On recent trends in climate dynamics

The dynamics of the atmosphere and the oceans are inherently complex. There are active entangled processes running on spatial scales from millimetres to thousands of kilometres, and temporal scales from seconds to millennia. Capturing the whole range of spatial and temporal scales is impossible given current computer power. A numerical forecaster must decide, depending on their specific objectives, what scales they want to resolve. A corollary of this decision is that each numerical scheme inevitably fails to resolve so-called unresolved scales or subgrid scales. Usually, the interesting information is carried by the slow and large scales. For example, for weather forecasts we want to resolve large-scale highand low-pressure fields rather than small-scale fast oscillations of the stratification surfaces, or for climate prediction in a coupled ocean-atmosphere model we want to learn about the slow dynamics of the ocean that are constantly kicked by fast-evolving weather systems swirling above.