Improving numerical weather prediction at urban to neighbourhood scales

In the current changing climate, national meteorological services are focussing more of their efforts on the prediction of weather extremes at local scales. Forecasting the weather where people live. In order to use the state-of-the-art numerical weather prediction (NWP) models at the scales of cities or individual neighbourhoods, a step-change in resolution is required. Where we currently run NWP models at tens of kilometres globally and a few kilometres regionally, modelling weather extremes in cities would require hectometric resolutions.  

Here, we highlight the development and research required to go towards these scales. These scales not only require changes in the urban canopy parameterisations and more detailed land-use characteristics, but they also require development in other parts of the model. For example, the way clouds and convection are parameterised are important for the triggering of resolved convection. These parameterisations also influence near-surface variables. Additionally, turbulence parameterisations should be adjusted to consider the grey zone of turbulence at 100-m scale resolution.  We also show how to use detailed (remote sensing) observations combined with idealised large eddy simulation (LES), to verify whether the NWP is resolving the right processes. The combination or LES and wind and turbulence profiles from remote sensing at an urban (Rotterdam) and a rural site (Cabauw) in the Netherlands is used to further develop NWP models, like HARMONIE-AROME, at urban scales.