Sensitivity of marine boundary layer clouds to idealized climate perturbations: The CGILS LES Intercomparison

The sensitivity of marine boundary layer clouds to idealized climate perturbations at three locations along the GCSS Pacific Cross-section is explored in a number of large eddy simulation models as part of CGILS (the CFMIP/GCSS Intercomparison of Large-Eddy and Single-Column Models). These locations vary from well-mixed stratocumulus close to the California coast to decoupled stratocumulus clouds farther offshore and finally a trade wind boundary layer dominated by shallow cumulus convection. The forcings for the present climate are based on the July climatology, while the perturbed climate assumes that sea surface temperature increases by two degrees Kelvin while the reference sounding is warmed moist adiabatically, and the relative humidity stays fixed. No perturbation to CO2 is made in the model intercomparison.

Several large-eddy simulation (LES) models have been run for ten days with both control and perturbed climate forcings at these locations, so that they have nearly reached equilibrium. While these models are subjected to the same large-scale forcings, they differ in their numerical discretizations and in their microphysical and sub-grid scale turbulence parameterizations. The models display broadly consistent behavior in cloud cover and boundary layer structure, and their response to the idealized climate perturbation, providing a useful benchmark for similarly forced single-column models. All models maintain full cloud cover at the two locations closer to the coast, and the inversion deepens in response to reduced subsidence in the perturbed (warmer) climate at these locations. The cloud albedo response to the climate perturbation has an uncertain sign in these cases, with some models having thicker clouds in the warmer climate and other having thinner clouds. However, if the subsidence is unchanged in the warmer climate at the coastal location, the models all agree and show a thinner cloud, implying positive cloud feedbacks at this location under those conditions. Similar responses are found in a mixed-layer model for the coastal location and in a representative LES for the decoupled stratocumulus location. Following the framework suggested by Bony et al (2004, doi:10.1007/s00382-003-0369-6), these results suggest that the thermodynamic component of cloud changes at these locations contributes a positive feedback, while the dynamic component of cloud changes is a negative feedback.

At the trade cumulus location, the cloud albedo and inversion depth change little in the perturbed climate, though some LES hint at a weak positive cloud feedback. In most of the models, the deepening of the inversion is restrained by the strengthening of precipitation at the surface. The precipitation increases in the warmer climate to offset the reduction in the subsidence warming of the boundary layer, so that the inversion height changes are proportionately smaller than those in the weakly or non-precipitating clouds at the other locations. Using a single LES model at this location, the simulated equilibrium inversion depth and the cloud response is found to be sensitive to an imposed change in cloud droplet concentration.