Researcher position at CNRM in Artificial Intelligence for Numerical Weather Prediction – Downscaling_ M/F

This position is part of the Destination Earth (DestinE) Tender ‘DE_376’. DestinE is an initiative of the European Commission under the EU Digital Europe programme, alongside with ESA and EUMETSAT as partners. DestinE aims to deploy several highly accurate thematic digital replicas of the Earth, called Digital Twins. The Digital Twins will help monitor and predict environmental change and human impact, in order to develop and test scenarios that would support sustainable development and corresponding European policies for the Green Deal. Artificial Intelligence (AI) and, more precisely, Machine Learning and Deep Learning are important for DestinE on many different levels. In particular, they will be used in DE_376 to develop and deliver a production-ready, open-source, probabilistic regional AI weather forecasting model that can be rapidly deployed on regional domains over Europe and regions of European interest. The resulting capability will complement the physics-based regional component of the DestinE Weather-Induced Extremes Digital Twin by providing fast, flexible ensemble forecasts and enabling efficient uncertainty quantification. Furthermore, it will incorporate additional Earth system components to account for interactions with land surfaces, waves and the upper ocean.

Currently most operational weather forecasts rely on physically-based modelling approaches, and Numerical Weather Prediction (NWP) models are operated to determine atmospheric conditions for the next hours and days. In DE_376, the proposed core system will deliver kilometre-scale, probabilistic data-driven forecasts with hourly temporal resolution and forecast horizons of at least five days, building up on existing European AI forecasting models. However these models currently only represent the atmosphere and do not fully capture interactions with land surface and ocean state, hampering the forecast of extreme events such as coastal flooding, compound storm surge, drought-driven wildfire (e.g. Camps-Valls et al., 2025). The forecasting system will be therefore augmented with representations of land surfaces and ocean waves. All the developments will rely on the Anemoi open-source framework (https://www.ecmwf.int/en/about/media-centre/news/2024/anemoi-new-framework-weather-forecasting-based-machine-learning).

The goal of the position is to contribute to the development of the regional core forecast system by introducing a representation of land surfaces. To do so, the researcher will expand an existing atmospheric training dataset with additional land surface data, and will train an initially atmosphere-only forecasting system prototype first on a single domain covering part of western Europe, then on other domains where sufficient land surface data is available, in collaboration with other partners. The researcher will have the following responsibilities :

• build a training dataset including in particular snowpack, soil wetness and temperature data. This dataset will be extracted from a readily available hourly, kilometre-scale land surface reanalysis produced by Météo-France over part of western Europe. An in-depth assessment of this dataset is required to select events of interest (e.g. extreme snow depth or drought) for training and inference;

• develop an open software framework that will allow running (training and inference) the data-driven core system including land surfaces within Anemoi, in close collaboration with other project partners and the ECMWF;

• demonstrate the capabilities and assess quality of the core system including land surfaces on the selected dataset;

• document his/her work to allow users understand the result;

• contribute to project reports and deliverable documents.

• Further information :

  • Duration of the contract : 22 months

  • Deadline application : 20/07/2026

  • Required level of education/ Diploma : PhD

  • Email contacts for any further information : david.salas@meteo.fr  and laure.raynaud@meteo.fr :

    - A curriculum vitae detailing experience and technical skills

    - Motivation letter explaining interests for the job

    - Recommendation letters will be appreciated

    All applications will be considered immediately after the application deadline, and all shortlisted candidates will be invited to an interview shortly afterwards.

Applications from candidates without a PhD in Artificial Intelligence will not be considered. The ideal candidate will have the following qualifications:

- A good background in deep learning algorithms, in particular convolution neural networks.

- Experience in geophysical problems would be appreciated, at least a strong interest for applied research in waves and surface ocean physics is highly recommended

- Proficiency with Python programming, and good knowledge of the Anemoi framework are highly recommended

- Experience with processing large volumes of data

- Experience of working in a Linux-based environment

- Aptitude for scientific work, written and oral communication in English

- A scientific curiosity, autonomy, rigor in the interpretation of the results