Current Large Eddy Simulation (LES) codes typically consider radiative transfer in one dimension only. The Intercomparison of 3D Radiation Codes (I3RC) Project has applied sophisticated 3D radiative transfer methods to compare fluxes and heating rates to those computed by their simple 1D, independent pixel approximation (IPA) counterparts. The I3RC results cast some measure of doubt on the reasonableness of using 1D codes to force LES models and imply that a full 3D treatment of radiative transfer might have an effect on simulated cloud fields. If this effect is significant, it would mean that neglecting multi-dimensional radiative transfer (MDRT) effects could lead to a systematic bias in LES results.

In order to address the interactive and evolutionary nature of this problem, we have coupled to an LES the sophisticated MDRT scheme of Evans (Spherical Harmonics Discrete Ordinate Method; 1998). Because of computational expense, we are at this time only able to run 2D experiments. Preliminary runs consider only the broadband longwave component, in large part because IR cloud top cooling is the significant forcing mechanism for marine stratocumulus.

Little difference is noted in the evolution of unbroken stratocumulus between three hour runs using MDRT and IPA for 2D domains of 50 km in the horizontal and 1.5 km in the vertical. Local heating rates differ slightly near undulating regions of cloud top, and a slight bias in mean heating rate from 1 to 3 h is present, yet the differences are never strong enough to result in a pronounced evolutionary bias in typical boundary layer metrics (e.g. inversion height, vertical velocity variance, TKE). Longer integration times may eventually produce a physical response to the bias in radiative cooling rates.

A low-CCN case, designed to produce significant drizzle and induce cloud breakup, does show subtle differences between MDRT and IPA. Over the course of the 6 hour simulations, entrainment is slightly less in the MDRT case, and the transition to the surface-based trade cumulus regime is delayed. Mean cooling rates appear systematically weaker in the MDRT case, indicative of a less energetic PBL and reflected in profiles of vertical velocity variance and TKE. If the effect is sustained over a longer duration, it may lead to less persistent and faster dissipating boundary layer clouds when full MDRT is considered, resulting in a smaller global cooling effect of marine stratocumulus.