Predicting future climate conditions on earth, and in particular, the impacts of climate change, would not be possible without climate models. Climate models are complex mathematical representations of the physical processes happening in each of the climate system components: ocean, atmosphere, sea-ice and land surface, as well as their interactions with each other.
Climate models divide the globe into three dimensional grid cells representing a specific location and elevation. The resolution of the models, their “level of detail”, is dependent on the size chosen for the grid cells: the smaller the grid size the higher the resolution. In practice, the grid size we can choose is limited by computer power.
The Machine Learning parameterizations developed by M²LInES for the ocean, atmosphere and sea-ice (and their interactions) will be implemented into coarse-resolution models (large sized grid) from leading modeling centers (NCAR, GFDL, IPSL). The aim will be to achieve comparable accuracy to high-resolution global simulations or observations, and therefore enhance climate projections.
Learn more: Learn more about the challenges in modelling the oceans for climate in this presentation by Alistair Adcroft, M²LInES head of modeling
