The CORE II mesh was initially used for experiments designed according to CORE II protocol. Later it become a basis mesh for climate simmulations with AWI-CM. Mesh resolution increased in equatorial and polar regions. 

The design of the HR mesh was presented in Sein et al., 2016. It is based on the idea to use finer resolution, and hence computational resources, in those regions where observed eddy variability is high. For this purpose, we use the pattern of sea surface height (SSH) variance derived from satellite altimetry (AVISO). This pattern is filtered to preserve the large-scale components and used as a basis for the mesh design. The resolution is set to be about 10 km in places of high variability, including the western boundary currents and ACC; and it is smoothly coarsened to the background resolution of 60 km in areas with relatively weak variability (see Sein et al. 2016, for details).  The mesh is additionally refined in the vicinity of coastlines and in certain passages (e.g. 2km in the Strait of Gibraltar). 

The XR mesh is based on varying resolution equal to half of the local Rossby radius, which is capped at 4 km (7km) in the Northern Hemisphere (Southern Hemisphere). The Rossby radius is calculated using the PHC3 climatology. Since the equatorial Rossby radius is rather large, an upper bound of 60km is employed for the coarsest resolution. This basic resolution is further improved in some key regions with observed high SSH variability. In other words, the resolution of the XR mesh is designed using a combination of two criteria. The first one is to enforce a grid size of half the Rossby radius, and the second one (mainly equatorial and tropical regions) is to scale the grid size by observed SSH variability. There are other necessary resolution adjustments on the XR mesh, but they are of geometrical character and aim at resolving the geometry of some passages. Although they introduce grid cell sizes smaller than 4 km (e.g. 2 km in the Strait of Gibraltar). Because of relatively high-resolution upper bounds for refinement used in high latitudes (4 and 7 km), the total number of nodes on the XR mesh turns out to be much larger than on the HR mesh, and in terms of degrees of freedom it fits into the same category as 1/8 –1/10 degree quasi-Mercator structured meshes. 

It contains a very simple (global) FESOM grid, mainly intended for technical tests and for spheRlab/pyfesom visualization examples. The mesh has 3.14*1000 2D surface nodes and it can be run with a timestep of 1.5 hours. Mesh is available at GitHub.