Hybrid computer simulation of turbulent diffusion in the atmosphere by Monte Carlo methods



Journal Title

Journal ISSN

Volume Title



Turbulent diffusion in the atmosphere was simulated by implementing a new Monte Carlo method on a hybrid computer. The new method involved the development of a stochastic Langevin equation which required the instantaneous wind velocity as input information to simulate the diffusion process. Although several established models were available for the mean wind profile, there were no models available for the fluctuating component of the velocity. Thus a model was developed by using empirical equations to describe the rms value as a function of position and by using independent Gaussian white noises of proper frequency range and power spectral densities. The present method was evaluated by comparing the results to the theoretical dispersion in a homogeneous flow and to experimental concentration profiles in a boundary layer and in the atmosphere. All of the experimental flow fields were nonhomogeneous. Good agreement was found in all cases. The simulated concentration distributions were found to have a 95% statistical reliability by a chi-square goodness-of-fit test. A few of the major advantages of the present method are (1) since the current method simulates the diffusion process directly, it has great flexibility and the concept of eddy diffusion coefficients is not used, and (2) essentially all meteorological effects can be fully utilized. The present method is also applicable to multiple sources of almost any type.