Cloud simulations improving in climate models

Cloud simulations improving in climate models. An improved understanding of Earth’s climate requires not only good data but also good computer models to make sense of the data. Currently there are about 20 climate models in use around the world that generate the indicators—including temperature, precipitation, clouds, and water vapor—on which the Intergovernmental Panel on Climate Change bases its projections. The largest uncertainties in the IPCC assessments—the most recent was in 2007—arise from how the models handle the complex feedback mechanisms of clouds and water vapor. Until recently, the “ground-truth” data for the simulations were sparse and provided climate scientists with incomplete knowledge. But that has changed with data from the A-Train constellation of satellites during 2006–09. (For an overview of the A-Train, see the article in Physics Today, July 2010, page 36 .) In particular, the observations from space unraveled the three-dimensional structure of clouds and water vapor over the entire globe, providing the basis for a deeper understanding of the relevant microphysics, such as the role of ice crystals. Using the new data, a collaboration of 30 scientists from 9 countries has now made a detailed and quantitative evaluation of both new and old models. The analyses reveal that in 12 cases, the new models are measurably improved over the older versions that contributed to the 2007 IPCC assessment. In general, water vapor is better simulated than clouds, and lower-altitude clouds fare better in the models than those very high up. Individual models were graded, but taking the mean of all models produces the best results, presumably because errors arising from different assumptions and treatments of microphysics average out. (J. H. Jiang et al., J. Geophys. Res. 117, D14105, 2012.)