Tropical intraseasonal variability improvements across the suite of CESM1 atmosphere models

By most measures tropical atmosphere intraseasonal variability is poorly simulated in all configurations of the Community Climate System Model, Version 3 (CCSM3). In particular the dominant model of variability associated with the Madden Julian Oscillation (MJO) is too weak and its phase speed too fast, so as to be indistinguishable from the wave speeds and length scales associated with atmospheric Kelvin waves. Advancing to the Community Atmosphere Model, version 4 (CAM4) used within CCSM4, incorporates a significant set of modifications to the deep convection parameterization that couples higher sensitivity to troposphere moisture with the dynamical shear impacts of convective momentum transports (CMT). We show that enhancing sensitivity to humidity increases tropical variance at most time and space scales. Coupled with CMT the enhanced variance is focused at MJO-specific time and space scales, resulting in an MJO phenomenon that is by several metrics comparable to observations. The differences in response between CAM3 and CAM4 are clearly revealed in Cloud-Associated Parameterization Testbed (CAPT) initialized experiments. CAM3 acts to rapidly damp initialized MJO-type wave structures via spurious convective heating, whereas CAM4 is able to better couple to the wave structures and maintain their fidelity through several days of forecast. In addition to changes associated with CAM4 physics we will summarize model performance across the suite of CESM1 models, demonstrating the response to coupling with an interactive ocean, varying horizontal resolution, varying model top, the inclusion of chemistry and using the advanced physics package in CAM5.