Southeast Pacific stratocumulus as depicted in two versions of the Community Atmosphere Model

There are two current versions of the Community Atmosphere Model (CAM versions 4 and 5), which are profoundly different in their parameterized physics. Both versions have biases in their representation of subtropical stratocumulus, clouds that are infamously difficult to capture in large-scale models. Using a short-term forecast framework, we investigate the processes involved in the formation and maintenance of stratocumulus in these two models. October 2006 in the southeast Pacific is chosen as the study period, corresponding to the annual peak in stratocumulus extent and also with a previous comparison of models (PreVOCA). The forecasts are initialized with a realistic state from operational analyses, giving the models a fair chance to foster realistic clouds. While the large-scale environment only slowly evolves over the course of the 5-day forecasts, the cloud fields are quickly established and evince the models' long-term biases. Thus the forecast framework allows an evaluation of the parameterized physics given a range of realistic large-scale flows. Understanding the evolution of the forecasts provides some insight into the origin of the models' climate biases.