We thank Dr Sorbjan for his comments ; S87) on our paper HN86). We do not think that a semantic discussion about 'governing parameters' or 'basic scaling parameters' is essential. We disagree, however, with Sorbjan's statement that L is the proper height scale in the surface layer instead of z.
We agree that a singularity appears in the temperature profile when cq? < al. (Here a2 and a, are the powers of the heat and momentum flux profiles.) But, let us restate our arguments for a better understanding. In , where incidentally local scaling was first proposed for the Cabauw data, the stress and heat flux profiles with a1 = $ and a2 = 1 were derived. This was done with a closure hypothesis based directly on local scaling. The fact that singularities appear near the top of the boundary layer (at z = h), signifies in our opinion that local scaling fails in that region. This is due to the fact that the underlying assumption: 'production of turbulence equals dissipation', breaks down. Of course, when one uses other values for a1 and a2, one can remove the singularity. The latter seems to be the only point of S87.
However, removing the singularity by choosing suitable values for a1 and cl,, gives no information about the validity of local scaling near the top of the stable boundary layer. In HN86 we have given a first, tentative estimate for the top of the z-less scaling region, until a theory for very stable and intermittent turbulence becomes available. As can be seen in Figure of HN86, the latter estimate leads to given sizes for the z-less scaling and intermittency regions, if the stability is given by h/L.
By analysing the stable Minnesota data, S87 arrives at a cubic heat flux profile (a2 = 3), while Cabauw observations indicate a linear profile (a2 = 1). The differences in the powers can be explained by the fact that the Minnesota data are collected near sunset in a strongly evolving boundary layer, while the Cabauw observations were made some hours after the transition period in more or less stationary conditions. Recent model calculations by Estoumel and Guedalia (1985) con6rm the above described behaviour for the heat flux profile. The fact that various profiles are needed to describe the structure of the boundary layer in different circumstances, should make the user Boundary-Layer Meteorology 38 (1987) 415-416.