Improving terrestrial carbon flux simulations with machine learning and global Earth observations

Seiler, Christian

doi:10.5194/esd-17-651-2026

Articles | Volume 17, issue 3

https://doi.org/10.5194/esd-17-651-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/esd-17-651-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 17, issue 3

Research article

|

01 Jun 2026

Research article |

| 01 Jun 2026

Improving terrestrial carbon flux simulations with machine learning and global Earth observations

Christian Seiler

Data sets

Improving Terrestrial Carbon Flux Simulations With Machine Learning and Global Earth Observations Christian Seiler https://doi.org/10.5281/zenodo.18358100

The Impact of Climate Forcing Biases and the Nitrogen Cycle on Land Carbon Balance Projections Christian Seiler https://doi.org/10.5281/zenodo.7799565

Model code and software

The Canadian Land Surface Scheme including Biogeochemical Cycles (1.0) J. R. Melton et al. https://doi.org/10.5281/zenodo.3522407

cseilerQueens/daisy: daisy (release-2026-05-29) Christian Seiler https://doi.org/10.5281/zenodo.20440065

Short summary

This study shows how machine learning combined with global Earth observations can improve simulations of the land carbon cycle. Optimizing key model parameters enhances the accuracy of historical carbon fluxes, while machine-learning tools help assess the robustness of these results in the presence of compensating parameter effects. The findings demonstrate that parameter optimization strongly influences simulated carbon fluxes, highlighting its importance for improving climate models.