The Danish Meteorological Institute (DMI) has an opening for a PhD position focused on implementing physical constraints in a machine learning model in the Weather Research Department. The successful applicant joins the work of the R&D teams in the Numerical Weather Prediction Unit.

To successfully complete this three-year PhD project on physics-constrained machine learning for weather prediction, the ideal candidate requires a specialised blend of expertise in atmospheric science, scientific computing, and deep learning. You need to be comfortable working at the intersection of high-performance numerical weather prediction (NWP) and modern generative AI. Specifically, this requires advanced programming proficiency in Python, the use of deep learning frameworks, and a strong foundation in managing and processing large-scale geospatial datasets such as ERA5 or CARRA2.

Furthermore, the research demands a deep understanding of atmospheric dynamics to effectively integrate physical laws—such as conservation of mass and momentum—into model architectures. You will need the analytical capacity to design and test complex neural network architectures, including Graph Neural Networks (GNNs) and diffusion models, while rigorously evaluating their performance against operational standards like the Harmonie-Arome ensemble system. Strong academic communication skills are also essential, as the project involves synthesising complex methodological findings into three high-impact research publications and collaborating across both DMI and the Technical University of Denmark (DTU) research environments.



We expect a candidate to fulfil the following qualifications:

• Master's degree in geophysics or similar
• Experience with processing big data using AI/ML
• Skills in high-performance computing, fully operational 24/7 environments, ensemble forecasting – all in a national MET-service provider context
• Ability to work independently, targeted towards project milestones and deliverables and with the ability to reach out to colleagues for assistance and collaboration
• Programming and code development skills in a UNIX environment
• Good communication skills
• Capacity for working in a team

The Weather Research Department at DMI employs a staff of approximately 60 scientists, of whom 20 currently work within the Weather Models unit. The department also hosts the Ocean & Dispersion modelling and Flooding & Hydrology units. The department has close collaboration with other research partners nationally and internationally, and internally at DMI with, e.g. the on-duty meteorologists and the neighbouring research department, the National Centre for Climate Research. We have a strong international profile, as reflected in the many nationalities and English as the working language.



Practical information:

Employment and salary: Will be in accordance with Danish Law and Agreements. DMI offers a flexible 37-hour workweek, including a paid lunch break, and six weeks of paid vacation annually. The position is for three years, which is the duration of the PhD project.

Employment area: The Ministry of Energy, Utilities and Climate and associated institutions.

Working time: 37 hours weekly, including a lunch break and 6 weeks of paid vacation annually.

Further information: For further details about the position, please contact the Head of Ocean and Dispersion Modelling, Morten A D Larsen, at +45 2173 0395 or by email at mla@dmi.dk.

Interview: We expect to start interviews immediately after the deadline.

Application: Please enclose a cover letter, a curriculum vitae (incl. publication list), full grade sheets and contacts for relevant professional references. To apply for the position, please use the official link and follow the instructions.