Abstract | A key component of a multiscale modeling framework (MMF) is a cloud resolving model (CRM) to simulate small scale atmospheric dynamics. As part of the E3SM-MMF project, we have been developing a new CRM: the Portable Atmospheric Model (PAM). This new model is designed specifically to run on accelerator architectures, and is written in C++ using the YAKL performance portability framework. Unlike most atmospheric models, PAM is based on a Hamiltonian formulation of the continuous equations, and uses novel structure-preserving spatial and temporal numerics that preserve the key elements of this formulation. This poster will describe both the Hamiltonian formulations and the new numerics used in PAM, and show results from a variety of commonly used dry and moist test cases. |
|---|