On the Nature of Cloud Property Errors in Contemporary Gcms: A Hindcast Approach

Abstract:

Contemporary GCMs exhibit a number of systematic errors in their simulations of cloud properties, as recently documented in Klein et al. (2013): Models do not simulate large enough total cloud fraction, especially over regions of marine stratocumulus on the eastern sides of subtropical ocean basins and throughout the midlatitudes over both ocean and land. Models also produce an over-­‐abundance of thick cloud (optical depth > 23) and an under-­‐abundance of thin cloud (optical depth < 23). It is important to determine the extent to which these biases are due to errors in model physics or due to errors in the simulated large-­‐scale ocean-­‐atmosphere state, as different causes require different solutions. Here we make use of novel simulations in which several CMIP5 GCMs are run in weather forecast mode as part of the Transpose-­AMIP II experiment. As each hindcast is initialized from the operational ECMWF YOTC analysis, large-­scale states remain very close to observations within a few days of model integration. Therefore, errors in the cloud fields with respect to instantaneous ISCCP observations can be largely attributed to model physics errors that arise relatively quickly in the simulations. We find that many of the systematic cloud biases identified in Klein et al. (2013) emerge within the first few days of these T-­AMIP hindcasts, indicating that errors in model physics are the dominant driver of the cloud biases. Using cloud radiative kernels designed specifically for the T-­AMIP runs, we quantify the relative importance of biases in cloud amount, altitude, and optical depth in driving errors in TOA radiation, focusing on key regions including Northern Hemisphere land areas where large summertime TOA SW radiation biases contribute to large warm biases at the surface.

This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-­‐AC52-­‐07NA27344. It is supported by the Regional and Global Climate Modeling Program of the Office of Science at the DOE. IM release number: LLNL-ABS-657939