Evaluation of local similarity theory in the wintertime nocturnal boundary layer over heterogeneous surface

•Local scaling approach is valid at a forested site with highly inhomogeneous fetch.•Flux-variance and flux-gradient relationships respond differently to inhomogeneity.•Roughness sublayer influence is observed only for the flux-variance relationships.•Above the RSL scaled TKE depends linearly on stability.•Surface inhomogeneity has insignificant effect on dimensionless wind shear. The local scaling approach was examined based on the multi-level measurements of atmospheric turbulence in the wintertime (December 2008–February 2009) stable atmospheric boundary layer (SBL) established over a heterogeneous surface influenced by mixed agricultural, industrial and forest surfaces. The heterogeneity of the surface was characterized by spatial variability of both roughness and topography. Nieuwstadt’s local scaling approach was found to be suitable for the representation of all three wind velocity components. For neutral conditions, values of all three non-dimensional velocity variances were found to be smaller at the lowest measurement level and larger at higher levels in comparison to classical values found over flat terrain. Influence of surface heterogeneity was reflected in the ratio of observed dimensionless standard deviation of the vertical wind component and corresponding values of commonly used similarity formulas for flat and homogeneous terrain showing considerable variation with wind direction. The roughness sublayer influenced wind variances, and consequently the turbulent kinetic energy and correlation coefficients at the lowest measurement level, but not the wind shear profile. The observations support the classical linear expressions for the dimensionless wind shear (ϕm) even over inhomogeneous terrain after removing data points associated with the flux Richardson number (Rf) greater than 0.25. Leveling-off of ϕm at higher stabilities was found to be a result of the large number of data characterized by small-scale turbulence (Rf>0.25). Deviations from linear expressions were shown to be mainly due to this small-scale turbulence rather than due to the surface heterogeneities, supporting the universality of this relationship. Additionally, the flux-gradient dependence on stability did not show different behavior for different wind regimes, indicating that the stability parameter is sufficient predictor for flux-gradient relationship. Data followed local z-less scaling for ϕm when the prerequisite Rf≤0.25 was imposed.