A numerical method for objective interpolation of boundary-layer data in the height-time domain is presented. The method is based on a diffusive transport hypothesis and allows for the determination of the optima1 ratio between the height and time scales to be used for non-dimensionalising the independent variables z and t. This ratio, with dimensions of a velocity, may be related to the vertical transfer properties of the atmosphere. A few cases during springtime with different stability conditions are discussed, showing that this ratio varies by at least one order of magnitude between day and night.Nomenclature strength of the diffusive source in the numerical experiments diffusion height scale gravity acceleration height of the atmospheric layer examined Von Karman constant turbulent diffusion coefficient Monin-Obukhov length number of empirical data utilized to compute the value of a given quantity in the grid points total number of data points nondimensional distance between the grid point and the actual one, r = { Kt,,,, -t,ctuaJ/Atlz + Ggric, -~act,,&/W* 1 "' radius of influence gradient Richardson number, Ri = g/s[(a$/az)/(au/az)2] nondimensional ratio between height step AZ and height scale AZ values of s determined by the proposed method duration time of the numerical experiments or time range of the actual data initial time of the numerical experiments horizontal wind velocity friction velocity velocity defined as: u = Az/sAt 'exchange velocity', V = AZ/At = Az/SAt analytical 'exchange velocity', o, = 4K/D (Case A) or IJ, = a (Case B) weighting functions (see Equation (2)) roughness height diffusion parameter (see Equation (4)): K = az time step (arbitrarily chosen) height step (arbitrarily chosen) true height scale potential temperature diffusion time scale T = Dz/4K (Case A) or z = D/a (Case B) Boundary-Layer Meteorology 25 (1983) 159-170. OOOS-8314/83/0252-Ol59$01.80.