Process-level improvements in CMIP5 models and their impact on Southern Ocean, Monsoon, and cloud climatology performance

• Main Authors: Lauer, Axel, Eyring, Veronika, Jones, Colin, Evaldsson, Martin, Hagemann, Stefan, Martin, Gill, Roehrig, Romain
• Format: Conference Object
• Language: unknown
• Published: 2016
• Subjects: Erdsystem-Modellierung; Southern Ocean
• Online Access: https://elib.dlr.de/105147/

The performance of 4 improved European climate models from the CNRM, EC-Earth, HadGEM, and MPI-ESM families is assessed in comparison to their predecessor versions used in the 5th Coupled Model Intercomparison Project (CMIP5). The ESMValTool (Eyring et al., 2016) is applied to evaluate a range of variables and climate phenomena in the models against a suite of observations. Analyzed simulations include AMIP-style experiments, with prescribed sea surface temperatures, as well as fully coupled experiments (historical runs). The study focuses on a number of key processes/variables that are known to exhibit systematic biases in present-day GCMs. These include: (a) Southern Ocean clouds and radiation, (b) The South Asian and West African monsoons, and (c) The global cloud climatology and cloud radiative forcing. The main aim is to quantify and document improvements made as a result of model development efforts over the period of the EU FP7 project EMBRACE. The analysis shows that the tropical precipitation in most models is significantly improved, which can be partly attributed to improved representation of convective precipitation in the models. Cloud amounts and simulated cloud-radiation interactions were also quite significantly improved over the Southern Ocean (latitude band 30°S to 65°S), although problems do still exist in the latitude band 50°S to 65°S. The simulated global cloud climatology, however, did not improve significantly. An amplified underestimate of the cloud liquid water path in tropical regions suggests that efficiency of convective precipitation formation is overestimated in some models. Some improvement was seen in the simulation of the South Asian and West African monsoons, although large systematic biases remain in regional details of precipitation and the overall timing of monsoon rainfall.