Research ArticleCLIMATOLOGY

Simulation of Eocene extreme warmth and high climate sensitivity through cloud feedbacks

See allHide authors and affiliations

Science Advances  18 Sep 2019:
Vol. 5, no. 9, eaax1874
DOI: 10.1126/sciadv.aax1874
  • Fig. 1 Model-data comparison of the Early Eocene GMST and relative meridional SST gradient.

    (A) GMST (in °C) in model simulations (markers) as a function of CO2 concentration compared with proxy estimates of the Early Eocene (gray patch) and the pre-PETM to PETM warming (black arrow). (B) Relative latitudinal gradient of SST in model simulations (markers) as a function of CO2 compared with estimates from proxies (gray patch). The CO2 ranges for the Early Eocene and PETM are from the latest estimates from proxies (1) and modeling (8), respectively. Eocene CESM1.2 simulations from this study are denoted as filled red circles. For reference, the PI simulation is marked as a black circle. Previous Eocene simulations (16, 19) including those from the Eocene Modeling Intercomparison Project (EoMIP) (3) are denoted by gray open markers. Note that methane concentrations in previous simulations that are different from the PI value have been converted to the equivalent CO2 concentration (Materials and Methods). In CCSM3KS simulations, the authors altered cloud properties substantially to account for presumed Eocene aerosol changes (16). Similarly, model physics are significantly altered in FAst Met Office/UK Universities Simulator (FAMOUS) (E17) (19).

  • Fig. 2 Model simulated zonal mean temperature over land and ocean compared with proxy estimates.

    (A) Zonal mean land surface temperature in the Eocene simulations compared with Early Eocene terrestrial proxy evidence (11). (B) Zonal mean SST in the Eocene simulations compared with published Early Eocene SST data, compiled in this study (table S1). Inferred temperatures using δ18O of planktic foraminifera, clumped isotopes, Mg/Ca of planktic foraminifera, and TEX86 are denoted as filled circles, upward-pointing triangles, squares, and downward-pointing triangles, respectively.

  • Fig. 3 ECS and climate feedback parameters in the Eocene simulations.

    (A) The ECS (in °C) in the PI (black circle) and Eocene 1×, 3×, and 6× simulations (red filled circles) as a function of the atmospheric CO2 concentration. For comparison, ECSs in previous Eocene simulations (3) and in CAM4 slab ocean simulations are shown as gray open markers and blue filled circles, respectively. ECS for Eocene 6× is shown as a smaller marker than the other Eocene simulations, as it is estimated from the 6× and 9× experiments instead of using slab ocean simulations (Materials and Methods). (B) Climate feedback parameters (in W m−2 K−1) diagnosed from the two-way PRP method as a function of the atmospheric CO2 concentration in the Eocene simulations. Surface albedo (ALB), cloud (CLD), lapse rate (LPR), Planck (PLK), and water vapor (WVP) feedback parameters are shown. The cloud feedback parameter is further decomposed into shortwave and longwave components. For reference, feedback parameters in the PI simulation are denoted as open markers. Detailed numbers and uncertainty are shown in table S3. (C) Contributions to the shortwave cloud feedback parameter from cloud fraction, scattering, and absorption properties calculated from the approximated PRP method using the Eocene 1× and 9× simulations.

  • Fig. 4 Zonal mean cloud cover, liquid water content, effective droplet size, and droplet number concentration in the Eocene 1× and 6× simulations.

    (A) Zonal mean cloud cover (in %) in the Eocene 1× simulation. (B) As in (A) but for the Eocene 6× experiment. (C) Zonal mean in-cloud liquid water content (in g kg−1) in the Eocene 1× simulation. (D) As in (C) but for the Eocene 6× experiment. (E) Zonal mean in-cloud liquid droplet radius (in μm) in the Eocene 1× simulation. (F) As in (E) but for the Eocene 6× experiment. (G) Zonal mean in-cloud liquid droplet number concentration (in cm−3) in the Eocene 1× simulation. (H) As in (G) but for the Eocene 6× experiment. Zonal means of in-cloud variables are calculated for grid points with cloud liquid water greater than 1 parts per million.

  • Fig. 5 Increase in in-cloud accretion and autoconversion rates with warming in the Eocene simulations.

    (A) Zonal mean in-cloud accretion rate (in g kg−1 day−1; conversion of cloud water to precipitation through rain droplets intercepting and collecting small cloud droplets) in the Eocene 1× simulation. (B) As in (A) but for the Eocene 6× experiment. (C) Zonal mean in-cloud autoconversion rate (in g kg−1 day−1; conversion of cloud water to precipitation by coalescence and vapor diffusion) in the Eocene 1× simulation. (D) As in (C) but for the Eocene 6× experiment. Zonal mean of in-cloud accretion and autoconversion rates are calculated for grid points with cloud cover greater than 1%.

  • Fig. 6 Role of CAM5 physical parameterizations in changing low-cloud cover.

    Changes in low-cloud cover (in %) versus the GMST (in °C) in the Eocene atmosphere-only simulations using CAM4 with individual CAM5 physical parameterizations switched on sequentially (Materials and Methods). The physical parameterizations of radiation transfer (rad), cloud microphysics (micro), turbulence and shallow convection (uw), and cloud macrophysics (macro) are tested. For comparison, values of low-cloud cover have been realigned by subtracting the corresponding value in the 1× simulation. For reference, standard atmosphere-only simulations using CAM5 (red filled circles) and CAM4 (orange filled circles) are also shown.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/9/eaax1874/DC1

    Fig. S1. Topography and bathymetry at model resolution.

    Fig. S2. Spin-up of the Eocene simulations.

    Fig. S3. Model-data comparison of Early Eocene surface temperature.

    Fig. S4. Model-data comparison of PETM warming in SST.

    Fig. S5. Comparison of CAM5 and CAM4 Eocene atmosphere-only simulations.

    Fig. S6. Comparison of temperature and shortwave cloud feedback parameters between CAM5 and CAM4 Eocene SOM simulations.

    Fig. S7. Increase in climate sensitivity and cloud feedback parameter with warming under modern conditions.

    Table S1. Compilation of SST proxies for the Early Eocene.

    Table S2. Compilation of SST proxies for the pre-PETM and PETM.

    Table S3. Feedback analysis from the PRP calculations.

  • Supplementary Materials

    The PDF file includes:

    • Fig. S1. Topography and bathymetry at model resolution.
    • Fig. S2. Spin-up of the Eocene simulations.
    • Fig. S3. Model-data comparison of Early Eocene surface temperature.
    • Fig. S4. Model-data comparison of PETM warming in SST.
    • Fig. S5. Comparison of CAM5 and CAM4 Eocene atmosphere-only simulations.
    • Fig. S6. Comparison of temperature and shortwave cloud feedback parameters between CAM5 and CAM4 Eocene SOM simulations.
    • Fig. S7. Increase in climate sensitivity and cloud feedback parameter with warming under modern conditions.
    • Table S1. Compilation of SST proxies for the Early Eocene.
    • Table S2. Compilation of SST proxies for the pre-PETM and PETM.
    • Table S3. feedback analysis from the PRP calculations.

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Table S1 (Microsoft Excel format). Compilation of SST proxies for the Early Eocene.
    • Table S2 (Microsoft Excel format). Compilation of SST proxies for the pre-PETM and PETM; feedback analysis from the PRP calculations.

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article