21 May 2013

Climate Model Simulations of Clouds Are Improving

Science

Predictions by climate models of the amount of warming that the planet resulting from an increase in greenhouse gases vary widely due to the different simulated responses of clouds to warming. Model cloud predictions are variable because clouds are among the least well-simulated components of climate models and much effort over many years has gone into improving their simulations by climate models. In this study, we measure whether or not the ability of climate models to simulate clouds has improved in the newest generation of climate models that are being assessed for reports of the Intergovernmental Panel on Climate Change. We examine the ability of 19 climate models to simulate the climatological cloud amount, reflectivity and altitude in comparison with satellite observations from the International Satellite Cloud Climatology Project (ISCCP) and contrast this ability between the newest simulations (from CFMIP2/CMIP5) produced in 2012 with older simulations (from CFMIP1/CMIP3) produced before 2005. An ISCCP simulator is applied to model clouds to facilitate the comparison between models and the ISCCP observations.

Approach

The simulations of clouds in climate models are improving. In the newest models, there is widespread reduction of a bias associated with too many highly reflective clouds, with the best models having eliminated this bias. Figure 1 shows this improvement by displaying how the amount of highly reflective clouds, indicated by the cloud optical depths τ > 23, has been reduced in newer models from individual model families (Figure 1). With increased amounts of clouds with lesser reflectivity, there is a significant reduction in the “too few – too bright” problem where the time-mean radiation balance is well-simulated by having the compensating errors of too few clouds which are too reflective.
 
Quantitatively, scalar measures of model performance for cloud amount and cloud properties demonstrative objective improvement (Figure 2). In this figure, each climate model is plotted with a symbol and stratified into the older (CFMIP1) and newer (CFMIP2) models. Arrows track progress in individual models or the ensemble mean and if they point to left indicate improvement with time. Model improvement is most marked for a scalar measure of cloud properties (right panel) that results mostly from the improvement in cloud reflectance rather than cloud altitude. Errors in cloud amount (left panel) on average show little or no improvement, although greater improvement can be found in individual models.
 

Impact

We conclude that the simulations of clouds in climate models are improving – a necessary but insufficient step towards increased confidence in their predictions.

Summary

This is funded by two projects, but I can't figure out how to insert two projects into the Project tab. The two projects are: "Improving the Characterization of Clouds, Aerosols and the Cryosphere in Climate Models (29)" and "Identifying Robust Cloud Feedbacks in Observations and Models (176)."

Contact
Siyu Chen
Acknowledgments

The efforts of authors from Lawrence Livermore National Laboratory were supported by the Regional and Global Earth and Environmental System Modeling programs of the United States Department of Energy’s Office of Science and were performed under the auspices of the United States Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344.