Longwave 3D Benchmark Calculations for Inhomogeneous Cloud Fields

As an extension of the shortwave phase of the third Intercomparison of Radiation Codes in Climate Models (ICRCCM III) by Barker et al. [1], this study establishes longwave benchmarks for the same inhomogeneous cloud fields used by that intercomparison and examines the performance of three approximate, one-dimensional (1D) radiation methods. The benchmarks are calculated using a correlated-k three-dimensional Monte Carlo (3DMC) algorithm that is validated against line-by-line calculations for simple atmospheres. The approximate methods include an independent pixel approximation (IPA) and two cloud-overlap schemes: maximum/random (MRO) and random (RO). With respect to the 3DMC benchmarks, the IPA consistently outperforms the MRO and RO since it preserves the horizontal inhomogeneity of the cloud field, whereas the two cloud-overlap techniques assume plane-parallel homogeneous atmospheres. The results from the open cellular convection case are presented here. This case has a horizontal domain of (50 km)² with 0.4 km resolution and 74 vertical layers with depths of 0.15 km through the clouds. The downward surface flux error is -4.7 W/m² for IPA, -14.7 W/m² for MRO and 5.4 W/m² for RO. These differences can be attributed to the neglected 3D effects of cloud radiation, and indicate the need to refine longwave 1D climate radiation codes so that they compensate for broken clouds.