The Project Hermès mi-parcours (HRMES Mid-Term Workshop) was held virtually on 29 January 2021. V. Balaji presented an introductory talk. The talk outlined the goals of Project Hermès as presented in the kickoff meeting in February 2019, using techniques from machine learning to build supermodels (where some model components are replaced by learning agents) and metamodels (low-dimensional emulators, or “fast approximate models” that can be used for exploring a model’s dependency on its parameters. In addition to its scientific thrusts, Project Hermès articulated the goals of building an interdisciplinary community of scientists equally at home in machine learning and Earth System Science. The introductory talk presented progress towards these goals, and was followed by presentations by MOPGA postdoctoral scientists Anna Sommer and Redouane Lguensat. The programme is below:
- Presentation by V. Balaji Project Hermès: High Resolution Modeling of the Earth System
- Presentation by Anna Sommer Reconstruction of Sub-grid-scale Buoyancy Fluxes from Large-Scale ocean Variables
- Presentation by Redouane Lguensat History Matching for the tuning of climate models
- Presentation by Redouane Lguensat Revealing changes in global ocean circulation under global heating using machine learning
The meeting was attended by, among others, Philippe Bousquet, Director LSCE, Olivier Boucher, Deputy Directory IPSL, Julie Deshayes LOCEAN, Imène Tabbi-Anneni and Yvan Lagadeuc, ANR. The project is on track toward its goals as outlined during the kickoff meeting.