PNNL

Abstract

Radiative heating associated with the variability of water vapor and clouds in the atmosphere is a principal driver of tropical circulation. Models must produce cloud and radiative heating rate profiles with realistic horizontal, vertical, and diurnal variability in order to produce realistic tropical circulations and cloud feedbacks. A recent study has indicated that the inability of many models to simulate realistic representations of the Madden-Julian Oscillation (MJO) may be caused by systematic diabatic heating profile errors. One of the primary difficulties in producing accurate heating rate profiles within a large-scale general circulation model (GCM) is the sub-grid scale nature of cloud processes and their interactions with radiation. A new approach to climate modeling, the Multi-Scale Modeling Framework (MMF), reduces the need for sub-grid scale cloud parameterizations by replacing the cloud and radiation parameterizations of a GCM with a 2-D cloud system resolving model. The long time series of cloud radar observations at the ARM tropical sites provide an unprecedented dataset for directly calculating radiative heating rate profiles with high temporal and vertical resolution. In this study, we compare radiative heating rate profiles calculated from ARM cloud observations at the Nauru and Manus sites to the model output from the MMF and its parent model, the NCAR Community Atmosphere Model (CAM 3.0). During the study period, the Nauru site was experiencing suppressed conditions while the Manus site had more active convection, leading to very different average radiative heating rate profiles at the two sites. We examine the differences in the observations and model output during these two meteorological regimes. Features of the average heating rates as well as the details of the diurnal cycle are examined. Initial results indicate that differences in the cloud amounts and cloud properties produced in the two models due to their different treatment of cloud processes lead to large differences in the average heating rate profiles. Both sets of model results fail to capture some of the structure of the observed heating because their vertical resolution is too coarse to fully resolve shallow boundary layer clouds and the observed mid-level cloud feature near the freezing level. The differences in the resulting radiative heating rate profiles may have important impacts on the model dynamics.