Effects of horizontal and vertical cloud structure in a ten-year climate simulation

The interaction of clouds and radiation presents a significant challenge in general circulation models (GCMs). Accurate representation of clouds is imperative if the model is to reliably describe this interaction. Most GCMs use horizontally homogeneous, plane-parallel clouds that are overlapped using maximum-random overlap, whereby vertically continuous clouds are overlapped maximally and clouds separated by layers of clear sky are overlapped randomly. These approximations have been shown in numerous studies to be unrealistic and to generate large biases in radiation budget.

Here, we make use of the new “Tripleclouds” scheme for horizontal cloud structure (Shonk and Hogan, 2008) in combination with “exponential-random” overlap (Shonk et al, 2010) to separately investigate the effect on global radiation budget of poor representation of horizontal and vertical cloud structure. Offline radiation calculations with cloud fields of re-analysis data from the ECMWF showed that the effect of introducing horizontal inhomogeneity (the “horizontal shift”) is to reduce global net cloud radiative forcing by 4.13 W m^—2, while the effect of improving the vertical overlap assumption (the “vertical shift”) is to increase global mean net cloud radiative forcing by 2.25 W m^—2. This implies that fixing just one element of cloud structure in a GCM could actually generate radiation fields that either overcompensate for the plane-parallel bias, or increase the magnitude of the overall bias.

These two parametrisations have recently been implemented into the Met Office Unified Model and ten-year climate simulations performed. When averaged over the ten years, the horizontal and vertical shifts in net global cloud radiative forcing are found to be smaller in magnitude (1.14 W m^—2 and 0.48 W m^—2 respectively). However, the mean surface temperature at the winter poles is found to be reduced by up to 3°C when horizontal inhomogeneity is included, and increased by up to 3°C when exponential-random overlap is applied. These are caused by changes in long-wave surface cooling instigated by the modifications to the clouds. Temperature shifts of the corresponding sign are also observed in the upper tropical troposphere, which are found to be primarily attributable to the differing radiative treatment of the clouds. It is also notable that the shifts in cloud fraction in the upper tropical troposphere act to oppose the temperature change, while the shifts in the polar boundary-layer cloud fraction act to enhance the change.