Project Summary: The E3SM-SMYLE is a set of initialized hindcasts produced using the E3SM-v21 low resolution model (initialization years 1980-2020). These hindcasts were specifically designed to explore Earth system predictability at multi-year timescales using forecast lead times ranging from one month to two years. This work contributes to specific goals of the DOE funded CATALYST project, and more broadly to RGMA, by providing climate information with uncertainty estimates to support DOE’s energy missions. The growing societal demand for advance warning of near-term climate extremes takes place against a backdrop of ever-increasing anthropogenic climate change. The field of initialized climate prediction on timescales from subseasonal to decadal has grown to meet this demand. Multi-year Earth system prediction falls ‘between the cracks’ of traditional seasonal and decadal prediction efforts, offering great promise for new research. An analogous large ensemble of multi-year initialized hindcasts using the Community Earth System Model (CESM2) (CESM-SMYLE) has provided significant contributions to our understanding of climate predictability on multi-year timescales [Yeager et al., 2022].

In addition to extensive hindcast output from all E3SM component models, E3SM-SMYLE includes historical reconstructions for the ocean, sea ice, and land component models; an experimental setup that can be replicated and/or modified; and python code for performing post-processing and skill assessment. Details are provided below.

The E3SM-SMYLE was produced by a collaboration between the Cooperative Agreement To Analyze variabiLity, change and predictabilitY in the Earth SysTem (CATALYST) group at the National Center for Atmospheric Research (NCAR) and DOE researchers at the Los Alamos National Lab (LANL). The project and the members of this collaboration were supported under the Department of Energy Office of Science Biological and Environmental Research (BER) Regional and Global Model Analysis (RGMA) and Earth System Model Development (ESMD) components of the Earth and Environmental System Modeling (EESM) Program. This work was also supported by the National Center for Atmospheric Research (NCAR). A 2023-2024 ASCR Leadership Computing Challenge (ALCC) HPC computing allocation award enabled the production of this dataset.


E3SM-SMYLE At A Glance

  • E3SM model: ocean (MPASO, 1°, 61L); atmosphere (EAM, 1°(ne30), 72L); land (ELM); sea ice (MPASSI)
  • Hindcasts initialized quarterly (1st of November) 1970 to 2019
  • 28-month simulations
  • 20-member ensembles
  • EAM initialization: ERA5 Reanalysis
  • MPASO initialization: GMPAS_JRA1-do forced ocean/sea-ice (FOSI) simulation
  • MPASSI initialization: GMPAS_JRA1-do forced ocean/sea-ice (FOSI) simulation
  • ELM initialization: CRU-NCEP forced land simulation
  • Data availability (TBD)
  • Data availability on NERSC CFS: /global/cfs/cdirs/mp9/archive/v21.LR.SMYLE
  • Data availability on NERSC CFS: /global/cfs/cdirs/mp9/archive/v21.LR.SMYLEsmbb

  • Reproducing E3SM-SMYLE:

    The E3SM-SMYLE simulations were conducted on NERSC supercomputing hardware. Bit-for-bit reproducibility may not be possible.
  • Machine: Perlmutter CPU nodes
  • Codebase: E3SMv2.1 (maint-2.1)
  • Initialization: Initialization files available on NERSC CFS: /global/cfs/cdirs/mp9/E3SMv2.1-SMYLE/inputdata/e3sm_init
  • Compsets: WCYCL20TR; WCYCLSSP370
  • Resolution: ne30pg2_EC30to60E2r2
  • Model_start_type: hybrid from initialized conditions
  • Confluence project website

  • Publications and Project Products:

    The following paper documents the E3SM-SMYLE experimental design and provides a broad overview of the E3SMv21 model system skill:

  • Yeager, S. G., N. Rosenbloom, S. Glanville, …. (2024): The E3SM-Seasonal-to-Multiyear Large Ensemble (SMYLE) prediction system using the Energy Exascale Earth System Model Version 21. In prep.
  • The following paper documents the companion CESM-SMYLE experimental design and provides a broad overview of the CESM2 model system skill:

  • Yeager, S. G., N. Rosenbloom, S. Glanville, X. Wu, I. Simpson, H. Li, M. J. Molina, K. Krumhardt, S. Mogen, K. Lindsay, D. Lombardozzi, W. Wieder, W. M. Kim, J. H. Richter, M. Long, G. Danabasoglu, D. Bailey, M. Holland, N. Lovenduski, W. G. Strand, and T. King (2022): The Seasonal-to-Multiyear Large Ensemble (SMYLE) prediction system using the Community Earth System Model Version 2, Geosci. Mod. Dev., 15, 6451-6493, https://doi.org/10.5194/gmd-15-6451-2022

  • Contacts:

  • Steve Yeager [ yeager@ucar.edu ]
  • Nan Rosenbloom [ nanr@ucar.edu ]

  • Acknowledging Use of E3SM-SMYLE Data

    E3SM-S2S uses E3SM version2.1. Reference: Smith et al.(2024): The DOE E3SM Version 2.1: Overview and Assessment of the Impacts of Parameterized Ocean Submesoscales. (Submitted to GMD).

    SMYLE data will be available for download at XXXX. Please add your name to the E3SM-SMYLE analysis registry to help us track who is using this resource. We request that you cite both the data DOI (XXX) as well as the dataset description paper (Yeager et al. 20xx) in work that makes use of E3SM-SMYLE. Finally, we welcome collaboration and information sharing, so please consider joining the CESM Earth System Prediction Working Group and presenting your work at upcoming ESPWG meetings.