Evolving turbulence realizations of atmospheric flow

A significant difference exists between estimates of contaminant atmospheric transport and dispersion calculated by an ensemble-averaged model and the turbulent details of any particular atmospheric transport and dispersion realization. In some cases, however, it is important to be able to make inferences of these realizations using ensemble-averaged models. It is possible to make such inferences if there are sensors in the field to report contaminant concentration observations. Any information determined about the atmospheric transport and dispersion realization can then be assimilated into a forecast model. This approach can enhance the accuracy of the atmospheric transport and dispersion forecast of a particular event. This work adopts that approach and reports on a genetic algorithm used to optimize the variational problem. Given contaminant sensor measurements and a transport and dispersion model, one can back-calculate unknown source and meteorological parameters. In this case, we demonstrate the dynamic recovery of unknown meteorological variables, including the transport variables that comprise the “outer variability” (wind speed and wind direction) and the dispersion variables that comprise the “inner variability” (contaminant spread). The optimization problem is set up in an Eulerian grid space, where the comparison of the concentration field variable between the predictions and the observations forms the cost function. The transport and dispersion parameters, which are determined from the optimization, are in Lagrangian space. This calculation is applied to continuous and instantaneous releases in a horizontally homogeneous wind field such as that observed during traditional transport and dispersion field experiments. The method proves to be successful at recovering the unknown transport and dispersion parameters for a numerical experiment.