The influence of surface texture on the effective roughness length

It has recently become common practice to parametrize vertical momentum transfer, due to subgridscale motions over inhomogeneous areas, by a grid‐cell averaged effective roughness length, z 0eff, based on the concept of a blending height. Here, it is argued that the requirement that z 0eff be determined above a blending height is not a sufficient condition for its parametrization. Results from a numerical experiment are presented which suggest that z 0eff is strongly modulated by the texture of surface variability. These results demonstrate that the neglect of subgridscale surface texture may lead to an underestimation of z 0eff by up to an order of magnitude. Surface texture, or the spatial relationship between different roughness patches, is interpreted as being of a ‘second‐order roughness’, by which the regular ‘first‐order roughness’ is enhanced owing to the asymmetry of local advection across surface‐type transitions. It is submitted for further discussion that texture analysis provides a framework for the objective quantification of this net effect of subgridscale processes or ‘second‐order roughness’. In the appendix, the flow models used in this study are verified by comparing them with the micrometeorological observations of Bradley.