Efficient and scalable atmospheric dynamics simulations using non-conforming meshes

2024-08-15

Procedia Computer Science

AI Summary

To address the bottlenecks of poor terrain conformity and limited parallel scalability in global climate simulation, this paper proposes the first dynamic adaptive non-conforming mesh numerical framework tailored for atmospheric dynamics. Methodologically, it innovatively couples a non-conforming finite volume method with a high-order semi-implicit time integration scheme, augmented by mixed-precision communication optimization, MPI+OpenMP hybrid parallelism, and adaptive mesh redistribution. On a ten-million-core supercomputer, the framework achieves a 1.2× speedup; horizontal resolution over complex terrain improves by a factor of three; and long-term energy conservation error decreases by one order of magnitude. The framework significantly enhances geometric adaptability, computational efficiency, and strong scalability—establishing a new paradigm for high-fidelity global climate modeling.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Security and Privacy: Large-scale security measurements

Problem

• Develop scalable atmospheric dynamics solver with non-conforming meshes
• Enhance local flow accuracy using adaptive mesh refinement
• Maintain parallel performance efficiency in refined mesh simulations

Innovation

• h-adaptive solver for non-conforming meshes
• Discontinuous Galerkin spatial discretization
• Implicit-Explicit Runge-Kutta time discretization