According to an arXiv abstract, morphodynamic simulations require repeated coupling between hydrodynamic solvers, sediment transport modules, and bed updating. The abstract notes that for long-duration, multi-scenario simulations on large unstructured meshes, this coupling is a central concern.
To address this, the abstract proposes an adaptive-horizon graph neural operator as a surrogate model. The approach is designed to handle the complexity of large unstructured meshes while supporting long-duration simulations, though the abstract is truncated and does not provide implementation details or performance results.
The source is a single arXiv listing with announcement type 'cross'; no other sources are available. The truncation of the abstract limits the information that can be reported here.