Evaluating similar large-eddy simulation representations of cloud regime transitions against DOE ARM observations

Poster PDF

Description

Global Climate Models (GCMs) continue to struggle representing important boundary-layer clouds and the transitions to deeper cloud types. Over the continent, the deepening of shallow cumulus clouds are a regular occurrence and part of the diurnal cycle of convection, as they deepen into congestus and eventually to deep, precipitating cumulonimbus. The difficulty in representing cloud and cloud transitions persists even for high-resolution cloud process models that are used to construct data sets employed for formulating and evaluating GCM parameterizations. Cumulus transitions in process models are predominantly forced using the diurnal cycle in surface fluxes, while making additional assumptions for large-scale homogeneity and hence weak horizontal advective tendencies. Relaxing the assumption of weak horizontal advective tendencies requires forcing in the form of advective tendencies of temperature and moisture and large-scale vertical motion. The ARM variational analysis product has often provided these forcing terms. Recently, Mechem et al. [2015] explored the impact of details in forcing on the temporal evolution of highly transient cloud system properties from using high-resolution large-eddy simulation of a case of shallow and congestus convection sampled during MC3E. Here we show results from a number of sensitivity experiments aimed at further improving simulation metrics and ameliorating the overly rapid transition to congestus convection found in Mechem et al. Based on typical integral cloud-system metrics (e.g., total precipitation), we find that simulations resulting from different forcing configurations can be remarkably similar, at least superficially. A deeper look into the cloud-system behavior reveals important differences in the evolution of the distribution of cloud depths, instability metrics, and cloud spatial organization. These results suggest that, indeed, the quality of the forcing matters and sophisticated instruments (such as those available in ARM) that can characterize both the spatial and temporal aspects of cloud system behavior are necessary to constrain process models.