Origin of climate sensitivity differences: role of selected radiative processes in two GCMs

• Published in: Tellus A: Dynamic Meteorology and Oceanography
• Main Authors: Cash, B., Schneider, E., Bengtsson, L.
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2007
• Subjects: Sea ice
• Online Access: http://hdl.handle.net/11858/00-001M-0000-0011-FBDF-7; http://hdl.handle.net/21.11116/0000-000B-6DE3-B

Model differences in projections of global mean and regional climate change due to increasing greenhouse gases are investigated using two atmospheric general circulation models (AGCMs): ECHAM4 (Max Planck Institute, version 4) and CCM3 (National Center for Atmospheric Research Community Climate Model version 3). We replace the ECHAM4 short-wave processes (including routines for short-wave radiation, aerosols, cloud liquid water path and cloud droplet size distribution) with the corresponding parametrizations from CCM3. We also eliminate sea-ice in both models. We find that the resulting 'hybrid'-ECHAM4 model has the same global mean temperature sensitivity (defined as the difference in temperature between the 2x CO2 and 1x CO2 integrations at equilibrium) and similar regional temperature change patterns as CCM3. The global mean precipitation sensitivity was only slightly affected; indicating different processes control this. Investigation of top of the atmosphere radiative feedbacks in the standard-ECHAM4 and hybrid-ECHAM4 models show that the differences in global mean temperature sensitivity and regional temperature change patterns can be attributed primarily to a stronger, negative, cloud short-wave feedback in the tropics of the hybrid-ECHAM4 model. However, comparison of the hybrid-ECHAM4 model to CCM3 reveals large differences in partitioning of the cloud feedbacks between long-wave and short-wave in the two models. This suggests that the global mean temperature sensitivity and regional temperature change patterns respond primarily to the magnitude and distribution of the top of the atmosphere feedbacks and are relatively insensitive to the partitioning between individual processes.