Paper Reviewed
Pithan, F., Medeiros, B. and Mauritsen, T. 2014. Mixed-phase clouds cause climate model biases in Arctic wintertime temperature inversions. Climate Dynamics 43: 289-303.

According to Pithan et al. (2014), "temperature inversions are a common feature of the Arctic wintertime boundary layer," and they say that "they have important impacts on both radiative and turbulent heat fluxes and partly determine local climate-change feedbacks," which leads them to further state that "understanding the spread in inversion strength modelled by current global climate models is therefore an important step in better understanding Arctic climate and its present and future changes." In a quest to obtain that "better understanding," Pithan et al. go on to show "how the formation of Arctic air masses leads to the emergence of a cloudy and a clear state of the Arctic winter boundary layer," and they describe the different climatic implications of each of these states.

In the Arctic's cloudy state, the three researchers find "little to no surface radiative cooling occurs and inversions are elevated and relatively weak," whereas in the Arctic's clear state, they find "surface radiative cooling leads to strong surface-based temperature inversions." And when comparing specific aspects of model output to real-world observations, they find that (1) the "freezing of super-cooled water at too warm temperatures that occurs in many CMIP5 models leads to a lack of high-emissivity mixed-phase clouds and thus of a cloudy state in these models," and that (2) "models lacking a cloudy state display excessive surface radiative cooling in Arctic winter, which tends to produce strong low-level stability and temperature inversions."

In addition, they report that (3) "few models that allow for cloud liquid water at very low temperatures reproduce both the clear and cloudy state of the boundary layer," that (4) "a second group of models lacks the cloudy state and exhibits strong stability and strong long-wave cooling, that (5) "other models also lack the cloudy state, but generate weak stability despite strong long-wave cooling," that (6) the CMIP5 inter-model spread of typical monthly-mean low-level stability over sea ice in winter is about 10 K, which is similar to that in CMIP3 models," that (7) "15 out of 21 CMIP5 models overestimate low-level stability over sea ice compared to reanalysis data," that (8) "this wide-spread model bias is linked to shortcomings in the representation of mixed-phase cloud microphysics," and that (9) "differences in cloud properties, energy fluxes and inversion strengths between land and sea ice domains remain to be investigated."

In light of these several findings, Pithan et al. state, in the concluding sentence of their paper, that "in order to better represent the Arctic winter boundary layer and surface energy budget in climate models, an important step would be to improve the mixed-phase cloud microphysics and to obtain an adequate representation of the cloudy state." And so we discover another glaring example of the fact that today's CMIP5 models are still not up to the task of adequately portraying Earth's current climate, which must surely be done before we can rely on them to provide valid portrayals of Earth's future climatic state.