Inter–annual Variability of Model Parameters Improves Simulation of Annual Gross Primary Production

Description

This repository contains codes to perform analysis and reproduce figures of our research paper:

De, R., Brenning, A., Reichstein, M., Šigut, L., Ruiz Reverter, B., Korkiakoski, M., Paul-Limoges, E., Blanken, P. D., Black, T. A., Gielen, B., Tagesson, T., Wohlfahrt, G., Montagnani, L., Wolf, S., Chen, J., Liddell, M., Desai, A. R., Koirala, S. and Carvalhais, N. (2026). Inter–annual Variability of Model Parameters Improves Simulation of Annual Gross Primary Production, ESS Open Archive, https://doi.org/10.22541/essoar.174349993.30198378/v2

This paper is a companion paper or the second part of our previous study. Further details on our methodology/ description of models can be found at:

De, R., Bao, S., Koirala, S., Brenning, A., Reichstein, M., Tagesson, T., Liddell, M., Ibrom, A., Wolf, S., Šigut, L., Hörtnagl, L., Woodgate, W., Korkiakoski, M., Merbold, L., Black, T. A., Roland, M., Klosterhalfen, A., Blanken, P. D., Knox, S., Sabbatini, S., Gielen, B., Montagnani, L., Fensholt, R., Wohlfahrt, G., Desai, A. R., Paul-Limoges, E., Galvagno, M., Hammerle, A., Jocher, G., Ruiz Reverter, B., Holl, D., Chen, J., Vitale, L., Arain, M. A., and Carvalhais, N. (2025). Addressing Challenges in Simulating Inter–annual Variability of Gross Primary Production. Journal of Advances in Modeling Earth Systems, 17(5), e2024MS004697. https://doi.org/10.1029/2024MS004697

We used broadly the following two models in our study. It is highly recommended to get acquainted with the following two research papers before using our codes.

1. Optimality-based model: P-model of Mengoli

Mengoli, G., Agustí-Panareda, A., Boussetta, S., Harrison, S. P., Trotta, C., and Prentice, I. C. (2022). Ecosystem photosynthesis in land-surface models: A first-principles approach incorporating acclimation, Journal of Advances in Modeling Earth Systems, 14, https://doi.org/10.1029/2021MS002767

2. Semi-empirical model: Bao model

Bao, S., Wutzler, T., Koirala, S., Cuntz, M., Ibrom, A., Besnard, S., Walther, S., Šigut, L., Moreno, A., Weber, U., Wohlfahrt,695 G., Cleverly, J., Migliavacca, M., Woodgate, W., Merbold, L., Veenendaal, E., and Carvalhais, N. (2022). Environment-sensitivity functions for gross primary productivity in light use efficiency models, Agricultural and Forest Meteorology, 312, 108 708, https://doi.org/10.1016/j.agrformet.2021.108708

Disclaimer

This repository should only be used for the experiments related to a given group of parameters varying per year, while other parameters remain fixed across years in a site. For all other experiments, such as optimizing all parameters across site-years, sites, plant functional types, globally, the repository from our previous study should be used. It is available at: https://github.com/de-ranit/iav_gpp_p_bao (https://doi.org/10.5281/zenodo.13729514).

The codes are written to be compatible with computing platforms and filestructure of MPI-BGC, Jena. It maybe necessary to adapt the certain parts of codes to make them compatible with other computing platforms. All the data should be prepared in NetCDF format and variables should be named as per the code. While the actual data used for analysis is not shared in this repository due to large sizes, all the data source are cited in the relevant paper and openly accessible. Corresponding author (Ranit De, rde@bgc-jena.mpg.de or de.ranit19@gmail.com) can be contacted in regards to code usage and data preparation. Any usage of codes are sole responsibility of the users.

Structure

• site_info: This folder contains two .csv files: (1) SiteInfo_BRKsite_list.csv, this one is necessary so that the code knows data for which all sites are available and can access site specific metadata for preparing results, such as data analysis and grouping of sites according to site characteristics, (2) site_year_list.csv lists all the site–years available for site–year specific optimization. This list also contains site–years which are not of good quality, and later gets excluded during data processing steps.
• src: This folder basically contains all source codes. It has four folders: (1) common folder contains all the scripts which are common for both the Optimality-based (P-model and its variations) and the semi-empirical model (Bao model and its variations), (2) lue_model contains model codes and cost function specific to the semi-empirical model (Bao model and its variations), (3) p_model contains model codes and cost function specific to the Optimality-based (P-model and its variations), and (4) postprocess contains all the scripts to prepare exploratory plots after parameterization and forward runs.
• optimize_lbgfs: This folder contains the code to further constrain model parameters obtained from CMA-ES (with a big population size) by using a gradient-based optimizer (L-BFGS-B).
• prep_figs: This folder contains all the scripts to reproduce the figures which are presented in our research paper and its supplementary document. All modelling experiments and their relevant data must be available to reproduce the figures and their relative paths should be correctly mentioned at result_path_coll.py.

How to run codes?

• Create a conda environment and install dependencies. Dependencies are listed in requirements.yml.
• Open model_settings.xlsx and specify all the experiment parameters from dropdown or by typing as described in the worksheet.
• Run main_opti_and_run_model.py to perform model parameter calibration or forward runs. If you want parallel processing on a high performance computing (HPC) platform, other settings are necessary based on the platform you are using. See send_slurm_job.sh for a sample job submission recipie to a HPC platform using slurm as a job scheduler.

How to cite?

Research paper:

• BibTeX

@article{De_2026_paramval,
author = {De, R. and Brenning, A. and Reichstein, M. and Šigut, L. and Ruiz Reverter, B. and Korkiakoski, M. and Paul-Limoges, E. and Blanken, P. D. and Black, T. A. and Gielen, B. and Tagesson, T. and Wohlfahrt, G. and Montagnani, L. and Wolf, S. and Chen, J. and Liddell, M. and Desai, A. R. and Koirala, S. and Carvalhais, N.},
doi = {10.22541/essoar.174349993.30198378/v2},
journal = {ESS Open Archive},
note = {preprint},
title = {{Inter--annual Variability of Model Parameters Improves Simulation of Annual Gross Primary Production}},
url = {https://essopenarchive.org/doi/full/10.22541/essoar.174349993.30198378/v2},
month = {jan},
year = {2026}
}

• APA

De, R., Brenning, A., Reichstein, M., Šigut, L., Ruiz Reverter, B., Korkiakoski, M., Paul-Limoges, E., Blanken, P. D., Black, T. A., Gielen, B., Tagesson, T., Wohlfahrt, G., Montagnani, L., Wolf, S., Chen, J., Liddell, M., Desai, A. R., Koirala, S. and Carvalhais, N. (2026). Inter–annual Variability of Model Parameters Improves Simulation of Annual Gross Primary Production, ESS Open Archive, https://doi.org/10.22541/essoar.174349993.30198378/v2

This repository:

• BibTeX

@software{de2026codes_param,
author = {De, Ranit},
title = {{Scripts for analyses presented in ``Inter--annual Variability of Model Parameters Improves Simulation of Annual Gross Primary Production''}},
month = jan,
year = 2026,
publisher = {Zenodo},
note = {v1.2-preprint},
doi = {10.5281/zenodo.15089289},
URL = {https://github.com/de-ranit/GPP-ModelParamVariation}
}

• APA

De, R. (2026). Scripts for analyses presented in “Inter–annual Variability of Model Parameters Improves Simulation of Annual Gross Primary Production” (v1.2-preprint). Zenodo. https://doi.org/10.5281/zenodo.15089289

Change Log:

v1.2-preprint

• CMA-ES optimization with default hyperparameters
• optimize with L-BFGS-B starting from CMA-ES with big population size
• additional analyses and updating figures

v1.1-preprint

• Contains code for model optimization in which a group of parameters were varied per year, while other parameters remain fixed.

License

This work is licensed under a MIT License.

License logo is created by ExcaliburZero, used under CC BY 3.0 license