CESM2 Single Forcing Large Ensemble Project

The CESM2 “Single Forcing” Large Ensemble Project is a publicly available set of model simulations useful for addressing the roles of individual forcings in historical and future change. This simulations use the same model and forcings as the CESM2 Large Ensemble Project and, therefore, can be used to parse the relative roles of different forcings in responses found in that ensemble where all forcings are applied together. The ensemble members are initialized from 1850 from the same initial conditions that were used to initialize the "macro" members of the CESM2 Large Ensemble and they extend to 2050, following CMIP6 historical forcings prior to 2015 and SSP3-7.0 forcings, thereafter. Note that the smoothed biomass burning emissions that were used in the second 50 members of the CESM2 large ensemble are used, so these simulations should be compared with the second 50 members of the CESM2 large ensemble. Four primary ensembles are available in which different forcings are time evolving while all other forcings are held fixed at 1850's values i.e, the "only" method is used. Note that this differs from the CESM1 Large Ensemble which used the "all-but-one" method where all forcings were evolving except the one of interest. In the CESM2 ensembles, only the forcing of interest is evolving. Four ensembles are available using the following time-evolving forcings:

  • GHG = only greenhouse gases evolving (15 members)
  • AAER = Only anthropogenic aerosols evolving (20 members)
  • BMB = Only biomass burning aerosols evolving (15 members)
  • EE = everything else evolving i.e., all forcings other than those that are time evolving in GHG, AAER or BMB are time evolving. Greenhouse gases and anthropogenic and biomass burning aerosols are held fixed (15 members)

Note that by "anthropogenic aerosols" here we refer to all industrial, agricultural, domestic and transport related emissions and acknowledge that this does not include any anthropogenic influences on biomass burning, which is instead present in the BMB ensemble. Each forcing that is time evolving in the CESM2 Large Ensemble is time evolving in one of these four sub-ensembles to allow additivity to be tested. As discussed in the release paper cited below, substantial non-linearities are present in the CESM2 Single Forcing Large Ensemble. As part of the investigation into this non-linearity, an additional ensemble was performed which was initialized in 1920 of the CESM2 Large Ensemble members and run with all forcings except anthropogenic aerosols i.e., following the same "all-but-one" approach that was used in CESM1. This demonstrated that there is considerable sensitivity of the anthropogenic aerosol forced response to the method used i.e., "all-but-one" versus "only". We have, therefore, expanded this ensemble to a 10 member ensemble, referred to as xAER, so that any sensitivity of the anthropogenic aerosol influence to the method used can be tested.

  • xAER = Everything time evolving except anthropogenic aerosols (10 members)

When presenting or publishing results based on the CESM2 Single Forcing Large Ensemble we ask that you acknowledge the CESM project and CISL supercomputing resources (doi:10.5065/D6RX99HX), as well as reference the CESM2 single forcing release paper and the dataset DOI (10.26024/yw4w-1w27). For the CESM1 simulations that this manuscript compares to, see the CESM1 Single Forcing Large Ensemble.

Project Details

  • Simulation Names:
    • AAER: b.e21.B1850cmip6.f09_g17.CESM2-SF-AAER.0XX / b.e21.B1850cmip6.f09_g17.CESM2-SF-AAER-SSP370.0XX, XX = 01-20
    • BMB: b.e21.B1850cmip6.f09_g17.CESM2-SF-BMB.0XX / b.e21.B1850cmip6.f09_g17.CESM2-SF-BMB-SSP370.0XX, XX = 01-15
    • EE: b.e21.B1850cmip6.f09_g17.CESM2-SF-EE.1XX / b.e21.B1850cmip6.f09_g17.CESM2-SF-EE-SSP370.1XX, XX = 01-15
    • GHG: b.e21.B1850cmip6.f09_g17.CESM2-SF-GHG.0XX / b.e21.B1850cmip6.f09_g17.CESM2-SF-GHG-SSP370.0XX, XX = 01-15
    • xAER: b.e21.BHISTcmip6.f09_g17.CESM2-SF-xAER.0XX / b.e21.BSSP370cmip6.f09_g17.CESM2-SF-xAER.0XX, XX = 01-10, note that despite these simulations having the compset name BHISTcmip6, they do use the smoothed biomass burning forcing i.e., the biomass burning forcing from the BHISTsmbb compset.
  • Model Version: CESM2.1.2 Codebase | Documentation
  • Resolution: 0.9x1.25_gx1v7 (CESM nominal 1o grid)
  • Years: 1850-2050 (AAER, BMB, EE and GHG), 1920-2050 (xAER)
  • Ensemble Size: 20 members (AAER), 15 members(BMB, EE and GHG), 10 members (xAER)
  • Time Frequencies Saved: Monthly, Daily
  • Machine: NCAR:Cheyenne
  • Compsets: B1850cmip6 with extensive modifications
  • Additional Notes: All simulations follow the CMIP6 forcing protocols for the forcing that is evolving. The only exception is the BMB simulations where the smoothed biomass burning forcing that was used in the second 50 members of the CESM2 large ensemble was used. This affects the late 20th and early 21st century forcing during the GFED era, so when comparing this ensemble during this time period with the "all forcing" large ensemble, the second 50 members of the "all forcing" large ensemble should be used.

Data Acquisition

  • NCAR Internal

    Location on NCAR's campaign store (accessible from NCAR CISL machines):  /glade/campaign/collections/rda/data/d651055

  • Web Access

    This data is available from the NSF NCAR Research Data Archive.

Known Issues

  • The historical portion of the EE simulations has an issue with the specification of land surface types over the ice shelves of Antarctica.  Over the historical portion, these regions were erroneously specified as wetland, whereas they should have been specified as Glacier.  Over the SSP portion of the simulations the land surface type in these regions is specified correctly.  The result is that there is a discontinuity in the specification of the land surface type at January 1st 2015 which will lead to discontinuity in quantities such as globally averaged terrestrial water storage and likely has some small impacts on the Earth System through effects such as Albedo.  This is not an issue in the other single forcing simulations or the CESM2 large ensemble, but the same issue is present in the historical simulations that are part of the CMIP6 DECK.