Coupling an off-line Lagrangian Dispersion Model with Large Eddy Simulations as a Tool for Vertical Mixing Parametrization Development in Mesoscale Applications

Lagrangian Particle Dispersion Models (LPDMs) are widely used for a number of applications, from emergency response to emission source apportionment and estimation. These models are often used in mesoscale applications and driven by Numerical Prediction Models (NWPs), which solve the Reynolds-Averaged Navier Stokes (RANS) equations and, thus, cannot be used to generate high-resolution, 4-D atmospheric boundary layer (ABL) turbulence fields. Thus, velocity variances must be parametrized to resolve the turbulent mixing in the LPDMs. Vertical (and horizontal) mixing parametrizations remain challenging, however.

By using Large Eddy Simulations (LES), velocity variances can be derived and, therefore, used as a reference to develop parametrizations that can be employed in mesoscale applications where the computational cost of LES would be prohibitive. In addition, the capability of driving an LPDM with previously computed LES turbulence fields allows us to quickly recompute dispersion and transport of different sources with a minimum computational cost as compared to a full LES. It also allows us to derive sensitivity functions (footprints) in a receptor-source framework by running the LPDM backward in time.

Here we present an application example of the LPDM-LES off-line coupling for a power plant plume. Turbulence characteristics derived from WRF-LES are shown and compared to airborne measurements, to gain confidence on the performance of both the LES and dispersion simulations. In addition, footprint examples are demonstrated and potential applications described. Next steps involve using this tool to develop new vertical mixing parametrizations in LPDMs for mesoscale applications when driven by RANS models.