E3.1 NonHydrostaticAtmosphere

                    

Poster TitleMultiscale interactions and non-hydrostatic modeling of the atmosphere
AuthorsJorge R. Urrego-Blanco, Balu Nadiga, and Mark Taylor
First AuthorJorge R. Urrego-Blanco
Session TypeE3SM session
Session IDE3
Submission TypePoster
GroupAtmosphere
ExperimentN/A
Poster Link




Abstract

We consider HOMME-NH, a variable-resolution, efficient and architecture-aware, non-hydrostatic dynamical core that is being developed under DOE’s earth system modeling initiative. With the algorithmic improvements included in this model and the current availability of high-performance computing resources, we can now begin to perform idealized global atmospheric simulations at resolutions at which non-hydrostatic effects become significant. While we are interested in studying and understanding the opportunities and challenges that are presented by such a capability, our studies also serve to assist in debugging and further developing and refining the model itself.

We consider high-resolution simulations of prototypical atmospheric flows with an aim of quantifying and characterizing non-hydrostatic effects. In particular, we consider two problems: the nonlinear evolution of an unstable baroclinic wave and mountain waves. In these problems, we focus on dynamical interactions across scales and study how linkages between the larger scale balanced modes and smaller scale, imbalanced modes differ when the hydrostatic approximation is made from when the full non-hydrostatic dynamics is resolved. Our ultimate goal in these studies is to demonstrate the impacts of resolving small-scale nonhydrostatic dynamics on the climate-relevant, global atmospheric circulation.

More practically, in conducting the above simulations, we find numerous issues that arise due to interactions between numerics and physics. For example, (a) The choice of a vertically-collocated computational grid can trigger spurious modes of instability---modes that resemble the physical symmetric instability; these instabilities are eliminated by the use of a (Lorenz) staggering in the vertical. (b) Choice of the nature of the vertical grid can lead to differences in some aspects of computation such as energy-conservation properties of the dycore. (c) We verify that the Hollingsworth instability that plagues numerous dycores and is exacerbated at high horizontal resolutions is absent in HOMME-NH.