In the light of Large Eddy Simulation and other recent work, this paper dis
cusses methods of stable boundary-layer modelling that are established in n
umerical weather prediction and potentially relevant to other applications,
e.g., to dispersion. It is argued that classical scaling theory is best ex
pressed in terms of the local Richardson numbers. The theory can then be ex
tended to treat cases where classical assumptions break down. There are ind
ications that Richardson number models with suitable boundary conditions ca
n mimic, at least qualitatively, some features of the `very stable boundary
layer'. Despite advances in research modelling, detailed observational val
idation remains essential.