in”
L168, 387: “In specific” -> “Specifically”
L181: “if all model parameters is considered” -> “if all model parameters are considered”
L184: “compensating with each other” -> “compensating each other”
L193-194: “Taking into account the influence of the prior distribution to the behavioral parameter sets” -> “Taking into account the influence of the prior distribution of the behavioral parameter sets”
L220 (twice): Fig. 1 -> Fig. 2
L225: “parameters related energy balance” -> “parameters related to energy balance’
L281: “Fig. 2” -> “Fig. 3”
L287: “e.g. Fig. 2d” -> “e.g., Fig. 3d”
L297: “a further underestimate of the” -> “a further underestimation of the”
L307: “the ensemble mean deviate remarkable from observations” -> “the ensemble mean strongly deviates from the observations”
L320, 352: “posterior (green)” -> It seems purple.
L322, 354: “blue dots” -> -> It seems gray.
L334: “Fig. 3” -> “Fig. 4”
L363-364: “by influence the modeling of stomatal conductance” -> “by influencing the modeling of the stomatal conductance”
L384: “in photosynthetic machinery” -> “in the photosynthetic machinery”
L397: “confident levels” -> “confidence levels”
L399: “A total of 14 parameter combinations demonstrate significantly correlated” -> “A total of 14 parameter combinations demonstrate significant correlations”
L411, 596: The red font looks weird, like mixed with a black one, could the authors improve that?
L419: “e.g.” -> “e.g.,”
L450: “exhibits low sensitivity” -> “exhibits a low sensitivity”
L468: “for COS simulation” -> “for COS simulations”
L475: “such as that” -> “for example considering that”
L478: “e.g.” -> “e.g.,”
L495: “interactions exists” -> “interactions exist”
L495-496: “In particularly” -> “Particularly” or “In particular”
L519: “according the” -> “according to the”
L525: In the first exponential of equation (A7), “Kn” should be “kn”.
L530: “is the is the” -> “is the”
L535: “(𝑔𝑠w in)” -> The unit is missing.
L538: “is intercept” -> “is the intercept”
L549: “the number of soil layer” -> “the number of soil layers”
L572: In equation (A19), shouldn’t “LAI” be “L” as in equations (A6/7)?
L590: “of the 9 parameters were” -> “of the 9 parameters that were”
L591: “to the parameter dependent” -> “to the parameter dependency”
L620: “Reference" -> “References”Citation: https://doi.org/10.5194/egusphere-2023-3032-RC2 - AC2: 'Reply on RC2', Huajie Zhu, 10 Apr 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 423 | 87 | 33 | 543 | 22 | 21 |
- HTML: 423
- PDF: 87
- XML: 33
- Total: 543
- BibTeX: 22
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 209 | 39 |
| China | 2 | 124 | 23 |
| France | 3 | 26 | 4 |
| Netherlands | 4 | 20 | 3 |
| Germany | 5 | 17 | 3 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 209
Huajie Zhu
Xiuli Xing
Mousong Wu
Weimin Ju
Fei Jiang
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4028 KB) - Metadata XML