Project Description: Physical glacier modelling has become an efficient and necessary tool to predict the future evolution of glaciers and the resulting sea-level rise under climate change scenarios, or to reconstruct Quaternary glaciations worldwide. However, high computational expenses associated with the modelling of complex physical processes (i.e. the ice dynamics) strongly limit the potential of these models -- especially in view of long time scales paleo applications. In recent years, deep-learning surrogate models have shown outstanding results at speeding-up physical models -- including glacial processes -- opening new perspectives for applications that are unreachable with traditional modelling.
Goal of the Thesis: i) to develop the Instructed Glacier Model (IGM, https://github.com/jouvetg/igm) – a newly introduced glacier model accelerated by deep-learning – by embedding new relevant physical processes in form of neural networks trained from data and/or state-of-the-art physical models and ii) to apply the improved model to the reconstruction of glacier extent, landscape and climate evolution in the European Alps during the last glacial cycles. The research involves a large diversity of fields including glaciology, physical and numerical modelling, machine learning, climatology, and geomorphology. The successful candidate will join the ICE (https://wp.unil.ch/ice/) group in the Institute of Earth Surface Dynamics which specializes in paleoenvironment and landscape evolution in a range of different environmental settings.
Candidate Profile: The chosen candidate will have a master degree either in Earth sciences, geophysics, physics, applied mathematics, machine learning, computer sciences, or a related field, and should have a sharp interest in the modelling of geophysical processes. Previous experience in machine learning, numerical modelling, and Python programming is an asset. Good writing and communication skills in English as well as the motivation to fruitfully collaborate within an interdisciplinary framework are essential. Knowledge in French language is preferable but not necessary.
Job description: The majority of the workload will be dedicated to the completion of the Ph.D. thesis, which includes model development, paleo glacier modelling applications, and the writing of peer-reviewed publications. Participation in internal and international meetings and conferences is expected, as well as the active participation in the research institute. A component of the workload will consist in assisting with teaching and research duties: teaching activities under the supervision of a professor, research work not directly related to the personal PhD topic, technical and administrative tasks related to the activities of the Institute.
Contact for further information: Prof. Guillaume Jouvet (firstname.lastname@example.org)
The full job description (entitled "Graduate Assistant in deep learning-aided glacier modelling" -- ID: 19184) can be found here: https://career5.successfactors.eu/career?career%5fns=job%5flisting&company=universitdP&navBarLevel=JOB%5fSEARCH&rcm%5fsite%5flocale=en%5fUS&career_job_req_id=19184&selected_lang=en_US&jobAlertController_jobAlertId=&jobAlertController_jobAlertName=&browserTimeZone=Europe/Zurich&_s.crb=yIvxSBQs%2bkpJEU1X4gyBhx6H60O6OKV%2fQFehn7pkJd0%3d
Approx. annual salary of CHF 50’000
Only complete applications made through this website will be considered. Review of applications will start on 14 March 2021, and will continue until the position is filled.