Modeling demographic-driven vegetation dynamics and ecosystem biogeochemical cycling in NASA GISS&rsquo;s Earth system model (ModelE-BiomeE v.1.0)

Weng, Ensheng; Aleinov, Igor; Singh, Ram; Puma, Michael J.; McDermid, Sonali S.; Kiang, Nancy Y.; Kelley, Maxwell; Wilcox, Kevin; Dybzinski, Ray; Farrior, Caroline E.; Pacala, Stephen W.; Cook, Benjamin I.

doi:https://doi.org/10.5194/gmd-2022-72

Preprints

https://doi.org/10.5194/gmd-2022-72

Preprints

Submitted as: model description paper

18 Mar 2022

Submitted as: model description paper | 18 Mar 2022

Status: a revised version of this preprint was accepted for the journal GMD.

Modeling demographic-driven vegetation dynamics and ecosystem biogeochemical cycling in NASA GISS’s Earth system model (ModelE-BiomeE v.1.0)

Ensheng Weng^1,2, Igor Aleinov^1,2, Ram Singh^1,2, Michael J. Puma^1,2, Sonali S. McDermid³, Nancy Y. Kiang², Maxwell Kelley², Kevin Wilcox⁴, Ray Dybzinski⁵, Caroline E. Farrior⁶, Stephen W. Pacala⁷, and Benjamin I. Cook² Ensheng Weng et al.

Show author details

¹Center for Climate Systems Research, Columbia University, New York, NY 10025, USA
²NASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025, USA
³Department of Environmental Studies, New York University, New York, NY 10003, USA
⁴Department of Ecosystem Science and Management, University of Wyoming, Laramie, WY 12 82071, USA
⁵Institute of Environmental Sustainability, Loyola University Chicago, Chicago, IL 60660, USA
⁶Department of Integrative Biology, University of Texas at Austin, Austin, TX 78712, USA
⁷Department of Ecology & Evolutionary Biology, Princeton University, Princeton, NJ 08544, USA

Received: 10 Mar 2022 – Discussion started: 18 Mar 2022

Abstract. We developed a new demographic vegetation model, BiomeE, to improve the representation of vegetation demographic dynamics and ecosystem biogeochemical cycles in the NASA Goddard Institute of Space Studies’ ModelE Earth system model. This model includes the processes of plant growth, mortality, reproduction, vegetation structural dynamics, and soil carbon and nitrogen storage and transformations. The model combines the plant physiological processes of ModelE’s original vegetation model, Ent, with minor adaptations to fit the new allometry and vegetation structure with the plant demographic and ecosystem nitrogen processes represented from Geophysical Fluid Dynamics Laboratory (GFDL)’s LM3-PPA. For global applications, we added a new set of plant functional types to represent global vegetation functional diversity, including trees, shrubs, and grasses, and a new phenology model to deal with seasonal changes in temperature and soil water availability. Competition for light and soil resources is individual based, which makes the modeling of transient compositional changes and vegetation succession possible. BiomeE will allow ModelE to simulate long-term biogeophysical and biogeochemical feedbacks between the climate system and land ecosystems. BiomeE simulates, with fidelity comparable to other models, the dynamics of vegetation and soil biogeochemistry, including leaf area index, vegetation structure (e.g., height, tree density, size distribution, crown organization), and ecosystem carbon and nitrogen storage and fluxes. Further, BiomeE will also allow for the simulations of transient vegetation dynamics and eco-evolutionary optimal community assemblage in response to past and future climate changes by incorporating core ecological processes, including demography, competition, and community assembly.

Ensheng Weng et al.

Status: final response (author comments only)

CEC1:
'Comment on gmd-2022-72', Juan Antonio Añel, 21 Apr 2022

Dear authors,
After checking your manuscript, it has come to our attention that it does not comply with our Code and Data Policy.

https://www.geoscientific-model-development.net/policies/code_and_data_policy.html

You have archived your code on GitHub. However, GitHub is not a suitable repository. GitHub itself instructs authors to use other alternatives for long-term archival and publishing, such as Zenodo (GitHub provides a direct way to copy your project to a Zenodo repository). Therefore, please, publish your code in one of the appropriate repositories. Also, in the GitHub repository, there is no license listed. If you do not include a license, the code continues to be your property and can not be used by others. Therefore, when uploading the model's code to the repository, you could want to choose a free software/open-source (FLOSS) license. We recommend the GPLv3. You only need to include the file 'https://www.gnu.org/licenses/gpl-3.0.txt' as LICENSE.txt with your code. Also, you can choose other options that Zenodo provides: GPLv2, Apache License, MIT License, etc.

Moreover, you state that the model code and data will be available after publication. First, your work is already published in GMD Discussions (though it is true that the topical editor oversaw this, as it should not have been published before solving the issues here pointed out). Secondly, we do not accept this. Our policy clearly states that all the code and data must be public during the review process; otherwise, anyone who wants to review your work and contribute to the public discussion can not do it.

Please, be aware that you must include in a potential reviewed version of your manuscript the modified 'Code and Data Availability' section and the DOI of the repositories.

Please, reply as soon as possible to this comment with the link for the code and data. In this way, the materials will be available for the peer-review process and the Discussions stage, as they must be.

And please, be aware that failing to comply with it can result in the rejection of your paper for publication.
Best regards,
Juan A. Añel

Geosci. Model Dev. Exec. Editor

Citation: https://doi.org/10.5194/gmd-2022-72-CEC1
- AC1: 'Reply on CEC1', Ensheng Weng, 25 Apr 2022
  
  Thanks for the comments.
  We have included a GPLv3 licence file by including 'https://www.gnu.org/licenses/gpl-3.0.txt and made the codes publicly available at https://doi.org/10.5281/zenodo.6476152.
  The simulated data have been archived at Zenodo (https://doi.org/10.5281/zenodo.6480411).
  The 'Code and Data Availability' section has been revised accordingly. We will submit it at the request of a revised version.
  Best regards,
  Ensheng Weng (on behalf of all coauthors)
  
  Citation: https://doi.org/10.5194/gmd-2022-72-AC1
RC1:
'Comment on gmd-2022-72', Anonymous Referee #1, 23 Apr 2022

This manuscript by Weng et al reported a model improvement. I am happy to read it and know many great improvements. However, there are some unclear things that need improve before publication.

Line 47-49: this sentence is not necessary.

Line 86-99: should this paragraph be moved to model introduction section?

Line 213-214: how to determine soil moisture threshold is quite important. However, it is difficult to understand how to determine this threshold. the authors need introduce more about this.

Line 219: what is structural biomass? What are the parameters of αc, αz, θc and θz.

For Eq. (5), the most important thing is how to simulate D?

Line 231: is that correct to keep a minimum growth rate of stems? Why do the authors set the equation like this?

Line 232-235: it is difficult to understand how the authors simulate carbon allocation. Carbon allocation is quite hard to simulate indeed, and especially for demo-type model, the authors should pay more attention to impacts stand age on carbon allocation. Please refer the Xia et al. 2019.

Line 246 “Reproduction and Mortality”. They are very important and hard to simulate. I am happy to see the authors made great contributions.

Line 258: cannot understand “U-shape” curve?

Eq. (9) it will be better to introduce the basic principle.

Figure 2. it will be helpful to give name of each vegetation type in the figure caption. Did you include cropland?

Line 355: what do you mean “The interpolation of radiation”?

Figure 4. can you explain why there are sharp decreases of simulated height? I am also confused why the crown area index increased first and then decrease?

Figure 5. simulated LAI is not good enough, but I totally understand it is very hard task. You may discuss this issue, and especially point out how we should improve LAI simulations in the further studies.

Figure 7. how did the authors treat cropland? If the model scheme impacts this global comparison?

Figure 8. these results are surprised for me. I thought the model can simulate plant carbon better than soil carbon. But it seems that I am not correct. Would you like please to explain the reason for large uncertainties of plant carbon simulations?

Citation: https://doi.org/10.5194/gmd-2022-72-RC1
- AC2: 'Reply on RC1', Ensheng Weng, 20 May 2022
  
  Reviewer 1 asks us to re-arrange the description of GISS ModelE, and improve the presentation of the model and results. We will do this if we are asked to submit a revised version. Detailed responses are below:
  
  This manuscript by Weng et al reported a model improvement. I am happy to read it and know many great improvements. However, there are some unclear things that need improve before publication.
  Thanks!
  Line 47-49: this sentence is not necessary.
  Removed
  
  Line 86-99: should this paragraph be moved to model introduction section?
  Agreed. We will move it to the model description section if we are asked to submit a revised version.
  
  Line 213-214: how to determine soil moisture threshold is quite important. However, it is difficult to understand how to determine this threshold. the authors need introduce more about this.
  In this paper, we just slightly tuned this parameter. We will parameterize it with data for different PFTs in future. A plant hydraulic module is being developed. It is not necessary to tune it too much. We will add some explanations for it.
  
  Line 219: what is structural biomass? What are the parameters of αc, αz, θc and θz.
  Sapwood plus heartwood. We will add the notation to these parameters in the text.
  
  For Eq. (5), the most important thing is how to simulate D?
  Yes. dD/dt (and its integral D(t)) is the process that links carbon fluxes to plant structure. It also bridges the traditional BGC model (fluxes and poots) to the demographic models that explicitly simulate three-dimensional growth of trees.
  We have a paragraph to explain it after Eq (7). We will clarify the meaning of predicting D in revised paper.
  
  Line 231: is that correct to keep a minimum growth rate of stems? Why do the authors set the equation like this?
  Yes. The purpose is to keep the consistency of the vascular system in leaves, sapwood and roots, though they are separated into three pools. We discussed this assumption in Weng et al. 2015 and Weng et al. 2019. A brief explanation will be added in the revised version.
  
  Line 232-235: it is difficult to understand how the authors simulate carbon allocation. Carbon allocation is quite hard to simulate indeed, and especially for demo-type model, the authors should pay more attention to impacts stand age on carbon allocation. Please refer the Xia et al. 2019.
  We will explain it clearly and put it in the background of other models, including Xia et al. 2019.
  
  Line 246 “Reproduction and Mortality”. They are very important and hard to simulate. I am happy to see the authors made great contributions.
  Thanks!
  
  Line 258: cannot understand “U-shape” curve?
  “U-shape” means high mortality rates at seedling and old trees. We will reword this sentence.
  
  Eq. (9) it will be better to introduce the basic principle.
  This equation delineates the mortality rate that changes with social status (crown layers), shade effects, and tree sizes. We will add this to the revised version.
  
  Figure 2. it will be helpful to give name of each vegetation type in the figure caption. Did you include cropland?
  Will do as suggested. Cropland is not included. We only simulated potential vegetation as represented by the 9 PFTs.
  
  Line 355: what do you mean “The interpolation of radiation”?
  The forcing data is at six-hour time step. We interpolate them into hourly/half-hourly time step. We will clarify it.
  
  Figure 4. can you explain why there are sharp decreases of simulated height? I am also confused why the crown area index increased first and then decrease?
  The sharp decrease in critical height indicates the transformation from even aged trees to mixed trees in canopy. “Critical height” is the shortest tree in canopy layer.
  We will clarify this model behavior in the revised version.
  
  Figure 5. simulated LAI is not good enough, but I totally understand it is very hard task. You may discuss this issue, and especially point out how we should improve LAI simulations in the further studies.
  Will do as suggested.
  
  Figure 7. how did the authors treat cropland? If the model scheme impacts this global comparison?
  We don’t include croplands in this study. We assumed the land is covered by potential vegetation that is represented by the 9 PFTs.
  
  Figure 8. these results are surprised for me. I thought the model can simulate plant carbon better than soil carbon. But it seems that I am not correct. Would you like please to explain the reason for large uncertainties of plant carbon simulations?
  
  We simulated the potential vegetation biomass. However, the data of biomass include effects of land use and disturbance. Soil carbon is disturbed too. From our simulations, biomass was far more away from equilibrium state. We will clarify it in the revised version.
  
  Citation: https://doi.org/10.5194/gmd-2022-72-AC2
RC2:
'Comment on gmd-2022-72', Anonymous Referee #2, 12 May 2022
The authors present a model development work on vegetation demographics and seek to implement it into an Earth system model. The new model features include a greater diversity of global plant functional types and a new phenological scheme. They also compare the behaviour of this model to the eight observational locations and six MsTMIP simulations. As seen from the results within the paper it is possible to capture vegetation structure and dynamics reasonably even within such a parasimonious model at a global scale. Overall, this work is well structured and is useful in highlighting some new issues in improving the representation of the terrestrial carbon cycle within ESMs. I think the manuscript could be publishable after some major revisions. I have a few comments below.

Major comments:

One of the most representative feature of this model is the fulldemographic processes. The authors mainly compare the differences between the simulations of the full demography and the single cohort settings of BiomeE. I think authors should separately compare simulations of these two versions of BiomE with observations, if possible, to show the advantages of full demography in reproducing ecosystem dynamics (Figs. 6-10).

Lines 157-158: “A set of continuous plant traits are used to define the distinctive plant types”. Please specify the continuous trait assignments for plant functional types, especially the differences with traditional PFT-based model.

To represent the major variations in plant functional diversity, the authorschose four plant traits as the primary axes to define PFTs: wood density, leaf mass per unit area (LMA), height growth parameter, and leaf maximum carboxylation rate (cmax). I would suggest a sensitivity analysis of these plant traits to different ecosystem functions, which would be very instructive for futher model improvement and localization.

Methods Line 184: it is unclear to me about the assumptions in the phenological scheme. Why to define the nine PFTs as these four phenological types? Regarding the comparative advantage and competitiveness of deciduous vs. evergreen trees, are there any basic theories that evergreen species are more resistant to cold and drought than deciduous tree species? According to the global vegetation distribution, evergreen broadleaf species are usually distributed in warm and moist environments. What kinds of functional traits suggest that evergreen species are adaptive under water limitation and cold conditions?

Methods Line 254: PFT-specific parameterisations for the mortality parameter are used, so are there different PFTs each with their own cohort structure? How are PFT-specific background mortality parameters set in the model? Are they all come from observations across different vegetation types? Related reference is missing in the main text. Since the most size-dependent mortality research focus on closed-canopy forest system, whether the “U-shaped mortality pattern” can be extended to other vegetation systems, including forbs, shrubs, grasslands, systems with open canopies and systems experiencing different risks in different environment?

Methods Line 374: how does disturbance history set in the MsTMIP simulations? I’m wondering whether the large inter-model discrepancy in simulating plant biomass is caused by disturbance dynamics? For clarity, can the authors be a bit more explicit about the experimental design of the MsTMIP.

Figure 4d: the authors point out that model analyses are based on equilibrium simulations without explicit disturbances. But the critical height across forests shows an abrupt decrease in the 100 years of model run. What reasons made this pattern happen in the model? Is that driven by the aging-related mortality of canopy trees? Could you discuss more the underlying mechanisms behind the emergent ecological phenomena?

Result Line 515-517: the formulation of allometry makes the tree height growth as a function of tree diameter (Eq. 5 in the main text). Since the two model versions have similar stem growth and tree size distribution, I would assume that tree height growth is stable as well. Why the full demography model shows higher tree height than the single-cohort model (Figure 11c)?

Result: the authors evaluate the model outputs with the MsTMIP simulations in the Result section. The simple intercomparison would be invaluable to help determine which model behaviour is more realistic. I think it would be interesting to have a section in the discussion tracing the variability that emerges among the models and informing what modeling structural choices or assumptions lead to improved model estimates. Since this paper is a model description paper, further discussion by model developers on the potential reasons for the biases would be much appreciated.

Minor comments:

The abbreviation of the term CAI on Line 409 should be put in parentheses for the first time on Line 407.

Lines 71-75: it is unclear to me what is “the legacy of land models and the technical requirements of reversibility in model development”? Could you explain or rephrase this sentence?

Lines 225, “H is tree height”should be modified to “Z is tree height”.

In equation(10), is ground area? Not defined.

Figure 3. How to understand the constant LAI value of KZ?

Figure 9. Please add units of LAI.
Citation: https://doi.org/10.5194/gmd-2022-72-RC2
- AC3:
  'Reply on RC2', Ensheng Weng, 20 May 2022
  Reviewer 2 asks us to do a detailed analysis of the comparison between the full demography and single cohort settings, test the sensitivity of PFT key parameters, improve the description of phenology, mortality, and clarify some model behavior. We can do these in a revised version. Detailed responses are blow:
  
  The authors present a model development work on vegetation demographics and seek to implement it into an Earth system model. The new model features include a greater diversity of global plant functional types and a new phenological scheme. They also compare the behaviour of this model to the eight observational locations and six MsTMIP simulations. As seen from the results within the paper it is possible to capture vegetation structure and dynamics reasonably even within such a parsimonious model at a global scale. Overall, this work is well structured and is useful in highlighting some new issues in improving the representation of the terrestrial carbon cycle within ESMs. I think the manuscript could be publishable after some major revisions. I have a few comments below.
  Thanks!
  
  Major comments:
  One of the most representative feature of this model is the full demographic processes. The authors mainly compare the differences between the simulations of the full demography and the single cohort settings of BiomeE. I think authors should separately compare simulations of these two versions of BiomeE with observations, if possible, to show the advantages of full demography in reproducing ecosystem dynamics (Figs. 6-10).
  
  We will do more analysis for the single-cohort model in the revised version if we are asked by the editor to resubmit this manuscript.
  
  Lines 157-158: “A set of continuous plant traits are used to define the distinctive plant types”. Please specify the continuous trait assignments for plant functional types, especially the differences with traditional PFT-based model.
  
  They are in table one. We will explain it clearly in the revised version.
  
  To represent the major variations in plant functional diversity, the authors chose four plant traits as the primary axes to define PFTs: wood density, leaf mass per unit area (LMA), height growth parameter, and leaf maximum carboxylation rate (Vcmax). I would suggest a sensitivity analysis of these plant traits to different ecosystem functions, which would be very instructive for further model improvement and localization.
  
  I agree with this suggestion. We will do the sensitivity analysis at one site to show model behavior.
  
  Methods Line 184: it is unclear to me about the assumptions in the phenological scheme. Why to define the nine PFTs as these four phenological types? Regarding the comparative advantage and competitiveness of deciduous vs. evergreen trees, are there any basic theories that evergreen species are more resistant to cold and drought than deciduous tree species? According to the global vegetation distribution, evergreen broadleaf species are usually distributed in warm and moist environments. What kinds of functional traits suggest that evergreen species are adaptive under water limitation and cold conditions?
  
  We only defined the possible factorial combinations of drought and cold deciduous, but did not discuss who will be more competitive. It is possible that the evergreen would be more competitive in high seasonality regions (e.g., evergreen in boreal regions), though the first response of plants to hash environments (e.g., cold or dry) is to shed off their leaves. Our simplified definition of phenology in here is to make it possible to evaluate the competitively optimal strategy in future studies. We will add a detailed explanation in this section.
  
  Methods Line 254: PFT-specific parameterisations for the mortality parameter are used, so are there different PFTs each with their own cohort structure? How are PFT-specific background mortality parameters set in the model? Are they all come from observations across different vegetation types? Related reference is missing in the main text. Since the most size-dependent mortality research focus on closed-canopy forest system, whether the “U-shaped mortality pattern” can be extended to other vegetation systems, including forbs, shrubs, grasslands, systems with open canopies and systems experiencing different risks in different environment?
  
  The mortality is delineated by the Eq. 9. We just fixed the default mortality rate according to the general mortality patterns of trees in the forests across the world. We didn’t extensively tune the parameters. We will make the mortality settings clear and validate with observational studies.
  
  Methods Line 374: how does disturbance history set in the MsTMIP simulations? I’m wondering whether the large inter-model discrepancy in simulating plant biomass is caused by disturbance dynamics? For clarity, can the authors be a bit more explicit about the experimental design of the MsTMIP.
  
  According to Huntzinger et al. 2013, MsTMIP only provided prescribed land use types. It is up to the participant models for the disturbances. We will clarify it.
  
  Figure 4d: the authors point out that model analyses are based on equilibrium simulations without explicit disturbances. But the critical height across forests shows an abrupt decrease in the 100 years of model run. What reasons made this pattern happen in the model? Is that driven by the aging-related mortality of canopy trees? Could you discuss more the underlying mechanisms behind the emergent ecological phenomena?
  
  It is a behavior of the model. The simulated forest is transforming from even aged to the mixed aged. In canopy layer, the trees are gradually replaced by younger trees as the old trees yield their space due to mortality. We will explain it in the revised version.
  
  Result Line 515-517: the formulation of allometry makes the tree height growth as a function of tree diameter (Eq. 5 in the main text). Since the two model versions have similar stem growth and tree size distribution, I would assume that tree height growth is stable as well. Why the full demography model shows higher tree height than the single-cohort model (Figure 11c)?
  
  For the full demography model, the height is from the tallest tree. For the single cohort model, all the trees have the same height. We will clarify this.
  
  Result: the authors evaluate the model outputs with the MsTMIP simulations in the Result section. The simple intercomparison would be invaluable to help determine which model behaviour is more realistic. I think it would be interesting to have a section in the discussion tracing the variability that emerges among the models and informing what modeling structural choices or assumptions lead to improved model estimates. Since this paper is a model description paper, further discussion by model developers on the potential reasons for the biases would be much appreciated.
  
  We will add the discussions about why these models perform differently, especially focusing on the assumptions of our model, BiomeE. Turnover rates and NPP are two key factors affecting vegetation dynamics. For the land models, they treat “turnover” in a high variation.
  
  Minor comments:
  The abbreviation of the term CAI on Line 409 should be put in parentheses for the first time on Line 407.
  
  Crown area index (CAI). Will add it to the text.
  
  Lines 71-75: it is unclear to me what is “the legacy of land models and the technical requirements of reversibility in model development”? Could you explain or rephrase this sentence?
  
  We will reword this sentence. Usually, it requires to turn off the new development and make the model performs like the old one.
  
  Lines 225, “H is tree height”should be modified to “Z is tree height”.
  
  Corrected. Thanks!
  
  In equation(10), k is ground area? Not defined.
  
  Added explanation.
  
  Figure 3. How to understand the constant LAI value of KZ?
  
  The model defined a maximum LAI. For Konza, the grasses can reach to the maximum each year, except the first year because of the low initial density. We will explain it in the new version of this manuscript.
  
  Figure 9. Please add units of LAI.
  
  Will do.
  
  Citation: https://doi.org/10.5194/gmd-2022-72-AC3

Ensheng Weng et al.

Viewed

Total article views: 1,009 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
820	166	23	1,009	5	5

HTML: 820
PDF: 166
XML: 23
Total: 1,009
BibTeX: 5
EndNote: 5

Views and downloads (calculated since 18 Mar 2022)

Month	HTML	PDF	XML	Total
Mar 2022	213	50	5	268
Apr 2022	146	29	6	181
May 2022	160	38	7	205
Jun 2022	74	10	2	86
Jul 2022	51	10	1	62
Aug 2022	100	15	2	117
Sep 2022	76	14	0	90

Cumulative views and downloads (calculated since 18 Mar 2022)

Month	HTML	PDF	XML	Total
Mar 2022	213	50	5	268
Apr 2022	146	29	6	181
May 2022	160	38	7	205
Jun 2022	74	10	2	86
Jul 2022	51	10	1	62
Aug 2022	100	15	2	117
Sep 2022	76	14	0	90

Viewed (geographical distribution)

Total article views: 877 (including HTML, PDF, and XML) Thereof 877 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 28 Sep 2022

Short summary

We developed a new demographic vegetation model to improve the representation of terrestrial vegetation dynamics and ecosystem biogeochemical cycles in an Earth system model. The individual-based competition for light and soil resources makes the modeling of eco-evolutionary optimality possible. This model will enable ModelE to simulate long-term biogeophysical and biogeochemical feedbacks between the climate system and land ecosystems at decadal to century temporal scales.


Total:	0
HTML:	0
PDF:	0
XML:	0