Influence of the turbulence typing scheme upon the cumulative frequency distributions of the calculated relative concentrations for different averaging times

Over a period of three consecutive years (1977-1979) hourly meteorological measurements at the Nuclear Energy Research Centre (SCK/CEN) Mol, Belgium and simultaneous synoptic observations at the nearby military airport of Kleine Brogel, have been compiled as input data for a bi-Gaussian dispersion model. The available information has first of all been used to determine hourly stability classes in ten widely used turbulent diffusion typing schemes. Systematic correlations between different systems were rare. Twelve different combinations of diffusion typing scheme-dispersion parameters were then used for calculating cumulative frequency distributions of 1 h, 8 h, 16 h, 3 d and 26 d average ground-level concentrations at receptors respectively at 500 m, l km, 2 km, 4 km and 8 km from a continuous ground-level release and an elevated release at 100 m height. Major differences were noted as well in the extreme values, the higher percentiles, as in the annual mean concentrations. These differences are almost entirely due to the differences in the numerical values (as a function of distance) of the various sets of dispersion parameters actually in use for impact assessment studies. Dispersion parameter sets giving the lowest normalized ground-level concentration values for groundlevel releases give the highest results for elevated releases and vice versa. While it was illustrated once again that the applicability of a given set of dispersion parameters is restricted due to the specific conditions under which the given set was derived (theoretically, experimentally and/or empirically), it was also concluded that systematic experimental work to validate certain assumptions is urgently needed.