In this talk, we first describe the important changes that need to be made to the scheme for it to be applicable in the longwave. This development is guided by the following theorem, found to be true from both geometric arguments and Monte-Carlo calculations: for an isolated, isothermal, homogeneous cubic cloud in vacuum, a third of the downwelling emitted radiation beneath the cloud originates from the cloud base and two-thirds from the cloud sides, regardless of the optical depth or scattering properties of the cloud. Furthermore, if this cloud is very optically thick then we can say that the downwelling emitted radiation beneath the cloud will be underestimated by a factor of 3 in calculations using the independent column approximation.
We have used consistency with this theorem as a constraint in the addition of a parameterization in SPARTACUS for the lateral distribution of longwave radiation within the cloud. Our modified 2-stream scheme reproduces well the results from theory and fully 3D calculations. As it is much cheaper than the fully 3D schemes - at only around twice the cost of the standard 2-stream scheme - this makes it suitable to include in a Global Climate Model.
Results will also be presented comparing the new scheme with full 3D radiation calculations for realistic cumulus cloud cases from cloud-resolving model simulations, exploring the role of cloud size, shape and in-cloud inhomogeneity. These results also highlight that the addition of 3D effects leads to a systematic and significant increase in the surface longwave radiative forcing of cumulus clouds, which runs counter to the view often aired that longwave 3D effects can always be safely ignored.