Variable Resolution Mesh Design

  • November 13, 2019
  • Home Page Feature,Science and Technical Highlights
  • Enhanced-resolution mesh prototypes: Coastal US Plus (left) and Southern Ocean refined (right) showing higher resolution in coastal regions (colorbars show approximate grid cell width).

     

    Background

    Global high-resolution simulations provide great detail in modeling physical processes, but are too computationally expensive to run regularly.  Hence scientists create meshes with higher resolution around areas of interest within a global, lower-resolution domain. Climate model components, such as the Model for Prediction Across Scales-Ocean (MPAS-Ocean), use unstructured meshes to enable such variable-resolution, regionally-enhanced configurations.

    Accurate modeling of coastal areas is particularly relevant to global mean sea level rise and coastal inundation, which pose risks to coastal populations, infrastructure, and natural resources.  These effects depend on the structure and evolution of water properties and currents along the coasts, so scientists place high resolution in these regions. The fidelity of the simulation depends on the details of the location, transition width, and grid cell size of these refined region.

    In this research, scientists are investigating the relationship between mesh quality and simulation statistics using the JIGSAW unstructured meshing library and MPAS-Ocean, which E3SM uses as the ocean model component.

    Examples of low-resolution (left), coastal-refined (center), and high-resolution (right) meshes showing North America in the top row and the region around Iceland in the lower row. The color represents the grid cell size, with red depicting finer cells and blue coarser cells. For the coastal-refined mesh, the estimated computational cost increase over the low resolution simulation is seven times greater, compared to 66 times greater for the global high-resolution simulation.

    Coastal US Plus (CUSP) Mesh

    In the base configuration, the Coastal United States ‘Plus’ (CUSP) mesh, the refined region employs 8 km cells and extends 400 km from the coast of North America. This coastal-refined region is embedded within a low-resolution global domain, with cell size varying latitudinally between 30 and 60 km. The resolution transition region between the refined region and background mesh is 600 km wide.

    In a first study, Hoch et al. (2019) conducted sensitivity tests where: 1) the quality of meshes are intentionally degraded so that horizontal cells are progressively more distorted; 2) the transition region from high to low resolution is steepened; and 3) resolution of the coastal refinement region is varied from 30 km to 8 km. The results show that overall, the ocean simulations are robust to mesh resolution and alterations in cell quality. Meshes that are substantially degraded still produce realistic currents, with Southern Ocean transports within 0.4% and Gulf Stream transports within 12% of high-quality mesh results. The narrowest transition case of 100 km did not produce any spurious effects. Refined regions with high resolution produce eddy kinetic energy and sea surface height variability that are similar to the high-resolution reference simulation.

    In a second study, Rosa et al. (2019) looked in detail at key ocean currents in the CUSP mesh using realistic atmospheric forcing. They found that many high-resolution features are captured by the coastal-refined domains, but it is critical that certain dynamically active regions, such as the Gulf Stream and its eastward extension, are fully contained within the high-resolution portion of the domain. These results provide heuristics for the design criteria of variable-resolution climate models.

    Sea Surface Temperature (SST) and SST gradient near California in MPAS-Ocean simulations with the low-resolution (left), coastal-refined (center) and high-resolution (right) meshes, showing that coastal refinement can provide the advantages of high resolution over limited areas.

    Impact

    • E3SM version 2 will use variable-resolution MPAS-Ocean meshes to provide high-resolution improvements in simulation quality at a fraction of the computational cost.
    • Variable resolution design and prototyping workflow is more than 100 times faster than previous versions.

    Status

    • Researchers at Los Alamos National Laboratory (LANL) are currently designing and testing variable resolution mesh configurations for both Water Cycle and Cryosphere E3SM version 2 science campaigns.
    • Research and preliminary testing of the North American mesh has led to an improved design with a more realistic Gulf Stream.

    Mentoring and Career Development

    Both first authors for these publications are early career researchers who are developing expertise in climate modeling relevant to E3SM and DOE. Kristen Hoch holds a post-bachelors position and Kevin Rosa is a graduate student in oceanography at the University of Rhode Island, working summers at LANL. They are mentored by Mark Petersen and others at LANL, and in collaboration with Darren Engwirda of Columbia University, the author of the Jigsaw mesh creation library.

     

    Variable Resolution Mesh Webinar

    This work was presented in the E3SM All-Hands Presentations and is available to all on Youtube, see https://e3sm.org/about/events/all-hands-presentations/

    Publications

    • Hoch, K.E., Petersen, M.R., Brus, S.R., Engwirda, D., Roberts, A.F., Rosa, K.L., Wolfram, P.J., (2019), MPAS-Ocean Simulation Quality for Variable-Resolution North American Coastal Meshes, in review, JAMES

    • Rosa, K.L., Petersen, M.R., Brus, S.R., Engwirda, D., Hoch, K.E., Maltrud, M.E., Van Roekel, L.P., Wolfram, P.J, (2019), Boundary current impacts of coastal refinement in the E3SM unstructured-mesh ocean model MPAS-Ocean, in prep.

    Contact

    Mark Petersen, Los Alamos National Laboratory

     

    Funding

    This research was supported as part of the Energy Exascale Earth System Model (E3SM) project, funded by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, as well as the U.S. Department of Energy Advanced Research Projects Agency – Energy (ARPA-E) Macroalgae Research Inspiring Novel Energy Resources (MARINER) program.

    Send this to a friend