Waveform Relaxation for Atmosphere-Ocean-Sea Ice Coupling

Earth system models are currently the best tools society has for simulating planetary processes and climate change. These are large computer codes which couple submodels in time and space; for example, models for the atmosphere, ocean, vegetation, or sea ice. The algorithm used to couple the submodels produces a
mathematical error at the interface between the models. We refer to this as the (numerical) coupling error and study how it behaves at the sea surface, where the atmosphere, ocean, and sea ice model interact.

In this project, we systematically study the magnitude and properties of the coupling error in a climate model of intermediate complexity, the EC-Earth single column model. For this we make use of iterative coupling algorithms, so-called waveform relaxation. This method provides us with a reference solution which we can compare standard simulation results to. Estimating the coupling error is crucial in the process of reducing it by means of developing new, energy-efficient coupling approaches.