Vitus Benson

Hi Friend, great to have you here!

I am Vitus, currently an ELLIS PhD candidate at Max Planck Institute for Biogeochemistry, ELLIS Unit Jena and ETH Zürich. I am conducting my research under the supervision of Markus Reichstein and Fanny Yang.

I train deep neural networks on data of the Earth system. More specifically, I am focusing on two applications. On the one hand, I am interested in the applicability of large-scale AI to Early Warning Systems. Within the EarthNet initiative we are leveraging high resolution satellite imagery to build data-driven impact early warning systems – with applications to anticipatory action for droughts and heat waves. Beyond my own work, I am collaborating within the RiskKAN Working Group AI for Complex Climate Risk Mitigation, which I happen to co-chair aswell. On the other hand, as part of the AI4Carbon Community Initiative I explore the use of deep neural networks for carbon cycle research. In particular, I am assessing the use of representation learning for atmospheric transport and inverse modeling – analogous to how recent works have revolutionized numerical weather prediction and other areas of scientific machine learning.

My research interests are broadly speaking Deep Learning, Biogeochemistry, Climate Risk Mitigation and Complex Systems. I believe solving global challenges requires working together. Let us do just that!

news

May 04, 2026	I’ll be at EGU26 in Vienna, presenting our recent work on Mech Interp for Earth Embeddings, contributing to multiple works by amazing colleagues (1, 2, 3, 4), co-hosting two sessions (ML + Carbon Cycle, Complex Climate Risks), and two splinter meetings (AI4Carbon, RiskKAN AI WG).
Apr 14, 2026	Glad to be able to convey Biogeoscience to a more general audience at Imaginata Osterüberraschungen.
Mar 16, 2026	Glad to be contributing as a challenge tutor at the ELLIS Winter School on AI for Earth System, Hazards & Climate Extremes in Athens.
Feb 09, 2026	Representing the Land model at the WeatherGenerator GA & Dissemination workshop in Oslo.
Dec 06, 2025	Happy to present our work on Atmospheric Transport Emulators at the EurIPS workshop on AI for Climate & Conservation in Copenhagen.
Dec 01, 2025	Happy to be speaking at the Global Dialogue Platform session on Guiding tomorrow’s Anticipatory Action: what role can artificial intelligence play and at Google Berlin for their AI for climate risk forecasting event with the Anticipation Hub!

TEDx talk

selected publications

Nat. Commun.

Early warning of complex climate risk with integrated artificial intelligence

Markus Reichstein* , Vitus Benson* , Jan Blunk , Gustau Camps-Valls , Felix Creutzig , and 14 more authors

Nature Communications, Mar 2025

Bib PDF

@article{reichstein2025ews,
  title = {Early warning of complex climate risk with integrated artificial intelligence},
  author = {Reichstein*, Markus and Benson*, Vitus and Blunk, Jan and Camps-Valls, Gustau and Creutzig, Felix and Fearnley, Carina J and Han, Boran and Kornhuber, Kai and Rahaman, Nasim and Sch{\"o}lkopf, Bernhard and Tárraga, José María and Frank, Dorothea and Martini, Giulia and Vinuesa, Ricardo and Dall, Karen and Denzler, Joachim and Nganga, Naomi and Maddix Robinson, Danielle and Weldemariam, Kommy},
  journal = {Nature Communications},
  volume = {16},
  number = {1},
  pages = {2564},
  year = {2025},
  month = mar,
  publisher = {Nature Publishing Group UK London},
}

JAMES

Atmospheric Transport Modeling of CO2 with Neural Networks

Vitus Benson , Ana Bastos , Christian Reimers , Alexander J. Winkler , Fanny Yang , and 1 more author

Journal of Advances in Modeling Earth Systems, Feb 2025

Bib PDF Code

@article{benson2024neuraltransport,
  title = {Atmospheric Transport Modeling of CO2 with Neural Networks},
  author = {Benson, Vitus and Bastos, Ana and Reimers, Christian and Winkler, Alexander J. and Yang, Fanny and Reichstein, Markus},
  year = {2025},
  month = feb,
  pages = {e2024MS004655},
  journal = {Journal of Advances in Modeling Earth Systems},
  volume = {17},
  number = {2},
  publisher = {Wiley Online Library},
}

CVPR

Multi-modal Learning for Geospatial Vegetation Forecasting

Vitus Benson , Claire Robin , Christian Requena-Mesa , Lazaro Alonso , Nuno Carvalhais , and 5 more authors

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2024

Bib PDF Code

@article{benson2024multimodal,
  title = {Multi-modal Learning for Geospatial Vegetation Forecasting},
  author = {Benson, Vitus and Robin, Claire and Requena-Mesa, Christian and Alonso, Lazaro and Carvalhais, Nuno and Cort\'es, Jos\'e and Gao, Zhihan and Linscheid, Nora and Weynants, M\'elanie and Reichstein, Markus},
  year = {2024},
  month = jun,
  pages = {27788-27799},
  journal = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
}