Researcher (M/F) Data Assimilation and Machine Learning for Arctic Sea Ice Forecasting

Organisation/Company CNRS Department Institut des géosciences de l'environnement Research Field Environmental science Environmental science » Earth science Environmental science » Global change Researcher Profile Recognised Researcher (R2) Application Deadline 13 Mar 2026 - 23:59 (UTC) Country France Type of Contract Temporary Job Status Full-time Hours Per Week 35 Offer Starting Date 1 May 2026 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description The overarching mission is to conduct research combining machine learning, data assimilation, and physical modeling to enhance short-term (days/weeks) forecasts of Arctic sea ice conditions.

The selected candidate will lead research efforts to explore how neural emulation strategies can be leveraged to enhance short-term forecasts of Arctic sea ice. The proposed approach will combine: (i) a neural emulator of sea ice dynamics, trained using high-fidelity numerical simulations, (ii) variational data assimilation methods, and (iii) a simplified representation of physical processes in the atmospheric boundary layer.

The candidate will first deploy a multivariate emulator of key variables governing Arctic sea ice dynamics, trained on high-fidelity simulations developed at IGE. This emulator will be implemented within an ensemble variational data assimilation system, enabling short-term forecasts based on sea ice concentration and thickness data while providing associated uncertainty estimates.

In a second phase, the focus will shift to assessing the impact of explicitly representing ice‑atmosphere interactions in the atmospheric boundary layer on forecast quality. This physical modeling will leverage differentiable programming techniques.

The selected candidate will be expected to publish findings in scientific journals, present results at international conferences, and contribute to dedicated working groups addressing these research questions.

Working context The selected candidate will work at the Institute of Environmental Geosciences (IGE) in Grenoble, located in the French Alps. IGE is a public research institute affiliated with CNRS, IRD, Université Grenoble Alpes, Grenoble-INP, and INRAE. It brings together approximately 250 people, including 150 permanent members (researchers, lecturer‑researchers, engineers) and 100 contractual staff (PhD students, postdoctoral researchers, engineers, and technicians). The institute also hosts dozens of interns and visiting scientists each year. IGE is spread across three sites on the Grenoble university campus, all within a 5‑minute walk of each other. It is one of the leading institutes of the Grenoble University Space Observatory (OSUG), a federative structure under the National Institute of Universe Sciences (INSU). The selected candidate will join OPERA, a new interdisciplinary team in computational geosciences at IGE. They will work under the supervision of Julien Le Sommer and in close collaboration with Pierre Rampal and Charlotte Durand.

Scientific context Short‑term forecasts—ranging from a few days to a few weeks—of sea ice conditions are critical for maritime navigation, environmental risk management, and understanding climate interactions in polar regions, particularly in the Arctic. Current forecasting systems, which rely on physical models and data assimilation approaches, face limitations in prediction quality due to initialization uncertainties, challenges in accounting for atmospheric forcing, and the representation of sea ice dynamics, especially its mechanical deformation and response to atmospheric conditions, which remain a major challenge for existing models. The use of neural emulators and differentiable programming techniques opens up new opportunities to (i) better integrate model and observational data, (ii) represent key physical processes in a short‑term forecasting context, and (iii) characterize the uncertainty of these simulators.

The selected candidate will hold a PhD in geosciences, applied machine learning, data assimilation, or applied mathematics.

Qualifications

Experience in numerical modeling (ocean, sea ice, atmosphere) or in physical model emulation;

Knowledge of data assimilation methods (4DVar, EnKF, En4DVar) and their practical implementation;

Research experience in machine learning applied to dynamic systems;

Proficiency in key machine learning libraries (PyTorch, JAX, etc.);

Mastery of Python and the software ecosystem for scientific data analysis and management (NumPy, Xarray, etc.);

Familiarity with collaborative software development tools and practices (Git, documentation, etc.);

Experience in international and interdisciplinary research contexts;

Experience in writing and communicating scientific results;

Ability to work in a team and in a multicultural environment;

Fluency in English (written and spoken).

The selection committee will also consider gender balance within the overall research team.

#J-18808-Ljbffr