Accounting for the subgrid-scale variability of surface radiative properties in GCMs

Within atmospheric radiative transfer codes, the surface boundary condition for each gridbox of a global climate model is assumed to be well represented by the mean surface properties (albedo and emissivity). This is assumed to be the case regardless of the complexity of the surface (multiple surface types) and presence of potentially well defined boundaries such as coast lines or sea-ice. We present an approach to sample the surface properties stochastically similar to that used to sample the unresolved cloud structure in the Monte Carlo Independent Column Approximation (McICA). Results of this approach from single column radiative transfer code will be presented a well as the effect of implementing this technique in the Canadian Atmospheric Model.