CUBE: Cubed-sphere projection for adaptive mesh generation in spherical coordinates
Year: 2026
Type: software-paper
Abstract: We present CUBE , an open-source Python framework for generating adaptive, non-singular meshes on the sphere using a cubed-sphere projection. The software maps spherical slices to Cartesian faces of an inscribed cube, avoiding the pole singularities inherent to latitude–longitude grids and producing quasi-uniform element sizes across the globe. A core feature of CUBE is error-driven spatial adaptation: the mesh is refined according to an estimator based on an approximation of the H 1 -seminorm of the topography discretization error, which concentrates resolution where terrain gradients are large. The implementation leverages numpy and scipy for efficient array operations, integrates gmsh via its Python API for meshing, and supports standard geospatial input (e.g., GTOPO30 digital elevation models). CUBE is intended as an extensible tool to produce high-quality input meshes for atmospheric and geophysical models, improving accuracy while reducing computational costs through targeted refinement.
Source: SoftwareX
Author: Federico Gatti
Institutions: ETH Zurich, Seminar for Applied Mathematics
Language: English
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,200
OpenAlex ID: W7124446590