The Community Atmosphere Model, version 4 (CAM4), was released as part of the Community Climate System Model, version 4 (CCSM4). The finite volume (FV) dynamical core is now the default because of its superior transport and conservation properties. Deep convection parameterization changes include a dilute plume calculation of convective available potential energy (CAPE) and the introduction of convective momentum transport (CMT). An additional cloud fraction calculation is now performed following macrophysical state updates to provide improved thermodynamic consistency. A freeze-drying modification is further made to the cloud fraction calculation in very dry environments (e.g., the Arctic), where cloud fraction and cloud water values were often inconsistent in CAM3. In CAM4 the FV dynamical core further degrades the excessive trade-wind simulation, but reduces zonal stress errors at higher latitudes. Plume dilution alleviates much of the midtropospheric tropical dry biases and reduces the persistent monsoon precipitation biases over the Arabian Peninsula and the southern Indian Ocean. CMT reduces much of the excessive trade-wind biases in eastern ocean basins. CAM4 shows a global reduction in cloud fraction compared to CAM3, primarily as a result of the freeze-drying and improved cloud fraction equilibrium modifications. Regional climate feature improvements include the propagation of stationary waves from the Pacific into midlatitudes and the seasonal frequency of Northern Hemisphere blocking events. A 1° versus 2° horizontal resolution of the FV dynamical core exhibits superior improvements in regional climate features of precipitation and surface stress. Improvements in the fully coupled mean climate between CAM3 and CAM4 are also more substantial than in forced sea surface temperature (SST) simulations.

The NCAR Community Climate System Model, version 3 (CCSM3) exhibits persistent errors in its simulation of the El Niño–Southern Oscillation (ENSO) mode of coupled variability. The amplitude of the oscillation is too strong, the dominant 2-yr period too regular, and the width of the sea surface temperature response in the Pacific too narrow, with positive anomalies extending too far into the western Pacific. Two changes in the parameterization of deep convection result in a significant improvement to many aspects of the ENSO simulation. The inclusion of convective momentum transport (CMT) and a dilution approximation for the calculation of convective available potential energy (CAPE) are used in development integrations, and a striking improvement in ENSO characteristics is seen. An increase in the periodicity of ENSO is achieved by a reduction in the strength of the existing “short-circuited” delayed-oscillator mode. The off-equatorial response is weaker and less tropically confined, largely as a result of the CMT and an associated redistribution of zonal momentum. The Pacific east–west structure is improved in response to the presence of convective dilution and cooling provided by increased surface fluxes. The initiation of El Niño events is fundamentally different. Enhanced intraseasonal surface stress variability leads to absolute surface westerlies and a cooling–warming dipole between the Philippine Sea and western Pacific. Lag-regression analysis shows that intraseasonal variability may play a significant role in event initiation and maintenance as opposed to being a benign response to increased SSTs. Recent observational evidence appears to support such a leading relationship.