Verifying urban canopy parameterization effects in a mesoscale model with wind tunnel data

Airflow within and above urban canopy layers are modelled by different approaches in a wind tunnel and in a numerical mesoscale model. For the experimental approaches in the wind tunnel, the combination of spires, roughness elements and a physical model generates a scaled boundary layer flow with velocity and turbulence characteristics that are consistent with microscale urban canopy flows in reality. A wind tunnel is comparable in resolution with an obstacle resolving microscale model, therefore data comparisons are frequently done for this scale. However, for many applications numerical models of 1 km resolution are used, resolving mesoscale atmospheric phenomena but not microscale ones. Parameterizations are then used to represent physical processes and obstacle influences on the atmospheres. Due to the coarse resolution, a direct comparison of mesoscale model results and wind tunnel is difficult.

In this study, we use wind tunnel data as validation datasets to evaluate the urban canopy parameterization effects on airflow in a mesoscale model. We have developed a multi-layer urban canopy parameterization using nudging, implemented in the atmospheric model METRAS. The extended model is tested in an idealized case, in which the model domain is designed using realistic topographical data for the Hamburg city center but not resolving buildings. To simplify the city structure, two important canopy morphological parameters are used: building surface fraction and building height. Experiments with a similar model configuration were carried out in parallel in the Blasius wind tunnel facility of the Environmental Wind-Tunnel Laboratory of the University of Hamburg at a model scale of 1:500. Based on the realistic building surface fraction and building height, a pyramid-like model for the urban canopy is placed in the wind tunnel. The set-ups of the numerical model runs and the wind tunnel experiments are designed following two principles: first, keeping the set-up in both approaches as equivalent as possible, in terms of meteorological conditions, roughness lengths, simulation durations, etc.; secondly, taking into account the limitations of the microscale wind tunnel datasets and keeping as many characteristics of atmospheric processes as possible.

The METRAS results show a good agreement with the wind tunnel datasets, in terms of representing building effects such as the reduction of mean wind speeds in the building wake, enhanced turbulence intensities and turbulent fluctuation characteristics for a sufficiently fine scale. However, for coarser resolution, the result comparability reduces and the agreement is less. Thus, we conclude that sub-grid scale canopy effects can be parameterized sufficiently well for their impacts on the average flow, but any detailed changes can only be simulated with a sufficiently high resolution.