Page Comparison

Abstract

Performance The target 3km horizontal resolution of the E3SM atmosphere model is under the 10km threshold at which nonhydrostatic effects become important to resolve.  This makes the development and performance of the HOMME-NH nonhydrostatic atmosphere dynamic dynamic core is critical especially for the success of the science campaigns of E3SM v2 and beyond. NGD nonhydrostatic atmosphere project.  In this poster talk we present results from the effort to develop efficient IMEX time-integration methods for the an overview of the formulation, performance, and specifications of the new HOMME-NH nonhydrostatic dycore. The HOMME-NH dycore is numerically stiff since it supports vertically propagating acoustic waves whose resolution is irrelevant for climate science and will destabilize explicit integrators without a stringent step-size restriction. Implicit-explicit (IMEX) Runge-Kutta (RK) methods for time-integration of the primitive equations in HOMME-NH can increase the CFL stability limit beyond that of traditional explicit methods at a much lower computational cost per time-step than fully implicit methods. The IMEX-KG methods are a family of second and third order accurate IMEX RK methods designed specifically for efficient time-integration in HOMME-NH. Analysis of the joint stability region of the IMEX-KG methods enabled the development of methods whose CFL stability limit is determined by the hydrostatic CFL limitatmosphere dynamic core.  Implicit-explict Runge-Kutta time integration and energy conservation is discussed in detail and the performance and computational efficiency of HOMME-NH is compared with that of the E3SM v1 hydrostatic HOMME dynamic core.  Results are presented showing the effectiveness of the HOMME-NH dynamics at resolving nonhydrostatic effects in various test cases including the DCMIP2016 super cell test case.