An improved cloud droplet number climatology and its application to the validation of climate models

The consequences of the interactions of aerosols and clouds are poorly understood and yet have the potential to play a major role in climate forcing. The most immediate effect of changing CCN concentrations in clouds is to alter the number of droplets that activate within the cloud, although this also depends on the cloud base updraft. There follows a complicated series of interactions and feedbacks that can involve precipitation, cloud thickness and coverage changes, entrainment, radiation, etc. These processes can feed back onto the droplet number concentrations. Thus the correct prediction of droplet numbers by climate models is vital, but requires the correct representation of this large range of processes, as well as the accurate simulation of the aerosol and dynamical fields. In this study we use data from the MODIS satellite to produce a droplet number climatology in which many potential problems with this type of retrieval are examined and remedied through careful data selection. For example, one problem that was recently identified occurred when the angle of the instrument sensor beam was too large resulting in the overestimate of cloud fractions and a lowering of the instrument resolution, which led to biases in the calculated droplet concentrations. The biases occurring in retrievals that take place when the Sun is too low in the sky have also been explored, as have ways to avoid broken or small scale clouds, which can all cause undesirable 3D radiative transfer effects. Thresholds for data removal will be specified through the quantification of these effects and the resulting impact on the robustness of the statistics around the globe will be calculated through error estimates. This improved dataset will be used to validate the new CLUBB unified cloud and turbulence scheme currently being tested in two of the leading US climate models (NCAR CAM & GFDL AM3), as part of the NSF/NOAA Climate Process Team.