8B.2 Numerical Analysis of Turbulence in an Idealized Urban Environment

Wednesday, 15 January 2020: 8:45 AM
104C (Boston Convention and Exhibition Center)
Tim Nagel, CNRM, Toulouse, France; and R. Schoetter, V. Masson, C. Lac, and F. Auguste

In the present work, the atmospheric model MESO-NH is used to investigate the influence of the mesoscale atmospheric turbulence on the mean flow and the turbulence in an idealized urban environment. The Mock Urban Setting Test (MUST) field experiment is numerically reproduced using Large Eddy Simulations (LES). In MUST, the idealized urban environment is represented by 120 ship-containers of 2.5 meters height settled on a flat area of Utah’s West desert. For the considered MUST experiment case, the atmospheric stability is nearly neutral.

Two configurations are considered. MESO-NH is first used as in a typical Computational Fluid Dynamics (CFD) configuration, i.e without accounting for the mesoscale atmospheric turbulence. A logarithmic velocity profile is imposed at the domain boudaries. The domain height is limited to eight times the obstacles elevation (20m). The vertical resolution is of 0.3m up to 6 meters height and increases linearly above. In this configuration, the turbulent fluctuations of the incoming flow are neglected.

In a second configuration, the atmospheric turbulence prevailing in the atmospheric boundary layer is accounted for. To achieve this, we use four nested domains with increasing horizontal resolution. The horizontal resolution of the coarsest model domain is 100 m, as a consequence only the largest eddies of the neutral boundary layer are resoved. An imposed geostrophic wind maintains the flow. The grid nesting method is used for the lateral boundaries of the finer domains (having an horizontal resolution of 20m, 4m and 0.3m). The vertical elevation of all the nested domains goes up to 3000 m in order to simulate the entire atmospheric boudary layer. The finest domain is similar to the CFD-like configuration except for the lateral boundary conditions and the vertical domain extension. In order to enhance the turbulence scale transition between two nested subdomains, a recycling turbulence method is used : at the subdomain inlet, velocity fluctuations are added to the large-scale velocities coming from the father domain.

Finally, in both configurations, the externalized surface SURFEX models the ground friction whereas the containers are represented using the immersed boundary method (IBM) recently developped in MESO-NH.

In the CFD-like case, the incoming flow is not turbulent. Therefore, the turbulence in between the containers only originates from the containers and not from the incoming flow. A previous study has shown that with the CFD-like model configuration, the mean flow and the turbulent fluctuations are consistent with the observations up to approximatively 10m above ground. This altitude corresponds to about 4 times the containers height. Above that height, large differences between the numerical predictions and the field observations have been found. This is due to the larger influence of atmospheric boundary layer turbulence in the incoming flow at this height.

Our work aims to quantify the influence of the mesoscale atmopheric turbulence upon the microscale turbulence in the MUST idealized urban environment. Is it possible to enhance the agreement between field observations and numerical simulation within the region directly influenced by the obstacles (below 10m) by accounting for the mesoscale turbulence ? If so, what about the pollutant dispersion ? How important is the boundary layer turbulence resolution to accurately predict the pollutant dispersion between the containers ?

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner