A33R-03
Do convective schemes substantially alter simulated global climate and cloud feedback?

Wednesday, 16 December 2015: 14:10
3012 (Moscone West)
Steven C Sherwood1, Mark Webb2, Adrian Lock3, Christopher Stephen Bretherton4, Sandrine Bony5, Jason N Cole6, Abderrahmane Idelkadi5, Sarah M Kang7, Tsuyoshi Koshiro8, Hideaki Kawai8, Tomoo Ogura9, Romain Roehrig10, Yechul Shin7, Thorsten Mauritsen11, Jessica Vial5, Masahiro Watanabe12, Matthew Woelfle4, Ming Zhao13, Cameron W Cairns14 and Geoffrey K Vallis15, (1)University of New South Wales, Climate Change Research Centre, Sydney, NSW, Australia, (2)UK MetOffice, Exeter, United Kingdom, (3)Met Office Hadley center for Climate Change, Exeter, United Kingdom, (4)University of Washington Seattle Campus, Seattle, WA, United States, (5)Laboratoire de Météorologie Dynamique UPMC, Paris, France, (6)Environment Canada Toronto, Toronto, ON, Canada, (7)Ulsan National Institute of Science and Technology, Ulsan, South Korea, (8)Meteorological Research Institute, Ibaraki, Japan, (9)NIES National Institute of Environmental Studies, Ibaraki, Japan, (10)CNRM (CNRS and Météo-France), Toulouse, France, (11)Max Planck Institute for Meteorology, Hamburg, Germany, (12)University of Tokyo, Atmosphere and Ocean Research Institute, Bunkyo-ku, Japan, (13)NOAA Geophysical Fluid Dynamis Laboratory (GFDL), Princeton, NJ, United States, (14)University of New South Wales, Climate Change Research Centre, Sydney, Australia, (15)University of Exeter, Exeter, United Kingdom
Abstract:
We investigate CFMIP-2 AMIP and AMIP+4K experiments with ten climate models which have had their convective parametrizations turned off. Previous studies have suggested that parameterized convection may be a leading source of inter-model spread in cloud feedbacks and other aspects. We find that model biases in ‘ConvOff’ runs (those with the convective schemes switched off), for example in tropical lapse rate and relative humidity, are generally smaller than intermodel differences in these fields, in spite of the fact that these fields are widely thought to be controlled by convective processes. This suggests that, to a considerable extent, explicitly resolved convection is able to do most of the overall job of convection in modern climate models at least with respect to phenomena at large scales.
We find moreover that ‘ConvOff’ models have a similar overall range of cloud feedbacks compared to the standard configurations. Furthermore, applying a simple bias correction method to allow for differences in present day global cloud radiative effects substantially reduces the differences between the cloud feedbacks with and without parametrized convection in the individual models. We conclude that, while parametrized convection influences the strength of the cloud feedbacks substantially in some models, other processes must also contribute substantially to the overall inter-model spread.

The positive shortwave cloud feedbacks seen in the models in subtropical regimes associated with shallow clouds are still present in the ConvOff experiments. Inter-model spread in shortwave cloud feedback increases slightly in regimes associated with trade cumulus in the ConvOff experiments but is quite similar in the most stable sub-tropical regimes associated with stratocumulus clouds. Inter-model spread in longwave cloud feedbacks in strongly precipitating regions of the tropics is substantially reduced in the ConvOff experiments however, indicating a considerable local contribution from differences in the details of convective parametrizations.

<< Previous Abstract | Next Abstract >>