Birgit Sützl, Gabriel Rooney, Anke Finnenkoetter, Sylvia Bohnenstengel, Sue Grimmond, and Maarten van Reeuwijk

Distributed urban drag parameterization in the sub-kilometre scale London Model

vEGU21

Distributed urban drag parameterization in the sub-kilometre scale London ModelSützl et al.vEGU21

• At last year’s EGU, we presented a model for the distribution of building drag within the urban canopy.

• The model characterises urban environments through vertical morphology profiles, rather than morphology parameters.

• More details on this model: https://doi.org/10.1007/s10546-020-00567-0

• We wondered what effect such a model would have when used in high-resolution numerical weather prediction…

A distributed urban drag model

vertical morphology profiles drag profiles

Distributed urban drag parameterization in the sub-kilometre scale London ModelSützl et al.vEGU21

• calculated vertical morphology profiles of urban neighbourhoods (333 m x 333 m) from Ordnance Survey data of 9 million buildings at 1 m resolution

• all profiles are categorised and averaged by their ratio of maximum to mean building height

• curve-fit to a modified exponential distribution with ratio of maximum to mean building height as input parameter

• distributed drag scheme then only requires the frontal area index, mean building height and maximum building height as input parameters

Extracting and parameterising vertical morphology profiles

averaged profiles parameterised profiles

Distributed urban drag parameterization in the sub-kilometre scale London ModelSützl et al.vEGU21

• the London Model is a quasi-operational model at ~333 m horizontal resolution, nested inside the 1.5 km UKV forecasting model.

• tile based land-surface exchange model JULES, urban surface scheme MORUSES, in which we replaced the urban momentum roughness length with the distributed drag model

• case study for 26 June 2018 comparing the standard model (control) with the distributed drag model

• analysis focusses on a 10 km x 15 km central London area (image)

Integration into the Met Office London Model

Data: Ordnance Survey MasterMap Topography

Distributed urban drag parameterization in the sub-kilometre scale London ModelSützl et al.vEGU21

• total surface stress is the same between control and distributed drag model runs, but there are large spatial variations!

• distributed drag-approach shows full range of urban features: high surface stress at densely built-up areas and high-rise building clusters; low surface stress at parks and along the river

• less detail in control run: low stress at parks, high stress at Canary Wharf (high-rise building cluster)

Time-averaged surface stress (total urban drag) over central London

Distributed urban drag parameterization in the sub-kilometre scale London ModelSützl et al.vEGU21

• distributed drag model shows effects of horizontal advection and urban drag with easterly winds: extended wakes with lower wind speeds in downwind direction of densely built-up areas, high wind speeds over the parks and along the river, approaching from the east

• control run wind speed mirrors the surface stress: high wind speed at parks, low wind speed at high-rise areas, but no obvious signs of accumulated downstream effects

Time-averaged wind speed over central London at 10 m model height

Distributed urban drag parameterization in the sub-kilometre scale London ModelSützl et al.vEGU21

• incorporating distributed drag has a profound effect on the wind in the lower atmospheric levels:

• more variation in the wind speed profiles than the standard parameterization,

• differences between neighbourhoods upwind (e.g. Canary Wharf) and downwind (e.g. Oxford Circus) of the city centre show that perturbations grow vertically in the downstream direction, suggesting the formation of internal boundary layers

Vertical wind speed profiles at 19:00 UTC

upwind of centredownwind of centre