Library of Regionally Refined Model Grids for the E3SM Atmosphere Model
North American RRM (NARRM) grids for the atmosphere dynamical core are shown in (a) a cylindrical equidistant projection and (b) an orthographic projection. One of the key features of E3SM is the ability to perform simulations with higher resolution in some parts of the globe and lower in others (see Fig. 1). This Regionally-Refined Model (RRM) capability is available for all components of the model (atmosphere, land, river, ocean, and sea ice), though the atmosphere capability is more mature and is the focus of this article. Regional refinement allows users to resolve storms, coastlines, mountains, and other features in the regions they care about at a fraction of the computational cost of global high-resolution simulations. In addition, RRM simulations allow the improved climate from the high-resolution region to flow back into the coarse-resolution portion of the domain. This upscale effect is a key benefit of RRM relative to more conventional regional models.
Atmospheric RRM capability has been available in E3SM since its inception (Tang et al., 2019), but has recently grown in popularity as a cost-effective way to make use of E3SM’s km-scale atmosphere model (Caldwell et al., 2021; Liu et al., 2023). Increased usage has caused an explosion of new RRM grids with refinement focused on different parts of the globe. The process of creating a new RRM grid is well documented on the publicly accessible confluence page: “Running E3SM on New Atmosphere Grids”. Still, there are a lot of steps in the process, and creating a grid that optimizes model skill in the region of interest is a bit of an art form. In particular, developers need to ensure that weather features flowing into the refined area have enough upstream fetch to develop fine-scale flow features before reaching the area of interest. This requires an understanding of regional climatology and an expansion of the refined region beyond the borders of the actual target zone.
To save time and avoid confusion caused by having several slightly different grids, E3SM has created a publicly-accessible library of Regionally-Refined Model (RRM) grids. The idea is that before creating a new grid, users will check this page to see if a suitable grid already exists. If it doesn’t, they will create one. In addition to adding this new grid to the library, a GitHub pull request will also be needed to make the grid and associated input data available to other users.
A challenge with a library of this type is that each useful grid is the product of many failed attempts. Grids can be problematic because they don’t extend far enough or extend too far beyond the region of interest, because the resolution transition zone is too sharp, or because input data files don’t work right. The goal of this library is to provide only trustworthy grids, so we ask that only grids based on published papers be contributed. So far the library includes cases for North America, the Antarctic, the Eastern US, California, and the Amazon.