Climate Processes in CMIP5: Precipitation dynamics in high-resolution CAM simulations

We examine simulated precipitation in the Community Atmosphere Model (CAM) at horizontal resolutions near 25 km. Overall, seasonal mean distributions of precipitation do not exhibit large improvements at this resolution compared with those obtained at typical climate model resolutions ~100 km. However, the high-resolution simulations do obtain realistic distributions of tropical cyclone activity, as well as, realistic structure and evolution for individual storms. Other phenomena such as organized propagating convection also begin to appear. Examination of the simulated precipitation fields shows that, as a rule, these “interesting” mesoscale features appear to be driven by the model's large-scale condensation routines, rather than parameterized convection. We analyze relationships between precipitation and other variables, such as low-level divergence, from hourly to seasonal time-scales. We also present results of sensitivity studies intended to probe the role of the model's deep convection parameterization. In particular, we will show that reducing the convective adjustment time-scale, i.e., making the model's deep scheme more active, improves seasonal mean precipitation, but nearly eliminates tropical cyclone activity.