As an innovative training network, iMIRACLI is focussed on delivering a comprehensive training programme tailored to a broad cohort of students from climate science and machine learning backgrounds. We aim to train a new generation of climate data scientists, tailored for employment in climate and data sciences as well as for the growing sector of commercial data and climate applications.
The training programme will be delivered through a wide range of mechanisms, including:
- Network-wide training through summer schools and a comprehensive e-learning platform
- Active supervision by experienced supervisors
- Local specialised training at the host organisations
- Inter-disciplinary secondments between climate and machine-learning academics
- Secondments to the non-academic parners
- External enrichment activities, such as workshops and conferences
Yearly summer schools will provide students with hands-on courses taught by world leading experts in their respective fields. The courses will cover everything from the fundamentals of climate change through to detailed cloud physics, and from simple linear regression techniques through to the latest machine learning algorithms.
Our first summer school was hosted virtually by the University of Oxford. This summer school provided a solid grounding in climate science, machine learning and the skills needed to successfully complete a PhD. It was a fantastic opportunity for the cohort to meet each other and the rest of the supervisory team. Specifically this summer school covered: Atmospheric, aerosol and cloud physics; an introduction to atmospheric modelling; concepts in probability and statistics; unsupervised machine learning techniques such as random forests and support vector machines; supervised learning methods such as neural networks; working with big data in Python; and professional skills development.
The final three days of the Oxford summer school culminated in the Climate Informatics conference, the leading subject conference in this area, hosted in 2020 by iMIRACLI at Oxford.
Our second summer school will be hosted in Valencia, Spain. This will cover more specialist methods such as: detection and attribution; inverse modelling and data assimilation; cutting-edge machine learning topics such as causal inference and multi-modal and multi-task learning; research highlight talks focussed on earth observation data and machine learning advances relevant for iMIRACLI.
The third summer school will be hosted in Stockholm, Sweden. At the end of year two the ESR projects will be well advanced so the emphasis will shift to career development, student-led activities and presentation of results and key methodologies.
International iMIRACLI workshop on machine learning in climate sciences 2023
The final meeting will take the form of a three-day workshop organised by the ESR cohort hosted by the University of Oxford. At this stage ESRs will be internationally well connected and will showcase and disseminate iMIRACLI results to the scientific community and commercial stakeholders.