This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| MPAS (Model for Prediction Across Scales) | |
|---|---|
| Name | MPAS (Model for Prediction Across Scales) |
| Developer | National Oceanic and Atmospheric Administration; Los Alamos National Laboratory; National Center for Atmospheric Research |
| Initial release | 2010s |
| Programming language | Fortran; C; Message Passing Interface |
| License | Open-source software |
MPAS (Model for Prediction Across Scales) is a numerical modeling framework for atmospheric, oceanic, and cryospheric prediction that supports unstructured meshes and multiscale simulations. The project integrates community research from National Oceanic and Atmospheric Administration, Los Alamos National Laboratory, National Center for Atmospheric Research, University of Washington, and Texas A&M University to address regional and global forecasting challenges. MPAS has been applied in operational and research settings interfacing with systems such as Weather Research and Forecasting Model, Global Forecast System, Community Earth System Model, and regional initiatives like HYCOM collaborations.
MPAS provides a unified dynamical core and shared libraries enabling simulations across scales for atmosphere and ocean, supporting unstructured Voronoi meshes and C-grid staggering. The framework emphasizes scalability for high-performance computing platforms developed by Oak Ridge National Laboratory, Lawrence Livermore National Laboratory, Argonne National Laboratory, and supercomputers such as Summit (supercomputer), Titan (supercomputer), Fugaku. MPAS integrates with data assimilation efforts from Data Assimilation Research Testbed, NOAA National Centers for Environmental Prediction, European Centre for Medium-Range Weather Forecasts, and observational programs like Argo (oceanography), Global Precipitation Measurement.
Development began through collaborations among Los Alamos National Laboratory, National Oceanic and Atmospheric Administration, and academic partners including University of Colorado Boulder and University of California, Los Angeles. Early funding and code exchanges involved agencies such as Department of Energy (United States), National Science Foundation, and projects affiliated with Office of Science and Technology Policy. MPAS evolved alongside numerical models like Community Atmosphere Model, Parallel Ocean Program, and community initiatives such as Earth System Modeling Framework. Major milestones include integration with Community Earth System Model workflows and adaptations for exascale efforts led by Exascale Computing Project.
The MPAS architecture centers on dynamical cores for atmosphere and ocean, mesh generators, physics libraries, and I/O drivers compatible with NetCDF. The atmospheric core uses rotating shallow-water and compressible nonhydrostatic formulations influenced by approaches in Finite-Volume Methods, Spectral Element Methods, and comparisons with MPDATA schemes. Mesh generation leverages Voronoi tessellations and Delaunay triangulations with tools developed alongside Generic Mapping Tools and grid resources from Earth System Grid Federation. I/O and coupling interfaces align with Model Coupling Toolkit and use community standards propagated by Unidata.
MPAS implements parameterizations for convection, cloud microphysics, radiation, turbulence, and surface fluxes, drawing on schemes from Goddard Institute for Space Studies, European Centre for Medium-Range Weather Forecasts parameter sets, and community packages such as WRF physics. Boundary layer treatments and land surface coupling incorporate models like Noah (land surface model), Community Land Model, and cryosphere modules informed by International Association of Cryospheric Sciences practices. Oceanic subgrid closures adopt eddy parameterizations comparable to Gent–McWilliams and mixing approaches from K-profile parameterization studies.
MPAS has been applied to global climate simulations in ensembles for Coupled Model Intercomparison Project, seasonal forecasting partnered with North American Multi-Model Ensemble, regional weather prediction for events monitored by National Weather Service, and ocean modeling supporting Intergovernmental Oceanographic Commission research. Use cases include hurricane and tropical cyclone research coordinated with National Hurricane Center, Arctic sea-ice studies linked to Arctic Council assessments, and mesoscale convection analysis supporting Field Campaigns like VORTEX and GPM Ground Validation. MPAS underpins academic projects at institutions such as Massachusetts Institute of Technology, Princeton University, University of Oxford, and ETH Zurich.
The codebase is written primarily in Fortran with parallelization using Message Passing Interface and support for hybrid MPI/OpenMP executions on architectures from Cray Inc. and Intel Corporation processors. Performance tuning has been performed for machines at National Energy Research Scientific Computing Center, Argonne Leadership Computing Facility, and systems managed by Texas Advanced Computing Center. Workflow integration uses containerization strategies with Singularity (software) and build systems aligned with CMake and continuous integration services provided by GitHub. Data formats and metadata follow conventions from Climate and Forecast (CF) metadata convention.
Validation efforts compare MPAS outputs against reanalyses like ERA5, observational datasets from Argo (oceanography), GPS Radio Occultation records, and intercomparison projects such as Model Intercomparison Project. Performance benchmarks assess scalability with strong and weak scaling tests on supercomputers including Mira (supercomputer) and Blue Waters, and evaluate accuracy against models like WRF, GFS, and NEMO (ocean model). Community-led verification campaigns involve institutions such as NOAA ESRL, NASA Goddard Space Flight Center, and academic groups at Scripps Institution of Oceanography.
MPAS development is governed by consortium-style collaborations among national laboratories, universities, and agencies including NOAA Research, Department of Energy (United States), and National Science Foundation with community coordination via workshops hosted by American Meteorological Society, European Geosciences Union, and steering inputs from project teams at Los Alamos National Laboratory and National Center for Atmospheric Research. Contribution workflows use version control on platforms supported by GitHub with community governance practices influenced by Open Source Initiative principles and outreach through training at conferences like AGU Fall Meeting and AMS Annual Meeting.