Evaluation of a novel WRF/PALM-4U coupling scheme incorporating a roughness-corrected surface layer representation

In this study we conduct realistic urban microclimate simulations using a multiscale approach that couples the PALM-4U microscale model to the WRF mesoscale model. The coupling is achieved by a one-way forcing of the initial and boundary conditions of the microscale model with data interpolated from the mesoscale model. To improve upon similar studies, our approach applies the roughness-corrected Monin–Obukhov similarity theory to fill the gap in the mesoscale results between the first model level and the ground. To assess the improvements of the coupled approach, we compare it to the standalone mesoscale simulation as well as to DWD measurements. Furthermore, the impact of the mesoscale model setup is investigated by comparing different inner domain grid sizes, coupling time steps, and urban parameterizations in an urban district of Berlin for a total of six different mesoscale setups. The comparison of the WRF mesoscale and PALM-4U microscale results indicate that the microscale results, in most cases, follow the mesoscale forcing and hence the microscale accuracy depends on the mesoscale setup. A comparison of simulations to measurements shows that the PALM-4U results tend to have a better agreement with measured data than the WRF results, in most but not all cases. It is also found that one mesoscale forcing setup may lead to better temperature accuracy of PALM-4U results while another one leads to better wind accuracy of PALM-4U results. However, the best overall results for temperature, mixing ratio and wind speed are achieved when forcing PALM-4U by the WRF multi-layer urban canopy model setup. Finally, decreasing the coupling time step from 1 h to 10 min or refining the WRF horizontal grid from 2 km down to 0.4 km do not lead to a significant improvement of PALM-4U simulation results. In all of the investigated cases, the PALM-4U results were sufficiently close to measurements to validate the coupled model approach.