2006b) Comments on "Flux-gradient relationship, self-correlation and intermittency in the stable boundary layer". Quart J Roy Meteor Soc (July B) 130: 2087-2103 Sorbjan Z (2007a) Gradient-based similarity in the atmospheric boundary layer. Accepted by Act

that the similarity function for the wind velocity gradient, Φm = κz/u * dU/dz, departs from the linear form 1 + 4.7 z/L (e.g., Businger et al., 1973) in very stable conditions (z/L> 1), where u * = τo1/2, T * =- Ho/u*, κ is the von Karman constant, L = u*2/[κ β T*] is the Monin-Obukhov length, τo and Ho are the surface fluxes of the momentum and temperature. The authors argue convincingly that the phenomenon is associated with self-correlation (i.e., a significant correlation coefficient between considered variables), since both expressions, Φm and z/L, contain a common divisor u*. Large random errors for very stable conditions increase the self-correlation, and consequently produce false results. They conclude that consequently a relationship between Φm and z/L in very stable conditions might never be empirically established with satisfactory confidence. The purpose of this note is to show that the reported departure can also be related to a deficiency of Monin-Obukhov's flux-based scaling in very stable conditions (L < 1 m), and the presence of another similarity regime, characterized by a gradient-based scaling. The Monin-Obukhov similarity yields the following predictions for mean values of the absolute temperature and wind velocity in stable (z-less) conditions: dT/dz ~ T*/L * ~ β Ho2/τo2 ~ cons