Clouds play a pivotal role in the surface energy budget of the polar regions. Here we use two largely independent data sets of cloud and surface downwelling radiation observations derived by satellite remote sensing (2007–2010) to evaluate simulated clouds and radiation over both
...
Clouds play a pivotal role in the surface energy budget of the polar regions. Here we use two largely independent data sets of cloud and surface downwelling radiation observations derived by satellite remote sensing (2007–2010) to evaluate simulated clouds and radiation over both polar ice sheets and oceans in state-of-the-art atmospheric reanalyses (ERA-Interim and Modern Era Retrospective-Analysis for Research and Applications-2) and the Coupled Model Intercomparison Project Phase 5 (CMIP5) climate model ensemble. First, we show that, compared to Clouds and the Earth's Radiant Energy System-Energy Balanced and Filled, CloudSat-CALIPSO better represents cloud liquid and ice water path over high latitudes, owing to its recent explicit determination of cloud phase that will be part of its new R05 release. The reanalyses and climate models disagree widely on the amount of cloud liquid and ice in the polar regions. Compared to the observations, we find significant but inconsistent biases in the model simulations of cloud liquid and ice water, as well as in the downwelling radiation components. The CMIP5 models display a wide range of cloud characteristics of the polar regions, especially with regard to cloud liquid water, limiting the representativeness of the multimodel mean. A few CMIP5 models (CNRM, GISS, GFDL, and IPSL_CM5b) clearly outperform the others, which enhances credibility in their projected future cloud and radiation changes over high latitudes. Given the rapid changes in polar regions and global feedbacks involved, future climate model developments should target improved representation of polar clouds. To that end, remote sensing observations are crucial, in spite of large remaining observational uncertainties, which is evidenced by the substantial differences between the two data sets.
@en
