A fast adaptive quadtree scheme for a two-layer shallow water model

This paper presents a dynamically adaptive quadtree grid generation system for the solution of a two-dimensional two-layer shallow water model. Roe-type two-layer shallow water solvers require numerical approximation of the system eigenvalues as well as numerical balancing, which increase computational cost considerably when a regular grid is used. In order to improve computational efficiency, we consider a dynamically adaptive quadtree grid generation system capable of increasing local resolution where high gradients occur in the physical flow variables. Test results show that satisfactory convergence can be obtained using the present scheme with the adaptive grid generator at a fraction of the cost incurred by a regular grid.