Postdoctoral Appointee - Machine Learning for Weather and Climate

Argonne National Laboratory, a U.S. Department of Energy National Laboratory located near Chicago, Illinois, has an opening for a highly motivated postdoctoral appointee in the Environmental Science Division.

Machine learning (ML), specifically deep learning (DL), has been demonstrated to successfully predict the weather for 1-14 days with skill on par with numerical weather prediction at a fraction of the computational cost.

A group of scientists at Argonne in collaboration with UCLA have successfully implemented a state-of-the-art ML weather model called Stormer. The candidate selected for this role will collaborate with this group of scientists to extend the predictability of Stormer to the subseasonal-to-seasonal (S2S) timeframe. This position will utilize generative AI to create a calibrated ensemble system for S2S at high resolution (30-km) to deliver probabilistic weather forecasts beyond 14 days to allow for actionable, local-scale impacts on infrastructure and communities.

In this role, you can expect to:

• Contribute technical expertise through analysis and support for programs and projects associated with machine learning, HPC, and computational problems related to earth system science and other dynamical systems.
• Develop, evaluate, and apply machine learning/computational approaches, synthesis activities, computational tools, compiling results, preparing reports, publications, and documentation.
• In particular, focus efforts on projects related to applying and developing machine learning-based weather models for the S2S timeframe with an emphasis on generative AI techniques, evaluating such models, and working with a team of scientists interested in pushing the boundary of predictability.

For more information, please see:

• Stormer modelICLR best paper award: (https://www.climatechange.ai/papers/iclr2024/7)
• Argonne press release: (https://www.anl.gov/article/argonne-develops-new-kind-of-ai-model-for-weather-prediction)

Position Requirements

Required skills and qualifications:

• Completed PhD (typically completed within the last 0-5 years) in geophysical sciences, atmospheric science, computer science, or related field
• Experienced in deep learning, PyTorch or JAX, and scaling deep learning models to large GPU-based machines
• Technical knowledge of large, dynamical systems (preferability the atmosphere and/or ocean)
• Expertise in data and model parallelisms for distributed training on large GPU-based machines
• Expertise in clear, concise writing of technical papers, and interacting and communicating verbally and orally effectively with colleagues
• Ability to model Argonne's core values of impact, safety, respect, integrity and teamwork

Preferred skills and qualifications:

• Candidates with experience using diffusion-based or other generative AI methods as well as experience in atmospheric science, especially weather modeling, are particularly sought after
• Technical knowledge in using HPC systems for visualization and analysis
• Experience in writing scientific code
• Effective problem-solving skills, organizational skills, and flexibility in coordinating a broad spectrum of activities
• Knowledge of atmospheric dynamics, process scale models, and numerical computation techniques
• Knowledge of data analysis
• Knowledge of using atmospheric observational datasets, data assimilation techniques, and statistics
• Familiarity subseasonal-to-seasonal modeling and or coupled atmosphere-ocean modeling
• Ability to work and communicate with stakeholders from public and private sectors

Job Family

Job Profile

Worker Type

Time Type