Search

Recently, our group developed a complete 3-D model system aimed to simulate atmospheric pollutant dispersion in complex terrain. The system includes the RAMS model (Pielke et al. 1992) providing the flow field, the Lagrangian Stochastic Models SPRAY (Tinarelli et al., in press; Ferrero and Anfossi 1998) computing the dispersion, and the interface code MIRS (Trini Castelli and Anfossi, 1997). Turbulent quantities can be calculated in MIRS by parameterisations from literature or by the turbulent kinetic energy (E) field, directly extracted by RAMS meteorological output.This model system is applied to the wind tunnel EPA-RUSVAL tracer experiment (Khurshudyan et al. 1990) in which a neutral flow is reproduced on a 2-D cross-wind valley and the source is placed near the bottom of the valley. In the version of RAMS here used, we have replaced the MellorYamada 2.5 turbulence closure with both E-l and closures, where l is the length scale and the mean rate of E dissipation. In a previous work (Trini Castelli et al. 1999) we showed that these last reproduce the flow and the turbulent fields in this wind tunnel experiment better than the Mellor-Yamada 2.5 model, originally implemented in RAMS. The influence of these different turbulent closures on the dispersion process is tested by comparing the SPRAY simulations with the observed tracer concentration fields. We focus our attention on the comparison between the simulations carried out using the Hanna (1982) parameterisation and the turbulent field supplied by the E-l or model. The turbulent quantities needed as input to SPRAY are: the three standard deviations of the wind fluctuation components and the three Lagrangian time scale The turbulent model gives the whilst the are obtained from the relationships